@livekit/agents 1.0.39 → 1.0.41

package/dist/tokenize/basic/sentence.cjs

@@ -27,7 +27,7 @@ const splitSentences = (text, minLength = 20, retainFormat = false) => {
   const suffixes = /(Inc|Ltd|Jr|Sr|Co)/g;
   const starters = /(Mr|Mrs|Ms|Dr|Prof|Capt|Cpt|Lt|He\s|She\s|It\s|They\s|Their\s|Our\s|We\s|But\s|However\s|That\s|This\s|Wherever)/g;
   const acronyms = /([A-Z][.][A-Z][.](?:[A-Z][.])?)/g;
-  const websites = /[.](com|net|org|io|gov|edu|me)/g;
+  const websites = /(\w+\.)+(com|net|org|io|gov|edu|me)/g;
   const digits = /([0-9])/g;
   const dots = /\.{2,}/g;
   if (retainFormat) {
@@ -36,7 +36,7 @@ const splitSentences = (text, minLength = 20, retainFormat = false) => {
     text = text.replaceAll("\n", " ");
   }
   text = text.replaceAll(prefixes, "$1<prd>");
-  text = text.replaceAll(websites, "<prd>$2");
+  text = text.replace(websites, (match) => match.replaceAll(".", "<prd>"));
   text = text.replaceAll(new RegExp(`${digits.source}[.]${digits.source}`, "g"), "$1<prd>$2");
   text = text.replaceAll(dots, (match) => "<prd>".repeat(match.length));
   text = text.replaceAll("Ph.D.", "Ph<prd>D<prd>");
@@ -60,7 +60,7 @@ const splitSentences = (text, minLength = 20, retainFormat = false) => {
   text = text.replaceAll('."', '".');
   text = text.replaceAll('!"', '"!');
   text = text.replaceAll('?"', '"?');
-  text = text.replaceAll(".", ".<stop>");
+  text = text.replace(/\.(?=\s|$)/g, ".<stop>");
   text = text.replaceAll("?", "?<stop>");
   text = text.replaceAll("!", "!<stop>");
   text = text.replaceAll("<prd>", ".");

package/dist/tokenize/basic/sentence.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../src/tokenize/basic/sentence.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\n/**\n * Split the text into sentences.\n */\nexport const splitSentences = (\n  text: string,\n  minLength = 20,\n  retainFormat: boolean = false,\n): [string, number, number][] => {\n  const alphabets = /([A-Za-z])/g;\n  const prefixes = /(Mr|St|Mrs|Ms|Dr)[.]/g;\n  const suffixes = /(Inc|Ltd|Jr|Sr|Co)/g;\n  const starters =\n    /(Mr|Mrs|Ms|Dr|Prof|Capt|Cpt|Lt|He\\s|She\\s|It\\s|They\\s|Their\\s|Our\\s|We\\s|But\\s|However\\s|That\\s|This\\s|Wherever)/g;\n  const acronyms = /([A-Z][.][A-Z][.](?:[A-Z][.])?)/g;\n  const websites = /[.](com|net|org|io|gov|edu|me)/g;\n  const digits = /([0-9])/g;\n  const dots = /\\.{2,}/g;\n\n  if (retainFormat) {\n    text = text.replaceAll('\\n', '<nel><stop>');\n  } else {\n    text = text.replaceAll('\\n', ' ');\n  }\n\n  text = text.replaceAll(prefixes, '$1<prd>');\n  text = text.replaceAll(websites, '<prd>$2');\n  text = text.replaceAll(new RegExp(`${digits.source}[.]${digits.source}`, 'g'), '$1<prd>$2');\n  text = text.replaceAll(dots, (match) => '<prd>'.repeat(match.length));\n  text = text.replaceAll('Ph.D.', 'Ph<prd>D<prd>');\n  text = text.replaceAll(new RegExp(`\\\\s${alphabets.source}[.] `, 'g'), ' $1<prd> ');\n  text = text.replaceAll(new RegExp(`${acronyms.source} ${starters.source}`, 'g'), '$1<stop> $2');\n  text = text.replaceAll(\n    new RegExp(`${alphabets.source}[.]${alphabets.source}[.]${alphabets.source}[.]`, 'g'),\n    '$1<prd>$2<prd>$3<prd>',\n  );\n  text = text.replaceAll(\n    new RegExp(`${alphabets.source}[.]${alphabets.source}[.]`, 'g'),\n    '$1<prd>$2<prd>',\n  );\n  text = text.replaceAll(\n    new RegExp(` ${suffixes.source}[.] ${starters.source}`, 'g'),\n    '$1<stop> $2',\n  );\n  text = text.replaceAll(new RegExp(` ${suffixes.source}[.]`, 'g'), '$1<prd>');\n  text = text.replaceAll(new RegExp(` ${alphabets.source}[.]`, 'g'), '$1<prd>');\n  text = text.replaceAll('.”', '”.');\n  text = text.replaceAll('.\"', '\".');\n  text = text.replaceAll('!\"', '\"!');\n  text = text.replaceAll('?\"', '\"?');\n  text = text.replaceAll('.', '.<stop>');\n  text = text.replaceAll('?', '?<stop>');\n  text = text.replaceAll('!', '!<stop>');\n  text = text.replaceAll('<prd>', '.');\n\n  if (retainFormat) {\n    text = text.replaceAll('<nel>', '\\n');\n  }\n\n  const split = text.split('<stop>');\n  text = text.replaceAll('<stop>', '');\n\n  const sentences: [string, number, number][] = [];\n  let buf = '';\n  let start = 0;\n  let end = 0;\n  const prePad = retainFormat ? '' : ' ';\n  for (const match of split) {\n    const sentence = retainFormat ? match : match.trim();\n    if (!sentence) continue;\n\n    buf += prePad + sentence;\n    end += match.length;\n    if (buf.length > minLength) {\n      sentences.push([buf.slice(prePad.length), start, end]);\n      start = end;\n      buf = '';\n    }\n  }\n\n  if (buf) {\n    sentences.push([buf.slice(prePad.length), start, text.length - 1]);\n  }\n\n  return sentences;\n};\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAOO,MAAM,iBAAiB,CAC5B,MACA,YAAY,IACZ,eAAwB,UACO;AAC/B,QAAM,YAAY;AAClB,QAAM,WAAW;AACjB,QAAM,WAAW;AACjB,QAAM,WACJ;AACF,QAAM,WAAW;AACjB,QAAM,WAAW;AACjB,QAAM,SAAS;AACf,QAAM,OAAO;AAEb,MAAI,cAAc;AAChB,WAAO,KAAK,WAAW,MAAM,aAAa;AAAA,EAC5C,OAAO;AACL,WAAO,KAAK,WAAW,MAAM,GAAG;AAAA,EAClC;AAEA,SAAO,KAAK,WAAW,UAAU,SAAS;AAC1C,SAAO,KAAK,WAAW,UAAU,SAAS;AAC1C,SAAO,KAAK,WAAW,IAAI,OAAO,GAAG,OAAO,MAAM,MAAM,OAAO,MAAM,IAAI,GAAG,GAAG,WAAW;AAC1F,SAAO,KAAK,WAAW,MAAM,CAAC,UAAU,QAAQ,OAAO,MAAM,MAAM,CAAC;AACpE,SAAO,KAAK,WAAW,SAAS,eAAe;AAC/C,SAAO,KAAK,WAAW,IAAI,OAAO,MAAM,UAAU,MAAM,QAAQ,GAAG,GAAG,WAAW;AACjF,SAAO,KAAK,WAAW,IAAI,OAAO,GAAG,SAAS,MAAM,IAAI,SAAS,MAAM,IAAI,GAAG,GAAG,aAAa;AAC9F,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,GAAG,UAAU,MAAM,MAAM,UAAU,MAAM,MAAM,UAAU,MAAM,OAAO,GAAG;AAAA,IACpF;AAAA,EACF;AACA,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,GAAG,UAAU,MAAM,MAAM,UAAU,MAAM,OAAO,GAAG;AAAA,IAC9D;AAAA,EACF;AACA,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,IAAI,SAAS,MAAM,OAAO,SAAS,MAAM,IAAI,GAAG;AAAA,IAC3D;AAAA,EACF;AACA,SAAO,KAAK,WAAW,IAAI,OAAO,IAAI,SAAS,MAAM,OAAO,GAAG,GAAG,SAAS;AAC3E,SAAO,KAAK,WAAW,IAAI,OAAO,IAAI,UAAU,MAAM,OAAO,GAAG,GAAG,SAAS;AAC5E,SAAO,KAAK,WAAW,WAAM,SAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,SAAS,GAAG;AAEnC,MAAI,cAAc;AAChB,WAAO,KAAK,WAAW,SAAS,IAAI;AAAA,EACtC;AAEA,QAAM,QAAQ,KAAK,MAAM,QAAQ;AACjC,SAAO,KAAK,WAAW,UAAU,EAAE;AAEnC,QAAM,YAAwC,CAAC;AAC/C,MAAI,MAAM;AACV,MAAI,QAAQ;AACZ,MAAI,MAAM;AACV,QAAM,SAAS,eAAe,KAAK;AACnC,aAAW,SAAS,OAAO;AACzB,UAAM,WAAW,eAAe,QAAQ,MAAM,KAAK;AACnD,QAAI,CAAC,SAAU;AAEf,WAAO,SAAS;AAChB,WAAO,MAAM;AACb,QAAI,IAAI,SAAS,WAAW;AAC1B,gBAAU,KAAK,CAAC,IAAI,MAAM,OAAO,MAAM,GAAG,OAAO,GAAG,CAAC;AACrD,cAAQ;AACR,YAAM;AAAA,IACR;AAAA,EACF;AAEA,MAAI,KAAK;AACP,cAAU,KAAK,CAAC,IAAI,MAAM,OAAO,MAAM,GAAG,OAAO,KAAK,SAAS,CAAC,CAAC;AAAA,EACnE;AAEA,SAAO;AACT;","names":[]}
+{"version":3,"sources":["../../../src/tokenize/basic/sentence.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\n/**\n * Split the text into sentences.\n */\nexport const splitSentences = (\n  text: string,\n  minLength = 20,\n  retainFormat: boolean = false,\n): [string, number, number][] => {\n  const alphabets = /([A-Za-z])/g;\n  const prefixes = /(Mr|St|Mrs|Ms|Dr)[.]/g;\n  const suffixes = /(Inc|Ltd|Jr|Sr|Co)/g;\n  const starters =\n    /(Mr|Mrs|Ms|Dr|Prof|Capt|Cpt|Lt|He\\s|She\\s|It\\s|They\\s|Their\\s|Our\\s|We\\s|But\\s|However\\s|That\\s|This\\s|Wherever)/g;\n  const acronyms = /([A-Z][.][A-Z][.](?:[A-Z][.])?)/g;\n  const websites = /(\\w+\\.)+(com|net|org|io|gov|edu|me)/g;\n  const digits = /([0-9])/g;\n  const dots = /\\.{2,}/g;\n\n  if (retainFormat) {\n    text = text.replaceAll('\\n', '<nel><stop>');\n  } else {\n    text = text.replaceAll('\\n', ' ');\n  }\n\n  text = text.replaceAll(prefixes, '$1<prd>');\n  text = text.replace(websites, (match) => match.replaceAll('.', '<prd>'));\n  text = text.replaceAll(new RegExp(`${digits.source}[.]${digits.source}`, 'g'), '$1<prd>$2');\n  text = text.replaceAll(dots, (match) => '<prd>'.repeat(match.length));\n  text = text.replaceAll('Ph.D.', 'Ph<prd>D<prd>');\n  text = text.replaceAll(new RegExp(`\\\\s${alphabets.source}[.] `, 'g'), ' $1<prd> ');\n  text = text.replaceAll(new RegExp(`${acronyms.source} ${starters.source}`, 'g'), '$1<stop> $2');\n  text = text.replaceAll(\n    new RegExp(`${alphabets.source}[.]${alphabets.source}[.]${alphabets.source}[.]`, 'g'),\n    '$1<prd>$2<prd>$3<prd>',\n  );\n  text = text.replaceAll(\n    new RegExp(`${alphabets.source}[.]${alphabets.source}[.]`, 'g'),\n    '$1<prd>$2<prd>',\n  );\n  text = text.replaceAll(\n    new RegExp(` ${suffixes.source}[.] ${starters.source}`, 'g'),\n    '$1<stop> $2',\n  );\n  text = text.replaceAll(new RegExp(` ${suffixes.source}[.]`, 'g'), '$1<prd>');\n  text = text.replaceAll(new RegExp(` ${alphabets.source}[.]`, 'g'), '$1<prd>');\n  text = text.replaceAll('.”', '”.');\n  text = text.replaceAll('.\"', '\".');\n  text = text.replaceAll('!\"', '\"!');\n  text = text.replaceAll('?\"', '\"?');\n  text = text.replace(/\\.(?=\\s|$)/g, '.<stop>');\n  text = text.replaceAll('?', '?<stop>');\n  text = text.replaceAll('!', '!<stop>');\n  text = text.replaceAll('<prd>', '.');\n\n  if (retainFormat) {\n    text = text.replaceAll('<nel>', '\\n');\n  }\n\n  const split = text.split('<stop>');\n  text = text.replaceAll('<stop>', '');\n\n  const sentences: [string, number, number][] = [];\n  let buf = '';\n  let start = 0;\n  let end = 0;\n  const prePad = retainFormat ? '' : ' ';\n  for (const match of split) {\n    const sentence = retainFormat ? match : match.trim();\n    if (!sentence) continue;\n\n    buf += prePad + sentence;\n    end += match.length;\n    if (buf.length > minLength) {\n      sentences.push([buf.slice(prePad.length), start, end]);\n      start = end;\n      buf = '';\n    }\n  }\n\n  if (buf) {\n    sentences.push([buf.slice(prePad.length), start, text.length - 1]);\n  }\n\n  return sentences;\n};\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAOO,MAAM,iBAAiB,CAC5B,MACA,YAAY,IACZ,eAAwB,UACO;AAC/B,QAAM,YAAY;AAClB,QAAM,WAAW;AACjB,QAAM,WAAW;AACjB,QAAM,WACJ;AACF,QAAM,WAAW;AACjB,QAAM,WAAW;AACjB,QAAM,SAAS;AACf,QAAM,OAAO;AAEb,MAAI,cAAc;AAChB,WAAO,KAAK,WAAW,MAAM,aAAa;AAAA,EAC5C,OAAO;AACL,WAAO,KAAK,WAAW,MAAM,GAAG;AAAA,EAClC;AAEA,SAAO,KAAK,WAAW,UAAU,SAAS;AAC1C,SAAO,KAAK,QAAQ,UAAU,CAAC,UAAU,MAAM,WAAW,KAAK,OAAO,CAAC;AACvE,SAAO,KAAK,WAAW,IAAI,OAAO,GAAG,OAAO,MAAM,MAAM,OAAO,MAAM,IAAI,GAAG,GAAG,WAAW;AAC1F,SAAO,KAAK,WAAW,MAAM,CAAC,UAAU,QAAQ,OAAO,MAAM,MAAM,CAAC;AACpE,SAAO,KAAK,WAAW,SAAS,eAAe;AAC/C,SAAO,KAAK,WAAW,IAAI,OAAO,MAAM,UAAU,MAAM,QAAQ,GAAG,GAAG,WAAW;AACjF,SAAO,KAAK,WAAW,IAAI,OAAO,GAAG,SAAS,MAAM,IAAI,SAAS,MAAM,IAAI,GAAG,GAAG,aAAa;AAC9F,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,GAAG,UAAU,MAAM,MAAM,UAAU,MAAM,MAAM,UAAU,MAAM,OAAO,GAAG;AAAA,IACpF;AAAA,EACF;AACA,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,GAAG,UAAU,MAAM,MAAM,UAAU,MAAM,OAAO,GAAG;AAAA,IAC9D;AAAA,EACF;AACA,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,IAAI,SAAS,MAAM,OAAO,SAAS,MAAM,IAAI,GAAG;AAAA,IAC3D;AAAA,EACF;AACA,SAAO,KAAK,WAAW,IAAI,OAAO,IAAI,SAAS,MAAM,OAAO,GAAG,GAAG,SAAS;AAC3E,SAAO,KAAK,WAAW,IAAI,OAAO,IAAI,UAAU,MAAM,OAAO,GAAG,GAAG,SAAS;AAC5E,SAAO,KAAK,WAAW,WAAM,SAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,QAAQ,eAAe,SAAS;AAC5C,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,SAAS,GAAG;AAEnC,MAAI,cAAc;AAChB,WAAO,KAAK,WAAW,SAAS,IAAI;AAAA,EACtC;AAEA,QAAM,QAAQ,KAAK,MAAM,QAAQ;AACjC,SAAO,KAAK,WAAW,UAAU,EAAE;AAEnC,QAAM,YAAwC,CAAC;AAC/C,MAAI,MAAM;AACV,MAAI,QAAQ;AACZ,MAAI,MAAM;AACV,QAAM,SAAS,eAAe,KAAK;AACnC,aAAW,SAAS,OAAO;AACzB,UAAM,WAAW,eAAe,QAAQ,MAAM,KAAK;AACnD,QAAI,CAAC,SAAU;AAEf,WAAO,SAAS;AAChB,WAAO,MAAM;AACb,QAAI,IAAI,SAAS,WAAW;AAC1B,gBAAU,KAAK,CAAC,IAAI,MAAM,OAAO,MAAM,GAAG,OAAO,GAAG,CAAC;AACrD,cAAQ;AACR,YAAM;AAAA,IACR;AAAA,EACF;AAEA,MAAI,KAAK;AACP,cAAU,KAAK,CAAC,IAAI,MAAM,OAAO,MAAM,GAAG,OAAO,KAAK,SAAS,CAAC,CAAC;AAAA,EACnE;AAEA,SAAO;AACT;","names":[]}

package/dist/tokenize/basic/sentence.js

@@ -4,7 +4,7 @@ const splitSentences = (text, minLength = 20, retainFormat = false) => {
   const suffixes = /(Inc|Ltd|Jr|Sr|Co)/g;
   const starters = /(Mr|Mrs|Ms|Dr|Prof|Capt|Cpt|Lt|He\s|She\s|It\s|They\s|Their\s|Our\s|We\s|But\s|However\s|That\s|This\s|Wherever)/g;
   const acronyms = /([A-Z][.][A-Z][.](?:[A-Z][.])?)/g;
-  const websites = /[.](com|net|org|io|gov|edu|me)/g;
+  const websites = /(\w+\.)+(com|net|org|io|gov|edu|me)/g;
   const digits = /([0-9])/g;
   const dots = /\.{2,}/g;
   if (retainFormat) {
@@ -13,7 +13,7 @@ const splitSentences = (text, minLength = 20, retainFormat = false) => {
     text = text.replaceAll("\n", " ");
   }
   text = text.replaceAll(prefixes, "$1<prd>");
-  text = text.replaceAll(websites, "<prd>$2");
+  text = text.replace(websites, (match) => match.replaceAll(".", "<prd>"));
   text = text.replaceAll(new RegExp(`${digits.source}[.]${digits.source}`, "g"), "$1<prd>$2");
   text = text.replaceAll(dots, (match) => "<prd>".repeat(match.length));
   text = text.replaceAll("Ph.D.", "Ph<prd>D<prd>");
@@ -37,7 +37,7 @@ const splitSentences = (text, minLength = 20, retainFormat = false) => {
   text = text.replaceAll('."', '".');
   text = text.replaceAll('!"', '"!');
   text = text.replaceAll('?"', '"?');
-  text = text.replaceAll(".", ".<stop>");
+  text = text.replace(/\.(?=\s|$)/g, ".<stop>");
   text = text.replaceAll("?", "?<stop>");
   text = text.replaceAll("!", "!<stop>");
   text = text.replaceAll("<prd>", ".");

package/dist/tokenize/basic/sentence.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../src/tokenize/basic/sentence.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\n/**\n * Split the text into sentences.\n */\nexport const splitSentences = (\n  text: string,\n  minLength = 20,\n  retainFormat: boolean = false,\n): [string, number, number][] => {\n  const alphabets = /([A-Za-z])/g;\n  const prefixes = /(Mr|St|Mrs|Ms|Dr)[.]/g;\n  const suffixes = /(Inc|Ltd|Jr|Sr|Co)/g;\n  const starters =\n    /(Mr|Mrs|Ms|Dr|Prof|Capt|Cpt|Lt|He\\s|She\\s|It\\s|They\\s|Their\\s|Our\\s|We\\s|But\\s|However\\s|That\\s|This\\s|Wherever)/g;\n  const acronyms = /([A-Z][.][A-Z][.](?:[A-Z][.])?)/g;\n  const websites = /[.](com|net|org|io|gov|edu|me)/g;\n  const digits = /([0-9])/g;\n  const dots = /\\.{2,}/g;\n\n  if (retainFormat) {\n    text = text.replaceAll('\\n', '<nel><stop>');\n  } else {\n    text = text.replaceAll('\\n', ' ');\n  }\n\n  text = text.replaceAll(prefixes, '$1<prd>');\n  text = text.replaceAll(websites, '<prd>$2');\n  text = text.replaceAll(new RegExp(`${digits.source}[.]${digits.source}`, 'g'), '$1<prd>$2');\n  text = text.replaceAll(dots, (match) => '<prd>'.repeat(match.length));\n  text = text.replaceAll('Ph.D.', 'Ph<prd>D<prd>');\n  text = text.replaceAll(new RegExp(`\\\\s${alphabets.source}[.] `, 'g'), ' $1<prd> ');\n  text = text.replaceAll(new RegExp(`${acronyms.source} ${starters.source}`, 'g'), '$1<stop> $2');\n  text = text.replaceAll(\n    new RegExp(`${alphabets.source}[.]${alphabets.source}[.]${alphabets.source}[.]`, 'g'),\n    '$1<prd>$2<prd>$3<prd>',\n  );\n  text = text.replaceAll(\n    new RegExp(`${alphabets.source}[.]${alphabets.source}[.]`, 'g'),\n    '$1<prd>$2<prd>',\n  );\n  text = text.replaceAll(\n    new RegExp(` ${suffixes.source}[.] ${starters.source}`, 'g'),\n    '$1<stop> $2',\n  );\n  text = text.replaceAll(new RegExp(` ${suffixes.source}[.]`, 'g'), '$1<prd>');\n  text = text.replaceAll(new RegExp(` ${alphabets.source}[.]`, 'g'), '$1<prd>');\n  text = text.replaceAll('.”', '”.');\n  text = text.replaceAll('.\"', '\".');\n  text = text.replaceAll('!\"', '\"!');\n  text = text.replaceAll('?\"', '\"?');\n  text = text.replaceAll('.', '.<stop>');\n  text = text.replaceAll('?', '?<stop>');\n  text = text.replaceAll('!', '!<stop>');\n  text = text.replaceAll('<prd>', '.');\n\n  if (retainFormat) {\n    text = text.replaceAll('<nel>', '\\n');\n  }\n\n  const split = text.split('<stop>');\n  text = text.replaceAll('<stop>', '');\n\n  const sentences: [string, number, number][] = [];\n  let buf = '';\n  let start = 0;\n  let end = 0;\n  const prePad = retainFormat ? '' : ' ';\n  for (const match of split) {\n    const sentence = retainFormat ? match : match.trim();\n    if (!sentence) continue;\n\n    buf += prePad + sentence;\n    end += match.length;\n    if (buf.length > minLength) {\n      sentences.push([buf.slice(prePad.length), start, end]);\n      start = end;\n      buf = '';\n    }\n  }\n\n  if (buf) {\n    sentences.push([buf.slice(prePad.length), start, text.length - 1]);\n  }\n\n  return sentences;\n};\n"],"mappings":"AAOO,MAAM,iBAAiB,CAC5B,MACA,YAAY,IACZ,eAAwB,UACO;AAC/B,QAAM,YAAY;AAClB,QAAM,WAAW;AACjB,QAAM,WAAW;AACjB,QAAM,WACJ;AACF,QAAM,WAAW;AACjB,QAAM,WAAW;AACjB,QAAM,SAAS;AACf,QAAM,OAAO;AAEb,MAAI,cAAc;AAChB,WAAO,KAAK,WAAW,MAAM,aAAa;AAAA,EAC5C,OAAO;AACL,WAAO,KAAK,WAAW,MAAM,GAAG;AAAA,EAClC;AAEA,SAAO,KAAK,WAAW,UAAU,SAAS;AAC1C,SAAO,KAAK,WAAW,UAAU,SAAS;AAC1C,SAAO,KAAK,WAAW,IAAI,OAAO,GAAG,OAAO,MAAM,MAAM,OAAO,MAAM,IAAI,GAAG,GAAG,WAAW;AAC1F,SAAO,KAAK,WAAW,MAAM,CAAC,UAAU,QAAQ,OAAO,MAAM,MAAM,CAAC;AACpE,SAAO,KAAK,WAAW,SAAS,eAAe;AAC/C,SAAO,KAAK,WAAW,IAAI,OAAO,MAAM,UAAU,MAAM,QAAQ,GAAG,GAAG,WAAW;AACjF,SAAO,KAAK,WAAW,IAAI,OAAO,GAAG,SAAS,MAAM,IAAI,SAAS,MAAM,IAAI,GAAG,GAAG,aAAa;AAC9F,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,GAAG,UAAU,MAAM,MAAM,UAAU,MAAM,MAAM,UAAU,MAAM,OAAO,GAAG;AAAA,IACpF;AAAA,EACF;AACA,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,GAAG,UAAU,MAAM,MAAM,UAAU,MAAM,OAAO,GAAG;AAAA,IAC9D;AAAA,EACF;AACA,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,IAAI,SAAS,MAAM,OAAO,SAAS,MAAM,IAAI,GAAG;AAAA,IAC3D;AAAA,EACF;AACA,SAAO,KAAK,WAAW,IAAI,OAAO,IAAI,SAAS,MAAM,OAAO,GAAG,GAAG,SAAS;AAC3E,SAAO,KAAK,WAAW,IAAI,OAAO,IAAI,UAAU,MAAM,OAAO,GAAG,GAAG,SAAS;AAC5E,SAAO,KAAK,WAAW,WAAM,SAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,SAAS,GAAG;AAEnC,MAAI,cAAc;AAChB,WAAO,KAAK,WAAW,SAAS,IAAI;AAAA,EACtC;AAEA,QAAM,QAAQ,KAAK,MAAM,QAAQ;AACjC,SAAO,KAAK,WAAW,UAAU,EAAE;AAEnC,QAAM,YAAwC,CAAC;AAC/C,MAAI,MAAM;AACV,MAAI,QAAQ;AACZ,MAAI,MAAM;AACV,QAAM,SAAS,eAAe,KAAK;AACnC,aAAW,SAAS,OAAO;AACzB,UAAM,WAAW,eAAe,QAAQ,MAAM,KAAK;AACnD,QAAI,CAAC,SAAU;AAEf,WAAO,SAAS;AAChB,WAAO,MAAM;AACb,QAAI,IAAI,SAAS,WAAW;AAC1B,gBAAU,KAAK,CAAC,IAAI,MAAM,OAAO,MAAM,GAAG,OAAO,GAAG,CAAC;AACrD,cAAQ;AACR,YAAM;AAAA,IACR;AAAA,EACF;AAEA,MAAI,KAAK;AACP,cAAU,KAAK,CAAC,IAAI,MAAM,OAAO,MAAM,GAAG,OAAO,KAAK,SAAS,CAAC,CAAC;AAAA,EACnE;AAEA,SAAO;AACT;","names":[]}
+{"version":3,"sources":["../../../src/tokenize/basic/sentence.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\n/**\n * Split the text into sentences.\n */\nexport const splitSentences = (\n  text: string,\n  minLength = 20,\n  retainFormat: boolean = false,\n): [string, number, number][] => {\n  const alphabets = /([A-Za-z])/g;\n  const prefixes = /(Mr|St|Mrs|Ms|Dr)[.]/g;\n  const suffixes = /(Inc|Ltd|Jr|Sr|Co)/g;\n  const starters =\n    /(Mr|Mrs|Ms|Dr|Prof|Capt|Cpt|Lt|He\\s|She\\s|It\\s|They\\s|Their\\s|Our\\s|We\\s|But\\s|However\\s|That\\s|This\\s|Wherever)/g;\n  const acronyms = /([A-Z][.][A-Z][.](?:[A-Z][.])?)/g;\n  const websites = /(\\w+\\.)+(com|net|org|io|gov|edu|me)/g;\n  const digits = /([0-9])/g;\n  const dots = /\\.{2,}/g;\n\n  if (retainFormat) {\n    text = text.replaceAll('\\n', '<nel><stop>');\n  } else {\n    text = text.replaceAll('\\n', ' ');\n  }\n\n  text = text.replaceAll(prefixes, '$1<prd>');\n  text = text.replace(websites, (match) => match.replaceAll('.', '<prd>'));\n  text = text.replaceAll(new RegExp(`${digits.source}[.]${digits.source}`, 'g'), '$1<prd>$2');\n  text = text.replaceAll(dots, (match) => '<prd>'.repeat(match.length));\n  text = text.replaceAll('Ph.D.', 'Ph<prd>D<prd>');\n  text = text.replaceAll(new RegExp(`\\\\s${alphabets.source}[.] `, 'g'), ' $1<prd> ');\n  text = text.replaceAll(new RegExp(`${acronyms.source} ${starters.source}`, 'g'), '$1<stop> $2');\n  text = text.replaceAll(\n    new RegExp(`${alphabets.source}[.]${alphabets.source}[.]${alphabets.source}[.]`, 'g'),\n    '$1<prd>$2<prd>$3<prd>',\n  );\n  text = text.replaceAll(\n    new RegExp(`${alphabets.source}[.]${alphabets.source}[.]`, 'g'),\n    '$1<prd>$2<prd>',\n  );\n  text = text.replaceAll(\n    new RegExp(` ${suffixes.source}[.] ${starters.source}`, 'g'),\n    '$1<stop> $2',\n  );\n  text = text.replaceAll(new RegExp(` ${suffixes.source}[.]`, 'g'), '$1<prd>');\n  text = text.replaceAll(new RegExp(` ${alphabets.source}[.]`, 'g'), '$1<prd>');\n  text = text.replaceAll('.”', '”.');\n  text = text.replaceAll('.\"', '\".');\n  text = text.replaceAll('!\"', '\"!');\n  text = text.replaceAll('?\"', '\"?');\n  text = text.replace(/\\.(?=\\s|$)/g, '.<stop>');\n  text = text.replaceAll('?', '?<stop>');\n  text = text.replaceAll('!', '!<stop>');\n  text = text.replaceAll('<prd>', '.');\n\n  if (retainFormat) {\n    text = text.replaceAll('<nel>', '\\n');\n  }\n\n  const split = text.split('<stop>');\n  text = text.replaceAll('<stop>', '');\n\n  const sentences: [string, number, number][] = [];\n  let buf = '';\n  let start = 0;\n  let end = 0;\n  const prePad = retainFormat ? '' : ' ';\n  for (const match of split) {\n    const sentence = retainFormat ? match : match.trim();\n    if (!sentence) continue;\n\n    buf += prePad + sentence;\n    end += match.length;\n    if (buf.length > minLength) {\n      sentences.push([buf.slice(prePad.length), start, end]);\n      start = end;\n      buf = '';\n    }\n  }\n\n  if (buf) {\n    sentences.push([buf.slice(prePad.length), start, text.length - 1]);\n  }\n\n  return sentences;\n};\n"],"mappings":"AAOO,MAAM,iBAAiB,CAC5B,MACA,YAAY,IACZ,eAAwB,UACO;AAC/B,QAAM,YAAY;AAClB,QAAM,WAAW;AACjB,QAAM,WAAW;AACjB,QAAM,WACJ;AACF,QAAM,WAAW;AACjB,QAAM,WAAW;AACjB,QAAM,SAAS;AACf,QAAM,OAAO;AAEb,MAAI,cAAc;AAChB,WAAO,KAAK,WAAW,MAAM,aAAa;AAAA,EAC5C,OAAO;AACL,WAAO,KAAK,WAAW,MAAM,GAAG;AAAA,EAClC;AAEA,SAAO,KAAK,WAAW,UAAU,SAAS;AAC1C,SAAO,KAAK,QAAQ,UAAU,CAAC,UAAU,MAAM,WAAW,KAAK,OAAO,CAAC;AACvE,SAAO,KAAK,WAAW,IAAI,OAAO,GAAG,OAAO,MAAM,MAAM,OAAO,MAAM,IAAI,GAAG,GAAG,WAAW;AAC1F,SAAO,KAAK,WAAW,MAAM,CAAC,UAAU,QAAQ,OAAO,MAAM,MAAM,CAAC;AACpE,SAAO,KAAK,WAAW,SAAS,eAAe;AAC/C,SAAO,KAAK,WAAW,IAAI,OAAO,MAAM,UAAU,MAAM,QAAQ,GAAG,GAAG,WAAW;AACjF,SAAO,KAAK,WAAW,IAAI,OAAO,GAAG,SAAS,MAAM,IAAI,SAAS,MAAM,IAAI,GAAG,GAAG,aAAa;AAC9F,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,GAAG,UAAU,MAAM,MAAM,UAAU,MAAM,MAAM,UAAU,MAAM,OAAO,GAAG;AAAA,IACpF;AAAA,EACF;AACA,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,GAAG,UAAU,MAAM,MAAM,UAAU,MAAM,OAAO,GAAG;AAAA,IAC9D;AAAA,EACF;AACA,SAAO,KAAK;AAAA,IACV,IAAI,OAAO,IAAI,SAAS,MAAM,OAAO,SAAS,MAAM,IAAI,GAAG;AAAA,IAC3D;AAAA,EACF;AACA,SAAO,KAAK,WAAW,IAAI,OAAO,IAAI,SAAS,MAAM,OAAO,GAAG,GAAG,SAAS;AAC3E,SAAO,KAAK,WAAW,IAAI,OAAO,IAAI,UAAU,MAAM,OAAO,GAAG,GAAG,SAAS;AAC5E,SAAO,KAAK,WAAW,WAAM,SAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,WAAW,MAAM,IAAI;AACjC,SAAO,KAAK,QAAQ,eAAe,SAAS;AAC5C,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,KAAK,SAAS;AACrC,SAAO,KAAK,WAAW,SAAS,GAAG;AAEnC,MAAI,cAAc;AAChB,WAAO,KAAK,WAAW,SAAS,IAAI;AAAA,EACtC;AAEA,QAAM,QAAQ,KAAK,MAAM,QAAQ;AACjC,SAAO,KAAK,WAAW,UAAU,EAAE;AAEnC,QAAM,YAAwC,CAAC;AAC/C,MAAI,MAAM;AACV,MAAI,QAAQ;AACZ,MAAI,MAAM;AACV,QAAM,SAAS,eAAe,KAAK;AACnC,aAAW,SAAS,OAAO;AACzB,UAAM,WAAW,eAAe,QAAQ,MAAM,KAAK;AACnD,QAAI,CAAC,SAAU;AAEf,WAAO,SAAS;AAChB,WAAO,MAAM;AACb,QAAI,IAAI,SAAS,WAAW;AAC1B,gBAAU,KAAK,CAAC,IAAI,MAAM,OAAO,MAAM,GAAG,OAAO,GAAG,CAAC;AACrD,cAAQ;AACR,YAAM;AAAA,IACR;AAAA,EACF;AAEA,MAAI,KAAK;AACP,cAAU,KAAK,CAAC,IAAI,MAAM,OAAO,MAAM,GAAG,OAAO,KAAK,SAAS,CAAC,CAAC;AAAA,EACnE;AAEA,SAAO;AACT;","names":[]}

package/dist/tokenize/tokenizer.test.cjs

@@ -2,13 +2,15 @@
 var import_vitest = require("vitest");
 var import_basic = require("./basic/index.cjs");
 var import_paragraph = require("./basic/paragraph.cjs");
-const TEXT = "Hi! LiveKit is a platform for live audio and video applications and services. R.T.C stands for Real-Time Communication... again R.T.C. Mr. Theo is testing the sentence tokenizer. This is a test. Another test. A short sentence. A longer sentence that is longer than the previous sentence. f(x) = x * 2.54 + 42. Hey! Hi! Hello! ";
+const TEXT = "Hi! LiveKit is a platform for live audio and video applications and services. R.T.C stands for Real-Time Communication... again R.T.C. Mr. Theo is testing the sentence tokenizer. This is a test. Another test. A short sentence. A longer sentence that is longer than the previous sentence. Find additional resources on livekit.com. Find additional resources on docs.livekit.com. f(x) = x * 2.54 + 42. Hey! Hi! Hello! ";
 const EXPECTED_MIN_20 = [
   "Hi! LiveKit is a platform for live audio and video applications and services.",
   "R.T.C stands for Real-Time Communication... again R.T.C.",
   "Mr. Theo is testing the sentence tokenizer.",
   "This is a test. Another test.",
   "A short sentence. A longer sentence that is longer than the previous sentence.",
+  "Find additional resources on livekit.com.",
+  "Find additional resources on docs.livekit.com.",
   "f(x) = x * 2.54 + 42.",
   "Hey! Hi! Hello!"
 ];

package/dist/tokenize/tokenizer.test.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/tokenize/tokenizer.test.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { describe, expect, it } from 'vitest';\nimport { SentenceTokenizer, WordTokenizer, hyphenateWord } from './basic/index.js';\nimport { splitParagraphs } from './basic/paragraph.js';\n\nconst TEXT =\n  'Hi! ' +\n  'LiveKit is a platform for live audio and video applications and services. ' +\n  'R.T.C stands for Real-Time Communication... again R.T.C. ' +\n  'Mr. Theo is testing the sentence tokenizer. ' +\n  'This is a test. Another test. ' +\n  'A short sentence. ' +\n  'A longer sentence that is longer than the previous sentence. ' +\n  'f(x) = x * 2.54 + 42. ' +\n  'Hey! Hi! Hello! ';\n\nconst EXPECTED_MIN_20 = [\n  'Hi! LiveKit is a platform for live audio and video applications and services.',\n  'R.T.C stands for Real-Time Communication... again R.T.C.',\n  'Mr. Theo is testing the sentence tokenizer.',\n  'This is a test. Another test.',\n  'A short sentence. A longer sentence that is longer than the previous sentence.',\n  'f(x) = x * 2.54 + 42.',\n  'Hey! Hi! Hello!',\n];\n\nconst WORDS_TEXT = 'This is a test. Blabla another test! multiple consecutive spaces:     done';\nconst WORDS_EXPECTED = [\n  'This',\n  'is',\n  'a',\n  'test',\n  'Blabla',\n  'another',\n  'test',\n  'multiple',\n  'consecutive',\n  'spaces',\n  'done',\n];\n\nconst WORDS_PUNCT_TEXT =\n  'This is <phoneme alphabet=\"cmu-arpabet\" ph=\"AE K CH UW AH L IY\">actually</phoneme> tricky to handle.';\nconst WORDS_PUNCT_EXPECTED = [\n  'This',\n  'is',\n  '<phoneme',\n  'alphabet=\"cmu-arpabet\"',\n  'ph=\"AE',\n  'K',\n  'CH',\n  'UW',\n  'AH',\n  'L',\n  'IY\">actually</phoneme>',\n  'tricky',\n  'to',\n  'handle.',\n];\n\nconst HYPHENATOR_TEXT = ['Segment', 'expected', 'communication', 'window', 'welcome', 'bedroom'];\nconst HYPHENATOR_EXPECTED = [\n  ['Seg', 'ment'],\n  ['ex', 'pect', 'ed'],\n  ['com', 'mu', 'ni', 'ca', 'tion'],\n  ['win', 'dow'],\n  ['wel', 'come'],\n  ['bed', 'room'],\n];\n\nconst PARAGRAPH_TEST_CASES: [string, [string, number, number][]][] = [\n  ['Single paragraph.', [['Single paragraph.', 0, 17]]],\n  [\n    'Paragraph 1.\\n\\nParagraph 2.',\n    [\n      ['Paragraph 1.', 0, 12],\n      ['Paragraph 2.', 14, 26],\n    ],\n  ],\n  [\n    'Para 1.\\n\\nPara 2.\\n\\nPara 3.',\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 9, 16],\n      ['Para 3.', 18, 25],\n    ],\n  ],\n  ['\\n\\nParagraph with leading newlines.', [['Paragraph with leading newlines.', 2, 34]]],\n  ['Paragraph with trailing newlines.\\n\\n', [['Paragraph with trailing newlines.', 0, 33]]],\n  [\n    '\\n\\n  Paragraph with leading and trailing spaces.  \\n\\n',\n    [['Paragraph with leading and trailing spaces.', 4, 47]],\n  ],\n  [\n    'Para 1.\\n\\n\\n\\nPara 2.', // Multiple newlines between paragraphs\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 11, 18],\n    ],\n  ],\n  [\n    'Para 1.\\n \\n \\nPara 2.', // Newlines with spaces between paragraphs\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 12, 19],\n    ],\n  ],\n  [\n    '', // Empty string\n    [],\n  ],\n  [\n    '\\n\\n\\n', // Only newlines\n    [],\n  ],\n  [\n    'Line 1\\nLine 2\\nLine 3', // Single paragraph with newlines\n    [['Line 1\\nLine 2\\nLine 3', 0, 20]],\n  ],\n];\n\ndescribe('tokenizer', () => {\n  describe('SentenceTokenizer', () => {\n    const tokenizer = new SentenceTokenizer();\n\n    it('should tokenize sentences correctly', () => {\n      expect(tokenizer.tokenize(TEXT).every((x, i) => EXPECTED_MIN_20[i] === x)).toBeTruthy();\n    });\n\n    it('should stream tokenize sentences correctly', async () => {\n      const pattern = [1, 2, 4];\n      let text = TEXT;\n      const chunks = [];\n      const patternIter = Array(Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)))\n        .fill(pattern)\n        .flat()\n        [Symbol.iterator]();\n\n      for (const size of patternIter) {\n        if (!text) break;\n        chunks.push(text.slice(undefined, size));\n        text = text.slice(size);\n      }\n      const stream = tokenizer.stream();\n      for (const chunk of chunks) {\n        stream.pushText(chunk);\n      }\n      stream.endInput();\n      stream.close();\n\n      for (const x of EXPECTED_MIN_20) {\n        await stream.next().then((value) => {\n          if (value.value) {\n            expect(value.value.token).toStrictEqual(x);\n          }\n        });\n      }\n    });\n  });\n  describe('WordTokenizer', () => {\n    const tokenizer = new WordTokenizer();\n\n    it('should tokenize words correctly', () => {\n      expect(tokenizer.tokenize(WORDS_TEXT).every((x, i) => WORDS_EXPECTED[i] === x)).toBeTruthy();\n    });\n\n    it('should stream tokenize words correctly', async () => {\n      const pattern = [1, 2, 4];\n      let text = WORDS_TEXT;\n      const chunks = [];\n      const patternIter = Array(Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)))\n        .fill(pattern)\n        .flat()\n        [Symbol.iterator]();\n\n      for (const size of patternIter) {\n        if (!text) break;\n        chunks.push(text.slice(undefined, size));\n        text = text.slice(size);\n      }\n      const stream = tokenizer.stream();\n      for (const chunk of chunks) {\n        stream.pushText(chunk);\n      }\n      stream.endInput();\n      stream.close();\n\n      for (const x of WORDS_EXPECTED) {\n        await stream.next().then((value) => {\n          if (value.value) {\n            expect(value.value.token).toStrictEqual(x);\n          }\n        });\n      }\n    });\n\n    describe('punctuation handling', () => {\n      const tokenizerPunct = new WordTokenizer(false);\n\n      it('should tokenize words correctly', () => {\n        expect(\n          tokenizerPunct.tokenize(WORDS_PUNCT_TEXT).every((x, i) => WORDS_PUNCT_EXPECTED[i] === x),\n        ).toBeTruthy();\n      });\n\n      it('should stream tokenize words correctly', async () => {\n        const pattern = [1, 2, 4];\n        let text = WORDS_PUNCT_TEXT;\n        const chunks = [];\n        const patternIter = Array(\n          Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)),\n        )\n          .fill(pattern)\n          .flat()\n          [Symbol.iterator]();\n\n        for (const size of patternIter) {\n          if (!text) break;\n          chunks.push(text.slice(undefined, size));\n          text = text.slice(size);\n        }\n        const stream = tokenizerPunct.stream();\n        for (const chunk of chunks) {\n          stream.pushText(chunk);\n        }\n        stream.endInput();\n        stream.close();\n\n        for (const x of WORDS_PUNCT_EXPECTED) {\n          await stream.next().then((value) => {\n            if (value.value) {\n              expect(value.value.token).toStrictEqual(x);\n            }\n          });\n        }\n      });\n    });\n  });\n  describe('hyphenateWord', () => {\n    it('should hyphenate correctly', () => {\n      HYPHENATOR_TEXT.forEach((x, i) => {\n        expect(hyphenateWord(x)).toStrictEqual(HYPHENATOR_EXPECTED[i]);\n      });\n    });\n  });\n  describe('splitParagraphs', () => {\n    it('should tokenize paragraphs correctly', () => {\n      PARAGRAPH_TEST_CASES.forEach(([a, b]) => {\n        expect(splitParagraphs(a)).toStrictEqual(b);\n      });\n    });\n  });\n});\n"],"mappings":";AAGA,oBAAqC;AACrC,mBAAgE;AAChE,uBAAgC;AAEhC,MAAM,OACJ;AAUF,MAAM,kBAAkB;AAAA,EACtB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,aAAa;AACnB,MAAM,iBAAiB;AAAA,EACrB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,mBACJ;AACF,MAAM,uBAAuB;AAAA,EAC3B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,kBAAkB,CAAC,WAAW,YAAY,iBAAiB,UAAU,WAAW,SAAS;AAC/F,MAAM,sBAAsB;AAAA,EAC1B,CAAC,OAAO,MAAM;AAAA,EACd,CAAC,MAAM,QAAQ,IAAI;AAAA,EACnB,CAAC,OAAO,MAAM,MAAM,MAAM,MAAM;AAAA,EAChC,CAAC,OAAO,KAAK;AAAA,EACb,CAAC,OAAO,MAAM;AAAA,EACd,CAAC,OAAO,MAAM;AAChB;AAEA,MAAM,uBAA+D;AAAA,EACnE,CAAC,qBAAqB,CAAC,CAAC,qBAAqB,GAAG,EAAE,CAAC,CAAC;AAAA,EACpD;AAAA,IACE;AAAA,IACA;AAAA,MACE,CAAC,gBAAgB,GAAG,EAAE;AAAA,MACtB,CAAC,gBAAgB,IAAI,EAAE;AAAA,IACzB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,GAAG,EAAE;AAAA,MACjB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA,CAAC,wCAAwC,CAAC,CAAC,oCAAoC,GAAG,EAAE,CAAC,CAAC;AAAA,EACtF,CAAC,yCAAyC,CAAC,CAAC,qCAAqC,GAAG,EAAE,CAAC,CAAC;AAAA,EACxF;AAAA,IACE;AAAA,IACA,CAAC,CAAC,+CAA+C,GAAG,EAAE,CAAC;AAAA,EACzD;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC;AAAA,EACH;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC;AAAA,EACH;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC,CAAC,0BAA0B,GAAG,EAAE,CAAC;AAAA,EACpC;AACF;AAAA,IAEA,wBAAS,aAAa,MAAM;AAC1B,8BAAS,qBAAqB,MAAM;AAClC,UAAM,YAAY,IAAI,+BAAkB;AAExC,0BAAG,uCAAuC,MAAM;AAC9C,gCAAO,UAAU,SAAS,IAAI,EAAE,MAAM,CAAC,GAAG,MAAM,gBAAgB,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;AAAA,IACxF,CAAC;AAED,0BAAG,8CAA8C,YAAY;AAC3D,YAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,UAAI,OAAO;AACX,YAAM,SAAS,CAAC;AAChB,YAAM,cAAc,MAAM,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC,CAAC,EAC1F,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,iBAAW,QAAQ,aAAa;AAC9B,YAAI,CAAC,KAAM;AACX,eAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,eAAO,KAAK,MAAM,IAAI;AAAA,MACxB;AACA,YAAM,SAAS,UAAU,OAAO;AAChC,iBAAW,SAAS,QAAQ;AAC1B,eAAO,SAAS,KAAK;AAAA,MACvB;AACA,aAAO,SAAS;AAChB,aAAO,MAAM;AAEb,iBAAW,KAAK,iBAAiB;AAC/B,cAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,cAAI,MAAM,OAAO;AACf,sCAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,UAC3C;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAAA,EACH,CAAC;AACD,8BAAS,iBAAiB,MAAM;AAC9B,UAAM,YAAY,IAAI,2BAAc;AAEpC,0BAAG,mCAAmC,MAAM;AAC1C,gCAAO,UAAU,SAAS,UAAU,EAAE,MAAM,CAAC,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;AAAA,IAC7F,CAAC;AAED,0BAAG,0CAA0C,YAAY;AACvD,YAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,UAAI,OAAO;AACX,YAAM,SAAS,CAAC;AAChB,YAAM,cAAc,MAAM,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC,CAAC,EAC1F,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,iBAAW,QAAQ,aAAa;AAC9B,YAAI,CAAC,KAAM;AACX,eAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,eAAO,KAAK,MAAM,IAAI;AAAA,MACxB;AACA,YAAM,SAAS,UAAU,OAAO;AAChC,iBAAW,SAAS,QAAQ;AAC1B,eAAO,SAAS,KAAK;AAAA,MACvB;AACA,aAAO,SAAS;AAChB,aAAO,MAAM;AAEb,iBAAW,KAAK,gBAAgB;AAC9B,cAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,cAAI,MAAM,OAAO;AACf,sCAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,UAC3C;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAED,gCAAS,wBAAwB,MAAM;AACrC,YAAM,iBAAiB,IAAI,2BAAc,KAAK;AAE9C,4BAAG,mCAAmC,MAAM;AAC1C;AAAA,UACE,eAAe,SAAS,gBAAgB,EAAE,MAAM,CAAC,GAAG,MAAM,qBAAqB,CAAC,MAAM,CAAC;AAAA,QACzF,EAAE,WAAW;AAAA,MACf,CAAC;AAED,4BAAG,0CAA0C,YAAY;AACvD,cAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,YAAI,OAAO;AACX,cAAM,SAAS,CAAC;AAChB,cAAM,cAAc;AAAA,UAClB,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC;AAAA,QACpE,EACG,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,mBAAW,QAAQ,aAAa;AAC9B,cAAI,CAAC,KAAM;AACX,iBAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,iBAAO,KAAK,MAAM,IAAI;AAAA,QACxB;AACA,cAAM,SAAS,eAAe,OAAO;AACrC,mBAAW,SAAS,QAAQ;AAC1B,iBAAO,SAAS,KAAK;AAAA,QACvB;AACA,eAAO,SAAS;AAChB,eAAO,MAAM;AAEb,mBAAW,KAAK,sBAAsB;AACpC,gBAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,gBAAI,MAAM,OAAO;AACf,wCAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,YAC3C;AAAA,UACF,CAAC;AAAA,QACH;AAAA,MACF,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACD,8BAAS,iBAAiB,MAAM;AAC9B,0BAAG,8BAA8B,MAAM;AACrC,sBAAgB,QAAQ,CAAC,GAAG,MAAM;AAChC,sCAAO,4BAAc,CAAC,CAAC,EAAE,cAAc,oBAAoB,CAAC,CAAC;AAAA,MAC/D,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACD,8BAAS,mBAAmB,MAAM;AAChC,0BAAG,wCAAwC,MAAM;AAC/C,2BAAqB,QAAQ,CAAC,CAAC,GAAG,CAAC,MAAM;AACvC,sCAAO,kCAAgB,CAAC,CAAC,EAAE,cAAc,CAAC;AAAA,MAC5C,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACH,CAAC;","names":[]}
+{"version":3,"sources":["../../src/tokenize/tokenizer.test.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { describe, expect, it } from 'vitest';\nimport { SentenceTokenizer, WordTokenizer, hyphenateWord } from './basic/index.js';\nimport { splitParagraphs } from './basic/paragraph.js';\n\nconst TEXT =\n  'Hi! ' +\n  'LiveKit is a platform for live audio and video applications and services. ' +\n  'R.T.C stands for Real-Time Communication... again R.T.C. ' +\n  'Mr. Theo is testing the sentence tokenizer. ' +\n  'This is a test. Another test. ' +\n  'A short sentence. ' +\n  'A longer sentence that is longer than the previous sentence. ' +\n  'Find additional resources on livekit.com. ' +\n  'Find additional resources on docs.livekit.com. ' +\n  'f(x) = x * 2.54 + 42. ' +\n  'Hey! Hi! Hello! ';\n\nconst EXPECTED_MIN_20 = [\n  'Hi! LiveKit is a platform for live audio and video applications and services.',\n  'R.T.C stands for Real-Time Communication... again R.T.C.',\n  'Mr. Theo is testing the sentence tokenizer.',\n  'This is a test. Another test.',\n  'A short sentence. A longer sentence that is longer than the previous sentence.',\n  'Find additional resources on livekit.com.',\n  'Find additional resources on docs.livekit.com.',\n  'f(x) = x * 2.54 + 42.',\n  'Hey! Hi! Hello!',\n];\n\nconst WORDS_TEXT = 'This is a test. Blabla another test! multiple consecutive spaces:     done';\nconst WORDS_EXPECTED = [\n  'This',\n  'is',\n  'a',\n  'test',\n  'Blabla',\n  'another',\n  'test',\n  'multiple',\n  'consecutive',\n  'spaces',\n  'done',\n];\n\nconst WORDS_PUNCT_TEXT =\n  'This is <phoneme alphabet=\"cmu-arpabet\" ph=\"AE K CH UW AH L IY\">actually</phoneme> tricky to handle.';\nconst WORDS_PUNCT_EXPECTED = [\n  'This',\n  'is',\n  '<phoneme',\n  'alphabet=\"cmu-arpabet\"',\n  'ph=\"AE',\n  'K',\n  'CH',\n  'UW',\n  'AH',\n  'L',\n  'IY\">actually</phoneme>',\n  'tricky',\n  'to',\n  'handle.',\n];\n\nconst HYPHENATOR_TEXT = ['Segment', 'expected', 'communication', 'window', 'welcome', 'bedroom'];\nconst HYPHENATOR_EXPECTED = [\n  ['Seg', 'ment'],\n  ['ex', 'pect', 'ed'],\n  ['com', 'mu', 'ni', 'ca', 'tion'],\n  ['win', 'dow'],\n  ['wel', 'come'],\n  ['bed', 'room'],\n];\n\nconst PARAGRAPH_TEST_CASES: [string, [string, number, number][]][] = [\n  ['Single paragraph.', [['Single paragraph.', 0, 17]]],\n  [\n    'Paragraph 1.\\n\\nParagraph 2.',\n    [\n      ['Paragraph 1.', 0, 12],\n      ['Paragraph 2.', 14, 26],\n    ],\n  ],\n  [\n    'Para 1.\\n\\nPara 2.\\n\\nPara 3.',\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 9, 16],\n      ['Para 3.', 18, 25],\n    ],\n  ],\n  ['\\n\\nParagraph with leading newlines.', [['Paragraph with leading newlines.', 2, 34]]],\n  ['Paragraph with trailing newlines.\\n\\n', [['Paragraph with trailing newlines.', 0, 33]]],\n  [\n    '\\n\\n  Paragraph with leading and trailing spaces.  \\n\\n',\n    [['Paragraph with leading and trailing spaces.', 4, 47]],\n  ],\n  [\n    'Para 1.\\n\\n\\n\\nPara 2.', // Multiple newlines between paragraphs\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 11, 18],\n    ],\n  ],\n  [\n    'Para 1.\\n \\n \\nPara 2.', // Newlines with spaces between paragraphs\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 12, 19],\n    ],\n  ],\n  [\n    '', // Empty string\n    [],\n  ],\n  [\n    '\\n\\n\\n', // Only newlines\n    [],\n  ],\n  [\n    'Line 1\\nLine 2\\nLine 3', // Single paragraph with newlines\n    [['Line 1\\nLine 2\\nLine 3', 0, 20]],\n  ],\n];\n\ndescribe('tokenizer', () => {\n  describe('SentenceTokenizer', () => {\n    const tokenizer = new SentenceTokenizer();\n\n    it('should tokenize sentences correctly', () => {\n      expect(tokenizer.tokenize(TEXT).every((x, i) => EXPECTED_MIN_20[i] === x)).toBeTruthy();\n    });\n\n    it('should stream tokenize sentences correctly', async () => {\n      const pattern = [1, 2, 4];\n      let text = TEXT;\n      const chunks = [];\n      const patternIter = Array(Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)))\n        .fill(pattern)\n        .flat()\n        [Symbol.iterator]();\n\n      for (const size of patternIter) {\n        if (!text) break;\n        chunks.push(text.slice(undefined, size));\n        text = text.slice(size);\n      }\n      const stream = tokenizer.stream();\n      for (const chunk of chunks) {\n        stream.pushText(chunk);\n      }\n      stream.endInput();\n      stream.close();\n\n      for (const x of EXPECTED_MIN_20) {\n        await stream.next().then((value) => {\n          if (value.value) {\n            expect(value.value.token).toStrictEqual(x);\n          }\n        });\n      }\n    });\n  });\n  describe('WordTokenizer', () => {\n    const tokenizer = new WordTokenizer();\n\n    it('should tokenize words correctly', () => {\n      expect(tokenizer.tokenize(WORDS_TEXT).every((x, i) => WORDS_EXPECTED[i] === x)).toBeTruthy();\n    });\n\n    it('should stream tokenize words correctly', async () => {\n      const pattern = [1, 2, 4];\n      let text = WORDS_TEXT;\n      const chunks = [];\n      const patternIter = Array(Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)))\n        .fill(pattern)\n        .flat()\n        [Symbol.iterator]();\n\n      for (const size of patternIter) {\n        if (!text) break;\n        chunks.push(text.slice(undefined, size));\n        text = text.slice(size);\n      }\n      const stream = tokenizer.stream();\n      for (const chunk of chunks) {\n        stream.pushText(chunk);\n      }\n      stream.endInput();\n      stream.close();\n\n      for (const x of WORDS_EXPECTED) {\n        await stream.next().then((value) => {\n          if (value.value) {\n            expect(value.value.token).toStrictEqual(x);\n          }\n        });\n      }\n    });\n\n    describe('punctuation handling', () => {\n      const tokenizerPunct = new WordTokenizer(false);\n\n      it('should tokenize words correctly', () => {\n        expect(\n          tokenizerPunct.tokenize(WORDS_PUNCT_TEXT).every((x, i) => WORDS_PUNCT_EXPECTED[i] === x),\n        ).toBeTruthy();\n      });\n\n      it('should stream tokenize words correctly', async () => {\n        const pattern = [1, 2, 4];\n        let text = WORDS_PUNCT_TEXT;\n        const chunks = [];\n        const patternIter = Array(\n          Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)),\n        )\n          .fill(pattern)\n          .flat()\n          [Symbol.iterator]();\n\n        for (const size of patternIter) {\n          if (!text) break;\n          chunks.push(text.slice(undefined, size));\n          text = text.slice(size);\n        }\n        const stream = tokenizerPunct.stream();\n        for (const chunk of chunks) {\n          stream.pushText(chunk);\n        }\n        stream.endInput();\n        stream.close();\n\n        for (const x of WORDS_PUNCT_EXPECTED) {\n          await stream.next().then((value) => {\n            if (value.value) {\n              expect(value.value.token).toStrictEqual(x);\n            }\n          });\n        }\n      });\n    });\n  });\n  describe('hyphenateWord', () => {\n    it('should hyphenate correctly', () => {\n      HYPHENATOR_TEXT.forEach((x, i) => {\n        expect(hyphenateWord(x)).toStrictEqual(HYPHENATOR_EXPECTED[i]);\n      });\n    });\n  });\n  describe('splitParagraphs', () => {\n    it('should tokenize paragraphs correctly', () => {\n      PARAGRAPH_TEST_CASES.forEach(([a, b]) => {\n        expect(splitParagraphs(a)).toStrictEqual(b);\n      });\n    });\n  });\n});\n"],"mappings":";AAGA,oBAAqC;AACrC,mBAAgE;AAChE,uBAAgC;AAEhC,MAAM,OACJ;AAYF,MAAM,kBAAkB;AAAA,EACtB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,aAAa;AACnB,MAAM,iBAAiB;AAAA,EACrB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,mBACJ;AACF,MAAM,uBAAuB;AAAA,EAC3B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,kBAAkB,CAAC,WAAW,YAAY,iBAAiB,UAAU,WAAW,SAAS;AAC/F,MAAM,sBAAsB;AAAA,EAC1B,CAAC,OAAO,MAAM;AAAA,EACd,CAAC,MAAM,QAAQ,IAAI;AAAA,EACnB,CAAC,OAAO,MAAM,MAAM,MAAM,MAAM;AAAA,EAChC,CAAC,OAAO,KAAK;AAAA,EACb,CAAC,OAAO,MAAM;AAAA,EACd,CAAC,OAAO,MAAM;AAChB;AAEA,MAAM,uBAA+D;AAAA,EACnE,CAAC,qBAAqB,CAAC,CAAC,qBAAqB,GAAG,EAAE,CAAC,CAAC;AAAA,EACpD;AAAA,IACE;AAAA,IACA;AAAA,MACE,CAAC,gBAAgB,GAAG,EAAE;AAAA,MACtB,CAAC,gBAAgB,IAAI,EAAE;AAAA,IACzB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,GAAG,EAAE;AAAA,MACjB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA,CAAC,wCAAwC,CAAC,CAAC,oCAAoC,GAAG,EAAE,CAAC,CAAC;AAAA,EACtF,CAAC,yCAAyC,CAAC,CAAC,qCAAqC,GAAG,EAAE,CAAC,CAAC;AAAA,EACxF;AAAA,IACE;AAAA,IACA,CAAC,CAAC,+CAA+C,GAAG,EAAE,CAAC;AAAA,EACzD;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC;AAAA,EACH;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC;AAAA,EACH;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC,CAAC,0BAA0B,GAAG,EAAE,CAAC;AAAA,EACpC;AACF;AAAA,IAEA,wBAAS,aAAa,MAAM;AAC1B,8BAAS,qBAAqB,MAAM;AAClC,UAAM,YAAY,IAAI,+BAAkB;AAExC,0BAAG,uCAAuC,MAAM;AAC9C,gCAAO,UAAU,SAAS,IAAI,EAAE,MAAM,CAAC,GAAG,MAAM,gBAAgB,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;AAAA,IACxF,CAAC;AAED,0BAAG,8CAA8C,YAAY;AAC3D,YAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,UAAI,OAAO;AACX,YAAM,SAAS,CAAC;AAChB,YAAM,cAAc,MAAM,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC,CAAC,EAC1F,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,iBAAW,QAAQ,aAAa;AAC9B,YAAI,CAAC,KAAM;AACX,eAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,eAAO,KAAK,MAAM,IAAI;AAAA,MACxB;AACA,YAAM,SAAS,UAAU,OAAO;AAChC,iBAAW,SAAS,QAAQ;AAC1B,eAAO,SAAS,KAAK;AAAA,MACvB;AACA,aAAO,SAAS;AAChB,aAAO,MAAM;AAEb,iBAAW,KAAK,iBAAiB;AAC/B,cAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,cAAI,MAAM,OAAO;AACf,sCAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,UAC3C;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAAA,EACH,CAAC;AACD,8BAAS,iBAAiB,MAAM;AAC9B,UAAM,YAAY,IAAI,2BAAc;AAEpC,0BAAG,mCAAmC,MAAM;AAC1C,gCAAO,UAAU,SAAS,UAAU,EAAE,MAAM,CAAC,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;AAAA,IAC7F,CAAC;AAED,0BAAG,0CAA0C,YAAY;AACvD,YAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,UAAI,OAAO;AACX,YAAM,SAAS,CAAC;AAChB,YAAM,cAAc,MAAM,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC,CAAC,EAC1F,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,iBAAW,QAAQ,aAAa;AAC9B,YAAI,CAAC,KAAM;AACX,eAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,eAAO,KAAK,MAAM,IAAI;AAAA,MACxB;AACA,YAAM,SAAS,UAAU,OAAO;AAChC,iBAAW,SAAS,QAAQ;AAC1B,eAAO,SAAS,KAAK;AAAA,MACvB;AACA,aAAO,SAAS;AAChB,aAAO,MAAM;AAEb,iBAAW,KAAK,gBAAgB;AAC9B,cAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,cAAI,MAAM,OAAO;AACf,sCAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,UAC3C;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAED,gCAAS,wBAAwB,MAAM;AACrC,YAAM,iBAAiB,IAAI,2BAAc,KAAK;AAE9C,4BAAG,mCAAmC,MAAM;AAC1C;AAAA,UACE,eAAe,SAAS,gBAAgB,EAAE,MAAM,CAAC,GAAG,MAAM,qBAAqB,CAAC,MAAM,CAAC;AAAA,QACzF,EAAE,WAAW;AAAA,MACf,CAAC;AAED,4BAAG,0CAA0C,YAAY;AACvD,cAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,YAAI,OAAO;AACX,cAAM,SAAS,CAAC;AAChB,cAAM,cAAc;AAAA,UAClB,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC;AAAA,QACpE,EACG,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,mBAAW,QAAQ,aAAa;AAC9B,cAAI,CAAC,KAAM;AACX,iBAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,iBAAO,KAAK,MAAM,IAAI;AAAA,QACxB;AACA,cAAM,SAAS,eAAe,OAAO;AACrC,mBAAW,SAAS,QAAQ;AAC1B,iBAAO,SAAS,KAAK;AAAA,QACvB;AACA,eAAO,SAAS;AAChB,eAAO,MAAM;AAEb,mBAAW,KAAK,sBAAsB;AACpC,gBAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,gBAAI,MAAM,OAAO;AACf,wCAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,YAC3C;AAAA,UACF,CAAC;AAAA,QACH;AAAA,MACF,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACD,8BAAS,iBAAiB,MAAM;AAC9B,0BAAG,8BAA8B,MAAM;AACrC,sBAAgB,QAAQ,CAAC,GAAG,MAAM;AAChC,sCAAO,4BAAc,CAAC,CAAC,EAAE,cAAc,oBAAoB,CAAC,CAAC;AAAA,MAC/D,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACD,8BAAS,mBAAmB,MAAM;AAChC,0BAAG,wCAAwC,MAAM;AAC/C,2BAAqB,QAAQ,CAAC,CAAC,GAAG,CAAC,MAAM;AACvC,sCAAO,kCAAgB,CAAC,CAAC,EAAE,cAAc,CAAC;AAAA,MAC5C,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACH,CAAC;","names":[]}

package/dist/tokenize/tokenizer.test.js

@@ -1,13 +1,15 @@
 import { describe, expect, it } from "vitest";
 import { SentenceTokenizer, WordTokenizer, hyphenateWord } from "./basic/index.js";
 import { splitParagraphs } from "./basic/paragraph.js";
-const TEXT = "Hi! LiveKit is a platform for live audio and video applications and services. R.T.C stands for Real-Time Communication... again R.T.C. Mr. Theo is testing the sentence tokenizer. This is a test. Another test. A short sentence. A longer sentence that is longer than the previous sentence. f(x) = x * 2.54 + 42. Hey! Hi! Hello! ";
+const TEXT = "Hi! LiveKit is a platform for live audio and video applications and services. R.T.C stands for Real-Time Communication... again R.T.C. Mr. Theo is testing the sentence tokenizer. This is a test. Another test. A short sentence. A longer sentence that is longer than the previous sentence. Find additional resources on livekit.com. Find additional resources on docs.livekit.com. f(x) = x * 2.54 + 42. Hey! Hi! Hello! ";
 const EXPECTED_MIN_20 = [
   "Hi! LiveKit is a platform for live audio and video applications and services.",
   "R.T.C stands for Real-Time Communication... again R.T.C.",
   "Mr. Theo is testing the sentence tokenizer.",
   "This is a test. Another test.",
   "A short sentence. A longer sentence that is longer than the previous sentence.",
+  "Find additional resources on livekit.com.",
+  "Find additional resources on docs.livekit.com.",
   "f(x) = x * 2.54 + 42.",
   "Hey! Hi! Hello!"
 ];

package/dist/tokenize/tokenizer.test.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/tokenize/tokenizer.test.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { describe, expect, it } from 'vitest';\nimport { SentenceTokenizer, WordTokenizer, hyphenateWord } from './basic/index.js';\nimport { splitParagraphs } from './basic/paragraph.js';\n\nconst TEXT =\n  'Hi! ' +\n  'LiveKit is a platform for live audio and video applications and services. ' +\n  'R.T.C stands for Real-Time Communication... again R.T.C. ' +\n  'Mr. Theo is testing the sentence tokenizer. ' +\n  'This is a test. Another test. ' +\n  'A short sentence. ' +\n  'A longer sentence that is longer than the previous sentence. ' +\n  'f(x) = x * 2.54 + 42. ' +\n  'Hey! Hi! Hello! ';\n\nconst EXPECTED_MIN_20 = [\n  'Hi! LiveKit is a platform for live audio and video applications and services.',\n  'R.T.C stands for Real-Time Communication... again R.T.C.',\n  'Mr. Theo is testing the sentence tokenizer.',\n  'This is a test. Another test.',\n  'A short sentence. A longer sentence that is longer than the previous sentence.',\n  'f(x) = x * 2.54 + 42.',\n  'Hey! Hi! Hello!',\n];\n\nconst WORDS_TEXT = 'This is a test. Blabla another test! multiple consecutive spaces:     done';\nconst WORDS_EXPECTED = [\n  'This',\n  'is',\n  'a',\n  'test',\n  'Blabla',\n  'another',\n  'test',\n  'multiple',\n  'consecutive',\n  'spaces',\n  'done',\n];\n\nconst WORDS_PUNCT_TEXT =\n  'This is <phoneme alphabet=\"cmu-arpabet\" ph=\"AE K CH UW AH L IY\">actually</phoneme> tricky to handle.';\nconst WORDS_PUNCT_EXPECTED = [\n  'This',\n  'is',\n  '<phoneme',\n  'alphabet=\"cmu-arpabet\"',\n  'ph=\"AE',\n  'K',\n  'CH',\n  'UW',\n  'AH',\n  'L',\n  'IY\">actually</phoneme>',\n  'tricky',\n  'to',\n  'handle.',\n];\n\nconst HYPHENATOR_TEXT = ['Segment', 'expected', 'communication', 'window', 'welcome', 'bedroom'];\nconst HYPHENATOR_EXPECTED = [\n  ['Seg', 'ment'],\n  ['ex', 'pect', 'ed'],\n  ['com', 'mu', 'ni', 'ca', 'tion'],\n  ['win', 'dow'],\n  ['wel', 'come'],\n  ['bed', 'room'],\n];\n\nconst PARAGRAPH_TEST_CASES: [string, [string, number, number][]][] = [\n  ['Single paragraph.', [['Single paragraph.', 0, 17]]],\n  [\n    'Paragraph 1.\\n\\nParagraph 2.',\n    [\n      ['Paragraph 1.', 0, 12],\n      ['Paragraph 2.', 14, 26],\n    ],\n  ],\n  [\n    'Para 1.\\n\\nPara 2.\\n\\nPara 3.',\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 9, 16],\n      ['Para 3.', 18, 25],\n    ],\n  ],\n  ['\\n\\nParagraph with leading newlines.', [['Paragraph with leading newlines.', 2, 34]]],\n  ['Paragraph with trailing newlines.\\n\\n', [['Paragraph with trailing newlines.', 0, 33]]],\n  [\n    '\\n\\n  Paragraph with leading and trailing spaces.  \\n\\n',\n    [['Paragraph with leading and trailing spaces.', 4, 47]],\n  ],\n  [\n    'Para 1.\\n\\n\\n\\nPara 2.', // Multiple newlines between paragraphs\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 11, 18],\n    ],\n  ],\n  [\n    'Para 1.\\n \\n \\nPara 2.', // Newlines with spaces between paragraphs\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 12, 19],\n    ],\n  ],\n  [\n    '', // Empty string\n    [],\n  ],\n  [\n    '\\n\\n\\n', // Only newlines\n    [],\n  ],\n  [\n    'Line 1\\nLine 2\\nLine 3', // Single paragraph with newlines\n    [['Line 1\\nLine 2\\nLine 3', 0, 20]],\n  ],\n];\n\ndescribe('tokenizer', () => {\n  describe('SentenceTokenizer', () => {\n    const tokenizer = new SentenceTokenizer();\n\n    it('should tokenize sentences correctly', () => {\n      expect(tokenizer.tokenize(TEXT).every((x, i) => EXPECTED_MIN_20[i] === x)).toBeTruthy();\n    });\n\n    it('should stream tokenize sentences correctly', async () => {\n      const pattern = [1, 2, 4];\n      let text = TEXT;\n      const chunks = [];\n      const patternIter = Array(Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)))\n        .fill(pattern)\n        .flat()\n        [Symbol.iterator]();\n\n      for (const size of patternIter) {\n        if (!text) break;\n        chunks.push(text.slice(undefined, size));\n        text = text.slice(size);\n      }\n      const stream = tokenizer.stream();\n      for (const chunk of chunks) {\n        stream.pushText(chunk);\n      }\n      stream.endInput();\n      stream.close();\n\n      for (const x of EXPECTED_MIN_20) {\n        await stream.next().then((value) => {\n          if (value.value) {\n            expect(value.value.token).toStrictEqual(x);\n          }\n        });\n      }\n    });\n  });\n  describe('WordTokenizer', () => {\n    const tokenizer = new WordTokenizer();\n\n    it('should tokenize words correctly', () => {\n      expect(tokenizer.tokenize(WORDS_TEXT).every((x, i) => WORDS_EXPECTED[i] === x)).toBeTruthy();\n    });\n\n    it('should stream tokenize words correctly', async () => {\n      const pattern = [1, 2, 4];\n      let text = WORDS_TEXT;\n      const chunks = [];\n      const patternIter = Array(Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)))\n        .fill(pattern)\n        .flat()\n        [Symbol.iterator]();\n\n      for (const size of patternIter) {\n        if (!text) break;\n        chunks.push(text.slice(undefined, size));\n        text = text.slice(size);\n      }\n      const stream = tokenizer.stream();\n      for (const chunk of chunks) {\n        stream.pushText(chunk);\n      }\n      stream.endInput();\n      stream.close();\n\n      for (const x of WORDS_EXPECTED) {\n        await stream.next().then((value) => {\n          if (value.value) {\n            expect(value.value.token).toStrictEqual(x);\n          }\n        });\n      }\n    });\n\n    describe('punctuation handling', () => {\n      const tokenizerPunct = new WordTokenizer(false);\n\n      it('should tokenize words correctly', () => {\n        expect(\n          tokenizerPunct.tokenize(WORDS_PUNCT_TEXT).every((x, i) => WORDS_PUNCT_EXPECTED[i] === x),\n        ).toBeTruthy();\n      });\n\n      it('should stream tokenize words correctly', async () => {\n        const pattern = [1, 2, 4];\n        let text = WORDS_PUNCT_TEXT;\n        const chunks = [];\n        const patternIter = Array(\n          Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)),\n        )\n          .fill(pattern)\n          .flat()\n          [Symbol.iterator]();\n\n        for (const size of patternIter) {\n          if (!text) break;\n          chunks.push(text.slice(undefined, size));\n          text = text.slice(size);\n        }\n        const stream = tokenizerPunct.stream();\n        for (const chunk of chunks) {\n          stream.pushText(chunk);\n        }\n        stream.endInput();\n        stream.close();\n\n        for (const x of WORDS_PUNCT_EXPECTED) {\n          await stream.next().then((value) => {\n            if (value.value) {\n              expect(value.value.token).toStrictEqual(x);\n            }\n          });\n        }\n      });\n    });\n  });\n  describe('hyphenateWord', () => {\n    it('should hyphenate correctly', () => {\n      HYPHENATOR_TEXT.forEach((x, i) => {\n        expect(hyphenateWord(x)).toStrictEqual(HYPHENATOR_EXPECTED[i]);\n      });\n    });\n  });\n  describe('splitParagraphs', () => {\n    it('should tokenize paragraphs correctly', () => {\n      PARAGRAPH_TEST_CASES.forEach(([a, b]) => {\n        expect(splitParagraphs(a)).toStrictEqual(b);\n      });\n    });\n  });\n});\n"],"mappings":"AAGA,SAAS,UAAU,QAAQ,UAAU;AACrC,SAAS,mBAAmB,eAAe,qBAAqB;AAChE,SAAS,uBAAuB;AAEhC,MAAM,OACJ;AAUF,MAAM,kBAAkB;AAAA,EACtB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,aAAa;AACnB,MAAM,iBAAiB;AAAA,EACrB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,mBACJ;AACF,MAAM,uBAAuB;AAAA,EAC3B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,kBAAkB,CAAC,WAAW,YAAY,iBAAiB,UAAU,WAAW,SAAS;AAC/F,MAAM,sBAAsB;AAAA,EAC1B,CAAC,OAAO,MAAM;AAAA,EACd,CAAC,MAAM,QAAQ,IAAI;AAAA,EACnB,CAAC,OAAO,MAAM,MAAM,MAAM,MAAM;AAAA,EAChC,CAAC,OAAO,KAAK;AAAA,EACb,CAAC,OAAO,MAAM;AAAA,EACd,CAAC,OAAO,MAAM;AAChB;AAEA,MAAM,uBAA+D;AAAA,EACnE,CAAC,qBAAqB,CAAC,CAAC,qBAAqB,GAAG,EAAE,CAAC,CAAC;AAAA,EACpD;AAAA,IACE;AAAA,IACA;AAAA,MACE,CAAC,gBAAgB,GAAG,EAAE;AAAA,MACtB,CAAC,gBAAgB,IAAI,EAAE;AAAA,IACzB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,GAAG,EAAE;AAAA,MACjB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA,CAAC,wCAAwC,CAAC,CAAC,oCAAoC,GAAG,EAAE,CAAC,CAAC;AAAA,EACtF,CAAC,yCAAyC,CAAC,CAAC,qCAAqC,GAAG,EAAE,CAAC,CAAC;AAAA,EACxF;AAAA,IACE;AAAA,IACA,CAAC,CAAC,+CAA+C,GAAG,EAAE,CAAC;AAAA,EACzD;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC;AAAA,EACH;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC;AAAA,EACH;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC,CAAC,0BAA0B,GAAG,EAAE,CAAC;AAAA,EACpC;AACF;AAEA,SAAS,aAAa,MAAM;AAC1B,WAAS,qBAAqB,MAAM;AAClC,UAAM,YAAY,IAAI,kBAAkB;AAExC,OAAG,uCAAuC,MAAM;AAC9C,aAAO,UAAU,SAAS,IAAI,EAAE,MAAM,CAAC,GAAG,MAAM,gBAAgB,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;AAAA,IACxF,CAAC;AAED,OAAG,8CAA8C,YAAY;AAC3D,YAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,UAAI,OAAO;AACX,YAAM,SAAS,CAAC;AAChB,YAAM,cAAc,MAAM,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC,CAAC,EAC1F,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,iBAAW,QAAQ,aAAa;AAC9B,YAAI,CAAC,KAAM;AACX,eAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,eAAO,KAAK,MAAM,IAAI;AAAA,MACxB;AACA,YAAM,SAAS,UAAU,OAAO;AAChC,iBAAW,SAAS,QAAQ;AAC1B,eAAO,SAAS,KAAK;AAAA,MACvB;AACA,aAAO,SAAS;AAChB,aAAO,MAAM;AAEb,iBAAW,KAAK,iBAAiB;AAC/B,cAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,cAAI,MAAM,OAAO;AACf,mBAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,UAC3C;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAAA,EACH,CAAC;AACD,WAAS,iBAAiB,MAAM;AAC9B,UAAM,YAAY,IAAI,cAAc;AAEpC,OAAG,mCAAmC,MAAM;AAC1C,aAAO,UAAU,SAAS,UAAU,EAAE,MAAM,CAAC,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;AAAA,IAC7F,CAAC;AAED,OAAG,0CAA0C,YAAY;AACvD,YAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,UAAI,OAAO;AACX,YAAM,SAAS,CAAC;AAChB,YAAM,cAAc,MAAM,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC,CAAC,EAC1F,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,iBAAW,QAAQ,aAAa;AAC9B,YAAI,CAAC,KAAM;AACX,eAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,eAAO,KAAK,MAAM,IAAI;AAAA,MACxB;AACA,YAAM,SAAS,UAAU,OAAO;AAChC,iBAAW,SAAS,QAAQ;AAC1B,eAAO,SAAS,KAAK;AAAA,MACvB;AACA,aAAO,SAAS;AAChB,aAAO,MAAM;AAEb,iBAAW,KAAK,gBAAgB;AAC9B,cAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,cAAI,MAAM,OAAO;AACf,mBAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,UAC3C;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAED,aAAS,wBAAwB,MAAM;AACrC,YAAM,iBAAiB,IAAI,cAAc,KAAK;AAE9C,SAAG,mCAAmC,MAAM;AAC1C;AAAA,UACE,eAAe,SAAS,gBAAgB,EAAE,MAAM,CAAC,GAAG,MAAM,qBAAqB,CAAC,MAAM,CAAC;AAAA,QACzF,EAAE,WAAW;AAAA,MACf,CAAC;AAED,SAAG,0CAA0C,YAAY;AACvD,cAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,YAAI,OAAO;AACX,cAAM,SAAS,CAAC;AAChB,cAAM,cAAc;AAAA,UAClB,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC;AAAA,QACpE,EACG,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,mBAAW,QAAQ,aAAa;AAC9B,cAAI,CAAC,KAAM;AACX,iBAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,iBAAO,KAAK,MAAM,IAAI;AAAA,QACxB;AACA,cAAM,SAAS,eAAe,OAAO;AACrC,mBAAW,SAAS,QAAQ;AAC1B,iBAAO,SAAS,KAAK;AAAA,QACvB;AACA,eAAO,SAAS;AAChB,eAAO,MAAM;AAEb,mBAAW,KAAK,sBAAsB;AACpC,gBAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,gBAAI,MAAM,OAAO;AACf,qBAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,YAC3C;AAAA,UACF,CAAC;AAAA,QACH;AAAA,MACF,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACD,WAAS,iBAAiB,MAAM;AAC9B,OAAG,8BAA8B,MAAM;AACrC,sBAAgB,QAAQ,CAAC,GAAG,MAAM;AAChC,eAAO,cAAc,CAAC,CAAC,EAAE,cAAc,oBAAoB,CAAC,CAAC;AAAA,MAC/D,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACD,WAAS,mBAAmB,MAAM;AAChC,OAAG,wCAAwC,MAAM;AAC/C,2BAAqB,QAAQ,CAAC,CAAC,GAAG,CAAC,MAAM;AACvC,eAAO,gBAAgB,CAAC,CAAC,EAAE,cAAc,CAAC;AAAA,MAC5C,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACH,CAAC;","names":[]}
+{"version":3,"sources":["../../src/tokenize/tokenizer.test.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { describe, expect, it } from 'vitest';\nimport { SentenceTokenizer, WordTokenizer, hyphenateWord } from './basic/index.js';\nimport { splitParagraphs } from './basic/paragraph.js';\n\nconst TEXT =\n  'Hi! ' +\n  'LiveKit is a platform for live audio and video applications and services. ' +\n  'R.T.C stands for Real-Time Communication... again R.T.C. ' +\n  'Mr. Theo is testing the sentence tokenizer. ' +\n  'This is a test. Another test. ' +\n  'A short sentence. ' +\n  'A longer sentence that is longer than the previous sentence. ' +\n  'Find additional resources on livekit.com. ' +\n  'Find additional resources on docs.livekit.com. ' +\n  'f(x) = x * 2.54 + 42. ' +\n  'Hey! Hi! Hello! ';\n\nconst EXPECTED_MIN_20 = [\n  'Hi! LiveKit is a platform for live audio and video applications and services.',\n  'R.T.C stands for Real-Time Communication... again R.T.C.',\n  'Mr. Theo is testing the sentence tokenizer.',\n  'This is a test. Another test.',\n  'A short sentence. A longer sentence that is longer than the previous sentence.',\n  'Find additional resources on livekit.com.',\n  'Find additional resources on docs.livekit.com.',\n  'f(x) = x * 2.54 + 42.',\n  'Hey! Hi! Hello!',\n];\n\nconst WORDS_TEXT = 'This is a test. Blabla another test! multiple consecutive spaces:     done';\nconst WORDS_EXPECTED = [\n  'This',\n  'is',\n  'a',\n  'test',\n  'Blabla',\n  'another',\n  'test',\n  'multiple',\n  'consecutive',\n  'spaces',\n  'done',\n];\n\nconst WORDS_PUNCT_TEXT =\n  'This is <phoneme alphabet=\"cmu-arpabet\" ph=\"AE K CH UW AH L IY\">actually</phoneme> tricky to handle.';\nconst WORDS_PUNCT_EXPECTED = [\n  'This',\n  'is',\n  '<phoneme',\n  'alphabet=\"cmu-arpabet\"',\n  'ph=\"AE',\n  'K',\n  'CH',\n  'UW',\n  'AH',\n  'L',\n  'IY\">actually</phoneme>',\n  'tricky',\n  'to',\n  'handle.',\n];\n\nconst HYPHENATOR_TEXT = ['Segment', 'expected', 'communication', 'window', 'welcome', 'bedroom'];\nconst HYPHENATOR_EXPECTED = [\n  ['Seg', 'ment'],\n  ['ex', 'pect', 'ed'],\n  ['com', 'mu', 'ni', 'ca', 'tion'],\n  ['win', 'dow'],\n  ['wel', 'come'],\n  ['bed', 'room'],\n];\n\nconst PARAGRAPH_TEST_CASES: [string, [string, number, number][]][] = [\n  ['Single paragraph.', [['Single paragraph.', 0, 17]]],\n  [\n    'Paragraph 1.\\n\\nParagraph 2.',\n    [\n      ['Paragraph 1.', 0, 12],\n      ['Paragraph 2.', 14, 26],\n    ],\n  ],\n  [\n    'Para 1.\\n\\nPara 2.\\n\\nPara 3.',\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 9, 16],\n      ['Para 3.', 18, 25],\n    ],\n  ],\n  ['\\n\\nParagraph with leading newlines.', [['Paragraph with leading newlines.', 2, 34]]],\n  ['Paragraph with trailing newlines.\\n\\n', [['Paragraph with trailing newlines.', 0, 33]]],\n  [\n    '\\n\\n  Paragraph with leading and trailing spaces.  \\n\\n',\n    [['Paragraph with leading and trailing spaces.', 4, 47]],\n  ],\n  [\n    'Para 1.\\n\\n\\n\\nPara 2.', // Multiple newlines between paragraphs\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 11, 18],\n    ],\n  ],\n  [\n    'Para 1.\\n \\n \\nPara 2.', // Newlines with spaces between paragraphs\n    [\n      ['Para 1.', 0, 7],\n      ['Para 2.', 12, 19],\n    ],\n  ],\n  [\n    '', // Empty string\n    [],\n  ],\n  [\n    '\\n\\n\\n', // Only newlines\n    [],\n  ],\n  [\n    'Line 1\\nLine 2\\nLine 3', // Single paragraph with newlines\n    [['Line 1\\nLine 2\\nLine 3', 0, 20]],\n  ],\n];\n\ndescribe('tokenizer', () => {\n  describe('SentenceTokenizer', () => {\n    const tokenizer = new SentenceTokenizer();\n\n    it('should tokenize sentences correctly', () => {\n      expect(tokenizer.tokenize(TEXT).every((x, i) => EXPECTED_MIN_20[i] === x)).toBeTruthy();\n    });\n\n    it('should stream tokenize sentences correctly', async () => {\n      const pattern = [1, 2, 4];\n      let text = TEXT;\n      const chunks = [];\n      const patternIter = Array(Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)))\n        .fill(pattern)\n        .flat()\n        [Symbol.iterator]();\n\n      for (const size of patternIter) {\n        if (!text) break;\n        chunks.push(text.slice(undefined, size));\n        text = text.slice(size);\n      }\n      const stream = tokenizer.stream();\n      for (const chunk of chunks) {\n        stream.pushText(chunk);\n      }\n      stream.endInput();\n      stream.close();\n\n      for (const x of EXPECTED_MIN_20) {\n        await stream.next().then((value) => {\n          if (value.value) {\n            expect(value.value.token).toStrictEqual(x);\n          }\n        });\n      }\n    });\n  });\n  describe('WordTokenizer', () => {\n    const tokenizer = new WordTokenizer();\n\n    it('should tokenize words correctly', () => {\n      expect(tokenizer.tokenize(WORDS_TEXT).every((x, i) => WORDS_EXPECTED[i] === x)).toBeTruthy();\n    });\n\n    it('should stream tokenize words correctly', async () => {\n      const pattern = [1, 2, 4];\n      let text = WORDS_TEXT;\n      const chunks = [];\n      const patternIter = Array(Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)))\n        .fill(pattern)\n        .flat()\n        [Symbol.iterator]();\n\n      for (const size of patternIter) {\n        if (!text) break;\n        chunks.push(text.slice(undefined, size));\n        text = text.slice(size);\n      }\n      const stream = tokenizer.stream();\n      for (const chunk of chunks) {\n        stream.pushText(chunk);\n      }\n      stream.endInput();\n      stream.close();\n\n      for (const x of WORDS_EXPECTED) {\n        await stream.next().then((value) => {\n          if (value.value) {\n            expect(value.value.token).toStrictEqual(x);\n          }\n        });\n      }\n    });\n\n    describe('punctuation handling', () => {\n      const tokenizerPunct = new WordTokenizer(false);\n\n      it('should tokenize words correctly', () => {\n        expect(\n          tokenizerPunct.tokenize(WORDS_PUNCT_TEXT).every((x, i) => WORDS_PUNCT_EXPECTED[i] === x),\n        ).toBeTruthy();\n      });\n\n      it('should stream tokenize words correctly', async () => {\n        const pattern = [1, 2, 4];\n        let text = WORDS_PUNCT_TEXT;\n        const chunks = [];\n        const patternIter = Array(\n          Math.ceil(text.length / pattern.reduce((sum, num) => sum + num, 0)),\n        )\n          .fill(pattern)\n          .flat()\n          [Symbol.iterator]();\n\n        for (const size of patternIter) {\n          if (!text) break;\n          chunks.push(text.slice(undefined, size));\n          text = text.slice(size);\n        }\n        const stream = tokenizerPunct.stream();\n        for (const chunk of chunks) {\n          stream.pushText(chunk);\n        }\n        stream.endInput();\n        stream.close();\n\n        for (const x of WORDS_PUNCT_EXPECTED) {\n          await stream.next().then((value) => {\n            if (value.value) {\n              expect(value.value.token).toStrictEqual(x);\n            }\n          });\n        }\n      });\n    });\n  });\n  describe('hyphenateWord', () => {\n    it('should hyphenate correctly', () => {\n      HYPHENATOR_TEXT.forEach((x, i) => {\n        expect(hyphenateWord(x)).toStrictEqual(HYPHENATOR_EXPECTED[i]);\n      });\n    });\n  });\n  describe('splitParagraphs', () => {\n    it('should tokenize paragraphs correctly', () => {\n      PARAGRAPH_TEST_CASES.forEach(([a, b]) => {\n        expect(splitParagraphs(a)).toStrictEqual(b);\n      });\n    });\n  });\n});\n"],"mappings":"AAGA,SAAS,UAAU,QAAQ,UAAU;AACrC,SAAS,mBAAmB,eAAe,qBAAqB;AAChE,SAAS,uBAAuB;AAEhC,MAAM,OACJ;AAYF,MAAM,kBAAkB;AAAA,EACtB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,aAAa;AACnB,MAAM,iBAAiB;AAAA,EACrB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,mBACJ;AACF,MAAM,uBAAuB;AAAA,EAC3B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAEA,MAAM,kBAAkB,CAAC,WAAW,YAAY,iBAAiB,UAAU,WAAW,SAAS;AAC/F,MAAM,sBAAsB;AAAA,EAC1B,CAAC,OAAO,MAAM;AAAA,EACd,CAAC,MAAM,QAAQ,IAAI;AAAA,EACnB,CAAC,OAAO,MAAM,MAAM,MAAM,MAAM;AAAA,EAChC,CAAC,OAAO,KAAK;AAAA,EACb,CAAC,OAAO,MAAM;AAAA,EACd,CAAC,OAAO,MAAM;AAChB;AAEA,MAAM,uBAA+D;AAAA,EACnE,CAAC,qBAAqB,CAAC,CAAC,qBAAqB,GAAG,EAAE,CAAC,CAAC;AAAA,EACpD;AAAA,IACE;AAAA,IACA;AAAA,MACE,CAAC,gBAAgB,GAAG,EAAE;AAAA,MACtB,CAAC,gBAAgB,IAAI,EAAE;AAAA,IACzB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,GAAG,EAAE;AAAA,MACjB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA,CAAC,wCAAwC,CAAC,CAAC,oCAAoC,GAAG,EAAE,CAAC,CAAC;AAAA,EACtF,CAAC,yCAAyC,CAAC,CAAC,qCAAqC,GAAG,EAAE,CAAC,CAAC;AAAA,EACxF;AAAA,IACE;AAAA,IACA,CAAC,CAAC,+CAA+C,GAAG,EAAE,CAAC;AAAA,EACzD;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA;AAAA,MACE,CAAC,WAAW,GAAG,CAAC;AAAA,MAChB,CAAC,WAAW,IAAI,EAAE;AAAA,IACpB;AAAA,EACF;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC;AAAA,EACH;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC;AAAA,EACH;AAAA,EACA;AAAA,IACE;AAAA;AAAA,IACA,CAAC,CAAC,0BAA0B,GAAG,EAAE,CAAC;AAAA,EACpC;AACF;AAEA,SAAS,aAAa,MAAM;AAC1B,WAAS,qBAAqB,MAAM;AAClC,UAAM,YAAY,IAAI,kBAAkB;AAExC,OAAG,uCAAuC,MAAM;AAC9C,aAAO,UAAU,SAAS,IAAI,EAAE,MAAM,CAAC,GAAG,MAAM,gBAAgB,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;AAAA,IACxF,CAAC;AAED,OAAG,8CAA8C,YAAY;AAC3D,YAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,UAAI,OAAO;AACX,YAAM,SAAS,CAAC;AAChB,YAAM,cAAc,MAAM,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC,CAAC,EAC1F,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,iBAAW,QAAQ,aAAa;AAC9B,YAAI,CAAC,KAAM;AACX,eAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,eAAO,KAAK,MAAM,IAAI;AAAA,MACxB;AACA,YAAM,SAAS,UAAU,OAAO;AAChC,iBAAW,SAAS,QAAQ;AAC1B,eAAO,SAAS,KAAK;AAAA,MACvB;AACA,aAAO,SAAS;AAChB,aAAO,MAAM;AAEb,iBAAW,KAAK,iBAAiB;AAC/B,cAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,cAAI,MAAM,OAAO;AACf,mBAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,UAC3C;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAAA,EACH,CAAC;AACD,WAAS,iBAAiB,MAAM;AAC9B,UAAM,YAAY,IAAI,cAAc;AAEpC,OAAG,mCAAmC,MAAM;AAC1C,aAAO,UAAU,SAAS,UAAU,EAAE,MAAM,CAAC,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAC,EAAE,WAAW;AAAA,IAC7F,CAAC;AAED,OAAG,0CAA0C,YAAY;AACvD,YAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,UAAI,OAAO;AACX,YAAM,SAAS,CAAC;AAChB,YAAM,cAAc,MAAM,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC,CAAC,EAC1F,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,iBAAW,QAAQ,aAAa;AAC9B,YAAI,CAAC,KAAM;AACX,eAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,eAAO,KAAK,MAAM,IAAI;AAAA,MACxB;AACA,YAAM,SAAS,UAAU,OAAO;AAChC,iBAAW,SAAS,QAAQ;AAC1B,eAAO,SAAS,KAAK;AAAA,MACvB;AACA,aAAO,SAAS;AAChB,aAAO,MAAM;AAEb,iBAAW,KAAK,gBAAgB;AAC9B,cAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,cAAI,MAAM,OAAO;AACf,mBAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,UAC3C;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAED,aAAS,wBAAwB,MAAM;AACrC,YAAM,iBAAiB,IAAI,cAAc,KAAK;AAE9C,SAAG,mCAAmC,MAAM;AAC1C;AAAA,UACE,eAAe,SAAS,gBAAgB,EAAE,MAAM,CAAC,GAAG,MAAM,qBAAqB,CAAC,MAAM,CAAC;AAAA,QACzF,EAAE,WAAW;AAAA,MACf,CAAC;AAED,SAAG,0CAA0C,YAAY;AACvD,cAAM,UAAU,CAAC,GAAG,GAAG,CAAC;AACxB,YAAI,OAAO;AACX,cAAM,SAAS,CAAC;AAChB,cAAM,cAAc;AAAA,UAClB,KAAK,KAAK,KAAK,SAAS,QAAQ,OAAO,CAAC,KAAK,QAAQ,MAAM,KAAK,CAAC,CAAC;AAAA,QACpE,EACG,KAAK,OAAO,EACZ,KAAK,EACL,OAAO,QAAQ,EAAE;AAEpB,mBAAW,QAAQ,aAAa;AAC9B,cAAI,CAAC,KAAM;AACX,iBAAO,KAAK,KAAK,MAAM,QAAW,IAAI,CAAC;AACvC,iBAAO,KAAK,MAAM,IAAI;AAAA,QACxB;AACA,cAAM,SAAS,eAAe,OAAO;AACrC,mBAAW,SAAS,QAAQ;AAC1B,iBAAO,SAAS,KAAK;AAAA,QACvB;AACA,eAAO,SAAS;AAChB,eAAO,MAAM;AAEb,mBAAW,KAAK,sBAAsB;AACpC,gBAAM,OAAO,KAAK,EAAE,KAAK,CAAC,UAAU;AAClC,gBAAI,MAAM,OAAO;AACf,qBAAO,MAAM,MAAM,KAAK,EAAE,cAAc,CAAC;AAAA,YAC3C;AAAA,UACF,CAAC;AAAA,QACH;AAAA,MACF,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACD,WAAS,iBAAiB,MAAM;AAC9B,OAAG,8BAA8B,MAAM;AACrC,sBAAgB,QAAQ,CAAC,GAAG,MAAM;AAChC,eAAO,cAAc,CAAC,CAAC,EAAE,cAAc,oBAAoB,CAAC,CAAC;AAAA,MAC/D,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACD,WAAS,mBAAmB,MAAM;AAChC,OAAG,wCAAwC,MAAM;AAC/C,2BAAqB,QAAQ,CAAC,CAAC,GAAG,CAAC,MAAM;AACvC,eAAO,gBAAgB,CAAC,CAAC,EAAE,cAAc,CAAC;AAAA,MAC5C,CAAC;AAAA,IACH,CAAC;AAAA,EACH,CAAC;AACH,CAAC;","names":[]}

package/dist/tts/stream_adapter.cjs

@@ -22,14 +22,16 @@ __export(stream_adapter_exports, {
   StreamAdapterWrapper: () => StreamAdapterWrapper
 });
 module.exports = __toCommonJS(stream_adapter_exports);
+var import_types = require("../types.cjs");
 var import_utils = require("../utils.cjs");
+var import_io = require("../voice/io.cjs");
 var import_tts = require("./tts.cjs");
 class StreamAdapter extends import_tts.TTS {
   #tts;
   #sentenceTokenizer;
   label;
   constructor(tts, sentenceTokenizer) {
-    super(tts.sampleRate, tts.numChannels, { streaming: true });
+    super(tts.sampleRate, tts.numChannels, { streaming: true, alignedTranscript: true });
     this.#tts = tts;
     this.#sentenceTokenizer = sentenceTokenizer;
     this.label = this.#tts.label;
@@ -59,6 +61,7 @@ class StreamAdapterWrapper extends import_tts.SynthesizeStream {
     this.label = `tts.StreamAdapterWrapper<${this.#tts.label}>`;
   }
   async run() {
+    let cumulativeDuration = 0;
     const forwardInput = async () => {
       for await (const input of this.input) {
         if (this.abortController.signal.aborted) break;
@@ -89,8 +92,19 @@ class StreamAdapterWrapper extends import_tts.SynthesizeStream {
       const audioStream = this.#tts.synthesize(token, this.connOptions, this.abortSignal);
       await (prevTask == null ? void 0 : prevTask.result);
       if (controller.signal.aborted) return;
+      const timedString = (0, import_io.createTimedString)({
+        text: token,
+        startTime: cumulativeDuration
+      });
+      let isFirstFrame = true;
       for await (const audio of audioStream) {
         if (controller.signal.aborted) break;
+        if (isFirstFrame) {
+          audio.frame.userdata[import_types.USERDATA_TIMED_TRANSCRIPT] = [timedString];
+          isFirstFrame = false;
+        }
+        const frameDuration = audio.frame.samplesPerChannel / audio.frame.sampleRate;
+        cumulativeDuration += frameDuration;
         this.queue.put(audio);
       }
     };

package/dist/tts/stream_adapter.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/tts/stream_adapter.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { SentenceStream, SentenceTokenizer } from '../tokenize/index.js';\nimport type { APIConnectOptions } from '../types.js';\nimport { Task } from '../utils.js';\nimport type { ChunkedStream } from './tts.js';\nimport { SynthesizeStream, TTS } from './tts.js';\n\nexport class StreamAdapter extends TTS {\n  #tts: TTS;\n  #sentenceTokenizer: SentenceTokenizer;\n  label: string;\n\n  constructor(tts: TTS, sentenceTokenizer: SentenceTokenizer) {\n    super(tts.sampleRate, tts.numChannels, { streaming: true });\n    this.#tts = tts;\n    this.#sentenceTokenizer = sentenceTokenizer;\n    this.label = this.#tts.label;\n    this.label = `tts.StreamAdapter<${this.#tts.label}>`;\n\n    this.#tts.on('metrics_collected', (metrics) => {\n      this.emit('metrics_collected', metrics);\n    });\n    this.#tts.on('error', (error) => {\n      this.emit('error', error);\n    });\n  }\n\n  synthesize(\n    text: string,\n    connOptions?: APIConnectOptions,\n    abortSignal?: AbortSignal,\n  ): ChunkedStream {\n    return this.#tts.synthesize(text, connOptions, abortSignal);\n  }\n\n  stream(options?: { connOptions?: APIConnectOptions }): StreamAdapterWrapper {\n    return new StreamAdapterWrapper(this.#tts, this.#sentenceTokenizer, options?.connOptions);\n  }\n}\n\nexport class StreamAdapterWrapper extends SynthesizeStream {\n  #tts: TTS;\n  #sentenceStream: SentenceStream;\n  label: string;\n\n  constructor(tts: TTS, sentenceTokenizer: SentenceTokenizer, connOptions?: APIConnectOptions) {\n    super(tts, connOptions);\n    this.#tts = tts;\n    this.#sentenceStream = sentenceTokenizer.stream();\n    this.label = `tts.StreamAdapterWrapper<${this.#tts.label}>`;\n  }\n\n  protected async run() {\n    const forwardInput = async () => {\n      for await (const input of this.input) {\n        if (this.abortController.signal.aborted) break;\n\n        if (input === SynthesizeStream.FLUSH_SENTINEL) {\n          this.#sentenceStream.flush();\n        } else {\n          this.#sentenceStream.pushText(input);\n        }\n      }\n      this.#sentenceStream.endInput();\n      this.#sentenceStream.close();\n    };\n\n    const synthesizeSentenceStream = async () => {\n      let task: Task<void> | undefined;\n      const tokenCompletionTasks: Task<void>[] = [];\n\n      for await (const ev of this.#sentenceStream) {\n        if (this.abortController.signal.aborted) break;\n\n        // this will enable non-blocking synthesis of the stream of tokens\n        task = Task.from(\n          (controller) => synthesize(ev.token, task, controller),\n          this.abortController,\n        );\n\n        tokenCompletionTasks.push(task);\n      }\n\n      await Promise.all(tokenCompletionTasks.map((t) => t.result));\n      this.queue.put(SynthesizeStream.END_OF_STREAM);\n    };\n\n    const synthesize = async (\n      token: string,\n      prevTask: Task<void> | undefined,\n      controller: AbortController,\n    ) => {\n      const audioStream = this.#tts.synthesize(token, this.connOptions, this.abortSignal);\n\n      // wait for previous audio transcription to complete before starting\n      // to queuing audio frames of the current token\n      await prevTask?.result;\n      if (controller.signal.aborted) return;\n\n      for await (const audio of audioStream) {\n        if (controller.signal.aborted) break;\n        this.queue.put(audio);\n      }\n    };\n\n    await Promise.all([forwardInput(), synthesizeSentenceStream()]);\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAKA,mBAAqB;AAErB,iBAAsC;AAE/B,MAAM,sBAAsB,eAAI;AAAA,EACrC;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,mBAAsC;AAC1D,UAAM,IAAI,YAAY,IAAI,aAAa,EAAE,WAAW,KAAK,CAAC;AAC1D,SAAK,OAAO;AACZ,SAAK,qBAAqB;AAC1B,SAAK,QAAQ,KAAK,KAAK;AACvB,SAAK,QAAQ,qBAAqB,KAAK,KAAK,KAAK;AAEjD,SAAK,KAAK,GAAG,qBAAqB,CAAC,YAAY;AAC7C,WAAK,KAAK,qBAAqB,OAAO;AAAA,IACxC,CAAC;AACD,SAAK,KAAK,GAAG,SAAS,CAAC,UAAU;AAC/B,WAAK,KAAK,SAAS,KAAK;AAAA,IAC1B,CAAC;AAAA,EACH;AAAA,EAEA,WACE,MACA,aACA,aACe;AACf,WAAO,KAAK,KAAK,WAAW,MAAM,aAAa,WAAW;AAAA,EAC5D;AAAA,EAEA,OAAO,SAAqE;AAC1E,WAAO,IAAI,qBAAqB,KAAK,MAAM,KAAK,oBAAoB,mCAAS,WAAW;AAAA,EAC1F;AACF;AAEO,MAAM,6BAA6B,4BAAiB;AAAA,EACzD;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,mBAAsC,aAAiC;AAC3F,UAAM,KAAK,WAAW;AACtB,SAAK,OAAO;AACZ,SAAK,kBAAkB,kBAAkB,OAAO;AAChD,SAAK,QAAQ,4BAA4B,KAAK,KAAK,KAAK;AAAA,EAC1D;AAAA,EAEA,MAAgB,MAAM;AACpB,UAAM,eAAe,YAAY;AAC/B,uBAAiB,SAAS,KAAK,OAAO;AACpC,YAAI,KAAK,gBAAgB,OAAO,QAAS;AAEzC,YAAI,UAAU,4BAAiB,gBAAgB;AAC7C,eAAK,gBAAgB,MAAM;AAAA,QAC7B,OAAO;AACL,eAAK,gBAAgB,SAAS,KAAK;AAAA,QACrC;AAAA,MACF;AACA,WAAK,gBAAgB,SAAS;AAC9B,WAAK,gBAAgB,MAAM;AAAA,IAC7B;AAEA,UAAM,2BAA2B,YAAY;AAC3C,UAAI;AACJ,YAAM,uBAAqC,CAAC;AAE5C,uBAAiB,MAAM,KAAK,iBAAiB;AAC3C,YAAI,KAAK,gBAAgB,OAAO,QAAS;AAGzC,eAAO,kBAAK;AAAA,UACV,CAAC,eAAe,WAAW,GAAG,OAAO,MAAM,UAAU;AAAA,UACrD,KAAK;AAAA,QACP;AAEA,6BAAqB,KAAK,IAAI;AAAA,MAChC;AAEA,YAAM,QAAQ,IAAI,qBAAqB,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC;AAC3D,WAAK,MAAM,IAAI,4BAAiB,aAAa;AAAA,IAC/C;AAEA,UAAM,aAAa,OACjB,OACA,UACA,eACG;AACH,YAAM,cAAc,KAAK,KAAK,WAAW,OAAO,KAAK,aAAa,KAAK,WAAW;AAIlF,aAAM,qCAAU;AAChB,UAAI,WAAW,OAAO,QAAS;AAE/B,uBAAiB,SAAS,aAAa;AACrC,YAAI,WAAW,OAAO,QAAS;AAC/B,aAAK,MAAM,IAAI,KAAK;AAAA,MACtB;AAAA,IACF;AAEA,UAAM,QAAQ,IAAI,CAAC,aAAa,GAAG,yBAAyB,CAAC,CAAC;AAAA,EAChE;AACF;","names":[]}
+{"version":3,"sources":["../../src/tts/stream_adapter.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { SentenceStream, SentenceTokenizer } from '../tokenize/index.js';\nimport type { APIConnectOptions } from '../types.js';\nimport { USERDATA_TIMED_TRANSCRIPT } from '../types.js';\nimport { Task } from '../utils.js';\nimport { createTimedString } from '../voice/io.js';\nimport type { ChunkedStream } from './tts.js';\nimport { SynthesizeStream, TTS } from './tts.js';\n\nexport class StreamAdapter extends TTS {\n  #tts: TTS;\n  #sentenceTokenizer: SentenceTokenizer;\n  label: string;\n\n  constructor(tts: TTS, sentenceTokenizer: SentenceTokenizer) {\n    super(tts.sampleRate, tts.numChannels, { streaming: true, alignedTranscript: true });\n    this.#tts = tts;\n    this.#sentenceTokenizer = sentenceTokenizer;\n    this.label = this.#tts.label;\n    this.label = `tts.StreamAdapter<${this.#tts.label}>`;\n\n    this.#tts.on('metrics_collected', (metrics) => {\n      this.emit('metrics_collected', metrics);\n    });\n    this.#tts.on('error', (error) => {\n      this.emit('error', error);\n    });\n  }\n\n  synthesize(\n    text: string,\n    connOptions?: APIConnectOptions,\n    abortSignal?: AbortSignal,\n  ): ChunkedStream {\n    return this.#tts.synthesize(text, connOptions, abortSignal);\n  }\n\n  stream(options?: { connOptions?: APIConnectOptions }): StreamAdapterWrapper {\n    return new StreamAdapterWrapper(this.#tts, this.#sentenceTokenizer, options?.connOptions);\n  }\n}\n\nexport class StreamAdapterWrapper extends SynthesizeStream {\n  #tts: TTS;\n  #sentenceStream: SentenceStream;\n  label: string;\n\n  constructor(tts: TTS, sentenceTokenizer: SentenceTokenizer, connOptions?: APIConnectOptions) {\n    super(tts, connOptions);\n    this.#tts = tts;\n    this.#sentenceStream = sentenceTokenizer.stream();\n    this.label = `tts.StreamAdapterWrapper<${this.#tts.label}>`;\n  }\n\n  protected async run() {\n    let cumulativeDuration = 0;\n\n    const forwardInput = async () => {\n      for await (const input of this.input) {\n        if (this.abortController.signal.aborted) break;\n\n        if (input === SynthesizeStream.FLUSH_SENTINEL) {\n          this.#sentenceStream.flush();\n        } else {\n          this.#sentenceStream.pushText(input);\n        }\n      }\n      this.#sentenceStream.endInput();\n      this.#sentenceStream.close();\n    };\n\n    const synthesizeSentenceStream = async () => {\n      let task: Task<void> | undefined;\n      const tokenCompletionTasks: Task<void>[] = [];\n\n      for await (const ev of this.#sentenceStream) {\n        if (this.abortController.signal.aborted) break;\n\n        // this will enable non-blocking synthesis of the stream of tokens\n        task = Task.from(\n          (controller) => synthesize(ev.token, task, controller),\n          this.abortController,\n        );\n\n        tokenCompletionTasks.push(task);\n      }\n\n      await Promise.all(tokenCompletionTasks.map((t) => t.result));\n      this.queue.put(SynthesizeStream.END_OF_STREAM);\n    };\n\n    const synthesize = async (\n      token: string,\n      prevTask: Task<void> | undefined,\n      controller: AbortController,\n    ) => {\n      const audioStream = this.#tts.synthesize(token, this.connOptions, this.abortSignal);\n\n      // wait for previous audio transcription to complete before starting\n      // to queuing audio frames of the current token\n      await prevTask?.result;\n      if (controller.signal.aborted) return;\n\n      // Create a TimedString with the sentence text and current cumulative duration\n      const timedString = createTimedString({\n        text: token,\n        startTime: cumulativeDuration,\n      });\n\n      let isFirstFrame = true;\n      for await (const audio of audioStream) {\n        if (controller.signal.aborted) break;\n\n        // Attach the TimedString to the first frame of this sentence\n        if (isFirstFrame) {\n          audio.frame.userdata[USERDATA_TIMED_TRANSCRIPT] = [timedString];\n          isFirstFrame = false;\n        }\n\n        // Track cumulative duration\n        const frameDuration = audio.frame.samplesPerChannel / audio.frame.sampleRate;\n        cumulativeDuration += frameDuration;\n\n        this.queue.put(audio);\n      }\n    };\n\n    await Promise.all([forwardInput(), synthesizeSentenceStream()]);\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAKA,mBAA0C;AAC1C,mBAAqB;AACrB,gBAAkC;AAElC,iBAAsC;AAE/B,MAAM,sBAAsB,eAAI;AAAA,EACrC;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,mBAAsC;AAC1D,UAAM,IAAI,YAAY,IAAI,aAAa,EAAE,WAAW,MAAM,mBAAmB,KAAK,CAAC;AACnF,SAAK,OAAO;AACZ,SAAK,qBAAqB;AAC1B,SAAK,QAAQ,KAAK,KAAK;AACvB,SAAK,QAAQ,qBAAqB,KAAK,KAAK,KAAK;AAEjD,SAAK,KAAK,GAAG,qBAAqB,CAAC,YAAY;AAC7C,WAAK,KAAK,qBAAqB,OAAO;AAAA,IACxC,CAAC;AACD,SAAK,KAAK,GAAG,SAAS,CAAC,UAAU;AAC/B,WAAK,KAAK,SAAS,KAAK;AAAA,IAC1B,CAAC;AAAA,EACH;AAAA,EAEA,WACE,MACA,aACA,aACe;AACf,WAAO,KAAK,KAAK,WAAW,MAAM,aAAa,WAAW;AAAA,EAC5D;AAAA,EAEA,OAAO,SAAqE;AAC1E,WAAO,IAAI,qBAAqB,KAAK,MAAM,KAAK,oBAAoB,mCAAS,WAAW;AAAA,EAC1F;AACF;AAEO,MAAM,6BAA6B,4BAAiB;AAAA,EACzD;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,mBAAsC,aAAiC;AAC3F,UAAM,KAAK,WAAW;AACtB,SAAK,OAAO;AACZ,SAAK,kBAAkB,kBAAkB,OAAO;AAChD,SAAK,QAAQ,4BAA4B,KAAK,KAAK,KAAK;AAAA,EAC1D;AAAA,EAEA,MAAgB,MAAM;AACpB,QAAI,qBAAqB;AAEzB,UAAM,eAAe,YAAY;AAC/B,uBAAiB,SAAS,KAAK,OAAO;AACpC,YAAI,KAAK,gBAAgB,OAAO,QAAS;AAEzC,YAAI,UAAU,4BAAiB,gBAAgB;AAC7C,eAAK,gBAAgB,MAAM;AAAA,QAC7B,OAAO;AACL,eAAK,gBAAgB,SAAS,KAAK;AAAA,QACrC;AAAA,MACF;AACA,WAAK,gBAAgB,SAAS;AAC9B,WAAK,gBAAgB,MAAM;AAAA,IAC7B;AAEA,UAAM,2BAA2B,YAAY;AAC3C,UAAI;AACJ,YAAM,uBAAqC,CAAC;AAE5C,uBAAiB,MAAM,KAAK,iBAAiB;AAC3C,YAAI,KAAK,gBAAgB,OAAO,QAAS;AAGzC,eAAO,kBAAK;AAAA,UACV,CAAC,eAAe,WAAW,GAAG,OAAO,MAAM,UAAU;AAAA,UACrD,KAAK;AAAA,QACP;AAEA,6BAAqB,KAAK,IAAI;AAAA,MAChC;AAEA,YAAM,QAAQ,IAAI,qBAAqB,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC;AAC3D,WAAK,MAAM,IAAI,4BAAiB,aAAa;AAAA,IAC/C;AAEA,UAAM,aAAa,OACjB,OACA,UACA,eACG;AACH,YAAM,cAAc,KAAK,KAAK,WAAW,OAAO,KAAK,aAAa,KAAK,WAAW;AAIlF,aAAM,qCAAU;AAChB,UAAI,WAAW,OAAO,QAAS;AAG/B,YAAM,kBAAc,6BAAkB;AAAA,QACpC,MAAM;AAAA,QACN,WAAW;AAAA,MACb,CAAC;AAED,UAAI,eAAe;AACnB,uBAAiB,SAAS,aAAa;AACrC,YAAI,WAAW,OAAO,QAAS;AAG/B,YAAI,cAAc;AAChB,gBAAM,MAAM,SAAS,sCAAyB,IAAI,CAAC,WAAW;AAC9D,yBAAe;AAAA,QACjB;AAGA,cAAM,gBAAgB,MAAM,MAAM,oBAAoB,MAAM,MAAM;AAClE,8BAAsB;AAEtB,aAAK,MAAM,IAAI,KAAK;AAAA,MACtB;AAAA,IACF;AAEA,UAAM,QAAQ,IAAI,CAAC,aAAa,GAAG,yBAAyB,CAAC,CAAC;AAAA,EAChE;AACF;","names":[]}

package/dist/tts/stream_adapter.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"stream_adapter.d.ts","sourceRoot":"","sources":["../../src/tts/stream_adapter.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAkB,iBAAiB,EAAE,MAAM,sBAAsB,CAAC;AAC9E,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,aAAa,CAAC;AAErD,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AAC9C,OAAO,EAAE,gBAAgB,EAAE,GAAG,EAAE,MAAM,UAAU,CAAC;AAEjD,qBAAa,aAAc,SAAQ,GAAG;;IAGpC,KAAK,EAAE,MAAM,CAAC;gBAEF,GAAG,EAAE,GAAG,EAAE,iBAAiB,EAAE,iBAAiB;IAe1D,UAAU,CACR,IAAI,EAAE,MAAM,EACZ,WAAW,CAAC,EAAE,iBAAiB,EAC/B,WAAW,CAAC,EAAE,WAAW,GACxB,aAAa;IAIhB,MAAM,CAAC,OAAO,CAAC,EAAE;QAAE,WAAW,CAAC,EAAE,iBAAiB,CAAA;KAAE,GAAG,oBAAoB;CAG5E;AAED,qBAAa,oBAAqB,SAAQ,gBAAgB;;IAGxD,KAAK,EAAE,MAAM,CAAC;gBAEF,GAAG,EAAE,GAAG,EAAE,iBAAiB,EAAE,iBAAiB,EAAE,WAAW,CAAC,EAAE,iBAAiB;cAO3E,GAAG;CAuDpB"}
+{"version":3,"file":"stream_adapter.d.ts","sourceRoot":"","sources":["../../src/tts/stream_adapter.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAkB,iBAAiB,EAAE,MAAM,sBAAsB,CAAC;AAC9E,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,aAAa,CAAC;AAIrD,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AAC9C,OAAO,EAAE,gBAAgB,EAAE,GAAG,EAAE,MAAM,UAAU,CAAC;AAEjD,qBAAa,aAAc,SAAQ,GAAG;;IAGpC,KAAK,EAAE,MAAM,CAAC;gBAEF,GAAG,EAAE,GAAG,EAAE,iBAAiB,EAAE,iBAAiB;IAe1D,UAAU,CACR,IAAI,EAAE,MAAM,EACZ,WAAW,CAAC,EAAE,iBAAiB,EAC/B,WAAW,CAAC,EAAE,WAAW,GACxB,aAAa;IAIhB,MAAM,CAAC,OAAO,CAAC,EAAE;QAAE,WAAW,CAAC,EAAE,iBAAiB,CAAA;KAAE,GAAG,oBAAoB;CAG5E;AAED,qBAAa,oBAAqB,SAAQ,gBAAgB;;IAGxD,KAAK,EAAE,MAAM,CAAC;gBAEF,GAAG,EAAE,GAAG,EAAE,iBAAiB,EAAE,iBAAiB,EAAE,WAAW,CAAC,EAAE,iBAAiB;cAO3E,GAAG;CA2EpB"}

package/dist/tts/stream_adapter.js

@@ -1,11 +1,13 @@
+import { USERDATA_TIMED_TRANSCRIPT } from "../types.js";
 import { Task } from "../utils.js";
+import { createTimedString } from "../voice/io.js";
 import { SynthesizeStream, TTS } from "./tts.js";
 class StreamAdapter extends TTS {
   #tts;
   #sentenceTokenizer;
   label;
   constructor(tts, sentenceTokenizer) {
-    super(tts.sampleRate, tts.numChannels, { streaming: true });
+    super(tts.sampleRate, tts.numChannels, { streaming: true, alignedTranscript: true });
     this.#tts = tts;
     this.#sentenceTokenizer = sentenceTokenizer;
     this.label = this.#tts.label;
@@ -35,6 +37,7 @@ class StreamAdapterWrapper extends SynthesizeStream {
     this.label = `tts.StreamAdapterWrapper<${this.#tts.label}>`;
   }
   async run() {
+    let cumulativeDuration = 0;
     const forwardInput = async () => {
       for await (const input of this.input) {
         if (this.abortController.signal.aborted) break;
@@ -65,8 +68,19 @@ class StreamAdapterWrapper extends SynthesizeStream {
       const audioStream = this.#tts.synthesize(token, this.connOptions, this.abortSignal);
       await (prevTask == null ? void 0 : prevTask.result);
       if (controller.signal.aborted) return;
+      const timedString = createTimedString({
+        text: token,
+        startTime: cumulativeDuration
+      });
+      let isFirstFrame = true;
       for await (const audio of audioStream) {
         if (controller.signal.aborted) break;
+        if (isFirstFrame) {
+          audio.frame.userdata[USERDATA_TIMED_TRANSCRIPT] = [timedString];
+          isFirstFrame = false;
+        }
+        const frameDuration = audio.frame.samplesPerChannel / audio.frame.sampleRate;
+        cumulativeDuration += frameDuration;
         this.queue.put(audio);
       }
     };

package/dist/tts/stream_adapter.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/tts/stream_adapter.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { SentenceStream, SentenceTokenizer } from '../tokenize/index.js';\nimport type { APIConnectOptions } from '../types.js';\nimport { Task } from '../utils.js';\nimport type { ChunkedStream } from './tts.js';\nimport { SynthesizeStream, TTS } from './tts.js';\n\nexport class StreamAdapter extends TTS {\n  #tts: TTS;\n  #sentenceTokenizer: SentenceTokenizer;\n  label: string;\n\n  constructor(tts: TTS, sentenceTokenizer: SentenceTokenizer) {\n    super(tts.sampleRate, tts.numChannels, { streaming: true });\n    this.#tts = tts;\n    this.#sentenceTokenizer = sentenceTokenizer;\n    this.label = this.#tts.label;\n    this.label = `tts.StreamAdapter<${this.#tts.label}>`;\n\n    this.#tts.on('metrics_collected', (metrics) => {\n      this.emit('metrics_collected', metrics);\n    });\n    this.#tts.on('error', (error) => {\n      this.emit('error', error);\n    });\n  }\n\n  synthesize(\n    text: string,\n    connOptions?: APIConnectOptions,\n    abortSignal?: AbortSignal,\n  ): ChunkedStream {\n    return this.#tts.synthesize(text, connOptions, abortSignal);\n  }\n\n  stream(options?: { connOptions?: APIConnectOptions }): StreamAdapterWrapper {\n    return new StreamAdapterWrapper(this.#tts, this.#sentenceTokenizer, options?.connOptions);\n  }\n}\n\nexport class StreamAdapterWrapper extends SynthesizeStream {\n  #tts: TTS;\n  #sentenceStream: SentenceStream;\n  label: string;\n\n  constructor(tts: TTS, sentenceTokenizer: SentenceTokenizer, connOptions?: APIConnectOptions) {\n    super(tts, connOptions);\n    this.#tts = tts;\n    this.#sentenceStream = sentenceTokenizer.stream();\n    this.label = `tts.StreamAdapterWrapper<${this.#tts.label}>`;\n  }\n\n  protected async run() {\n    const forwardInput = async () => {\n      for await (const input of this.input) {\n        if (this.abortController.signal.aborted) break;\n\n        if (input === SynthesizeStream.FLUSH_SENTINEL) {\n          this.#sentenceStream.flush();\n        } else {\n          this.#sentenceStream.pushText(input);\n        }\n      }\n      this.#sentenceStream.endInput();\n      this.#sentenceStream.close();\n    };\n\n    const synthesizeSentenceStream = async () => {\n      let task: Task<void> | undefined;\n      const tokenCompletionTasks: Task<void>[] = [];\n\n      for await (const ev of this.#sentenceStream) {\n        if (this.abortController.signal.aborted) break;\n\n        // this will enable non-blocking synthesis of the stream of tokens\n        task = Task.from(\n          (controller) => synthesize(ev.token, task, controller),\n          this.abortController,\n        );\n\n        tokenCompletionTasks.push(task);\n      }\n\n      await Promise.all(tokenCompletionTasks.map((t) => t.result));\n      this.queue.put(SynthesizeStream.END_OF_STREAM);\n    };\n\n    const synthesize = async (\n      token: string,\n      prevTask: Task<void> | undefined,\n      controller: AbortController,\n    ) => {\n      const audioStream = this.#tts.synthesize(token, this.connOptions, this.abortSignal);\n\n      // wait for previous audio transcription to complete before starting\n      // to queuing audio frames of the current token\n      await prevTask?.result;\n      if (controller.signal.aborted) return;\n\n      for await (const audio of audioStream) {\n        if (controller.signal.aborted) break;\n        this.queue.put(audio);\n      }\n    };\n\n    await Promise.all([forwardInput(), synthesizeSentenceStream()]);\n  }\n}\n"],"mappings":"AAKA,SAAS,YAAY;AAErB,SAAS,kBAAkB,WAAW;AAE/B,MAAM,sBAAsB,IAAI;AAAA,EACrC;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,mBAAsC;AAC1D,UAAM,IAAI,YAAY,IAAI,aAAa,EAAE,WAAW,KAAK,CAAC;AAC1D,SAAK,OAAO;AACZ,SAAK,qBAAqB;AAC1B,SAAK,QAAQ,KAAK,KAAK;AACvB,SAAK,QAAQ,qBAAqB,KAAK,KAAK,KAAK;AAEjD,SAAK,KAAK,GAAG,qBAAqB,CAAC,YAAY;AAC7C,WAAK,KAAK,qBAAqB,OAAO;AAAA,IACxC,CAAC;AACD,SAAK,KAAK,GAAG,SAAS,CAAC,UAAU;AAC/B,WAAK,KAAK,SAAS,KAAK;AAAA,IAC1B,CAAC;AAAA,EACH;AAAA,EAEA,WACE,MACA,aACA,aACe;AACf,WAAO,KAAK,KAAK,WAAW,MAAM,aAAa,WAAW;AAAA,EAC5D;AAAA,EAEA,OAAO,SAAqE;AAC1E,WAAO,IAAI,qBAAqB,KAAK,MAAM,KAAK,oBAAoB,mCAAS,WAAW;AAAA,EAC1F;AACF;AAEO,MAAM,6BAA6B,iBAAiB;AAAA,EACzD;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,mBAAsC,aAAiC;AAC3F,UAAM,KAAK,WAAW;AACtB,SAAK,OAAO;AACZ,SAAK,kBAAkB,kBAAkB,OAAO;AAChD,SAAK,QAAQ,4BAA4B,KAAK,KAAK,KAAK;AAAA,EAC1D;AAAA,EAEA,MAAgB,MAAM;AACpB,UAAM,eAAe,YAAY;AAC/B,uBAAiB,SAAS,KAAK,OAAO;AACpC,YAAI,KAAK,gBAAgB,OAAO,QAAS;AAEzC,YAAI,UAAU,iBAAiB,gBAAgB;AAC7C,eAAK,gBAAgB,MAAM;AAAA,QAC7B,OAAO;AACL,eAAK,gBAAgB,SAAS,KAAK;AAAA,QACrC;AAAA,MACF;AACA,WAAK,gBAAgB,SAAS;AAC9B,WAAK,gBAAgB,MAAM;AAAA,IAC7B;AAEA,UAAM,2BAA2B,YAAY;AAC3C,UAAI;AACJ,YAAM,uBAAqC,CAAC;AAE5C,uBAAiB,MAAM,KAAK,iBAAiB;AAC3C,YAAI,KAAK,gBAAgB,OAAO,QAAS;AAGzC,eAAO,KAAK;AAAA,UACV,CAAC,eAAe,WAAW,GAAG,OAAO,MAAM,UAAU;AAAA,UACrD,KAAK;AAAA,QACP;AAEA,6BAAqB,KAAK,IAAI;AAAA,MAChC;AAEA,YAAM,QAAQ,IAAI,qBAAqB,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC;AAC3D,WAAK,MAAM,IAAI,iBAAiB,aAAa;AAAA,IAC/C;AAEA,UAAM,aAAa,OACjB,OACA,UACA,eACG;AACH,YAAM,cAAc,KAAK,KAAK,WAAW,OAAO,KAAK,aAAa,KAAK,WAAW;AAIlF,aAAM,qCAAU;AAChB,UAAI,WAAW,OAAO,QAAS;AAE/B,uBAAiB,SAAS,aAAa;AACrC,YAAI,WAAW,OAAO,QAAS;AAC/B,aAAK,MAAM,IAAI,KAAK;AAAA,MACtB;AAAA,IACF;AAEA,UAAM,QAAQ,IAAI,CAAC,aAAa,GAAG,yBAAyB,CAAC,CAAC;AAAA,EAChE;AACF;","names":[]}
+{"version":3,"sources":["../../src/tts/stream_adapter.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { SentenceStream, SentenceTokenizer } from '../tokenize/index.js';\nimport type { APIConnectOptions } from '../types.js';\nimport { USERDATA_TIMED_TRANSCRIPT } from '../types.js';\nimport { Task } from '../utils.js';\nimport { createTimedString } from '../voice/io.js';\nimport type { ChunkedStream } from './tts.js';\nimport { SynthesizeStream, TTS } from './tts.js';\n\nexport class StreamAdapter extends TTS {\n  #tts: TTS;\n  #sentenceTokenizer: SentenceTokenizer;\n  label: string;\n\n  constructor(tts: TTS, sentenceTokenizer: SentenceTokenizer) {\n    super(tts.sampleRate, tts.numChannels, { streaming: true, alignedTranscript: true });\n    this.#tts = tts;\n    this.#sentenceTokenizer = sentenceTokenizer;\n    this.label = this.#tts.label;\n    this.label = `tts.StreamAdapter<${this.#tts.label}>`;\n\n    this.#tts.on('metrics_collected', (metrics) => {\n      this.emit('metrics_collected', metrics);\n    });\n    this.#tts.on('error', (error) => {\n      this.emit('error', error);\n    });\n  }\n\n  synthesize(\n    text: string,\n    connOptions?: APIConnectOptions,\n    abortSignal?: AbortSignal,\n  ): ChunkedStream {\n    return this.#tts.synthesize(text, connOptions, abortSignal);\n  }\n\n  stream(options?: { connOptions?: APIConnectOptions }): StreamAdapterWrapper {\n    return new StreamAdapterWrapper(this.#tts, this.#sentenceTokenizer, options?.connOptions);\n  }\n}\n\nexport class StreamAdapterWrapper extends SynthesizeStream {\n  #tts: TTS;\n  #sentenceStream: SentenceStream;\n  label: string;\n\n  constructor(tts: TTS, sentenceTokenizer: SentenceTokenizer, connOptions?: APIConnectOptions) {\n    super(tts, connOptions);\n    this.#tts = tts;\n    this.#sentenceStream = sentenceTokenizer.stream();\n    this.label = `tts.StreamAdapterWrapper<${this.#tts.label}>`;\n  }\n\n  protected async run() {\n    let cumulativeDuration = 0;\n\n    const forwardInput = async () => {\n      for await (const input of this.input) {\n        if (this.abortController.signal.aborted) break;\n\n        if (input === SynthesizeStream.FLUSH_SENTINEL) {\n          this.#sentenceStream.flush();\n        } else {\n          this.#sentenceStream.pushText(input);\n        }\n      }\n      this.#sentenceStream.endInput();\n      this.#sentenceStream.close();\n    };\n\n    const synthesizeSentenceStream = async () => {\n      let task: Task<void> | undefined;\n      const tokenCompletionTasks: Task<void>[] = [];\n\n      for await (const ev of this.#sentenceStream) {\n        if (this.abortController.signal.aborted) break;\n\n        // this will enable non-blocking synthesis of the stream of tokens\n        task = Task.from(\n          (controller) => synthesize(ev.token, task, controller),\n          this.abortController,\n        );\n\n        tokenCompletionTasks.push(task);\n      }\n\n      await Promise.all(tokenCompletionTasks.map((t) => t.result));\n      this.queue.put(SynthesizeStream.END_OF_STREAM);\n    };\n\n    const synthesize = async (\n      token: string,\n      prevTask: Task<void> | undefined,\n      controller: AbortController,\n    ) => {\n      const audioStream = this.#tts.synthesize(token, this.connOptions, this.abortSignal);\n\n      // wait for previous audio transcription to complete before starting\n      // to queuing audio frames of the current token\n      await prevTask?.result;\n      if (controller.signal.aborted) return;\n\n      // Create a TimedString with the sentence text and current cumulative duration\n      const timedString = createTimedString({\n        text: token,\n        startTime: cumulativeDuration,\n      });\n\n      let isFirstFrame = true;\n      for await (const audio of audioStream) {\n        if (controller.signal.aborted) break;\n\n        // Attach the TimedString to the first frame of this sentence\n        if (isFirstFrame) {\n          audio.frame.userdata[USERDATA_TIMED_TRANSCRIPT] = [timedString];\n          isFirstFrame = false;\n        }\n\n        // Track cumulative duration\n        const frameDuration = audio.frame.samplesPerChannel / audio.frame.sampleRate;\n        cumulativeDuration += frameDuration;\n\n        this.queue.put(audio);\n      }\n    };\n\n    await Promise.all([forwardInput(), synthesizeSentenceStream()]);\n  }\n}\n"],"mappings":"AAKA,SAAS,iCAAiC;AAC1C,SAAS,YAAY;AACrB,SAAS,yBAAyB;AAElC,SAAS,kBAAkB,WAAW;AAE/B,MAAM,sBAAsB,IAAI;AAAA,EACrC;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,mBAAsC;AAC1D,UAAM,IAAI,YAAY,IAAI,aAAa,EAAE,WAAW,MAAM,mBAAmB,KAAK,CAAC;AACnF,SAAK,OAAO;AACZ,SAAK,qBAAqB;AAC1B,SAAK,QAAQ,KAAK,KAAK;AACvB,SAAK,QAAQ,qBAAqB,KAAK,KAAK,KAAK;AAEjD,SAAK,KAAK,GAAG,qBAAqB,CAAC,YAAY;AAC7C,WAAK,KAAK,qBAAqB,OAAO;AAAA,IACxC,CAAC;AACD,SAAK,KAAK,GAAG,SAAS,CAAC,UAAU;AAC/B,WAAK,KAAK,SAAS,KAAK;AAAA,IAC1B,CAAC;AAAA,EACH;AAAA,EAEA,WACE,MACA,aACA,aACe;AACf,WAAO,KAAK,KAAK,WAAW,MAAM,aAAa,WAAW;AAAA,EAC5D;AAAA,EAEA,OAAO,SAAqE;AAC1E,WAAO,IAAI,qBAAqB,KAAK,MAAM,KAAK,oBAAoB,mCAAS,WAAW;AAAA,EAC1F;AACF;AAEO,MAAM,6BAA6B,iBAAiB;AAAA,EACzD;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,mBAAsC,aAAiC;AAC3F,UAAM,KAAK,WAAW;AACtB,SAAK,OAAO;AACZ,SAAK,kBAAkB,kBAAkB,OAAO;AAChD,SAAK,QAAQ,4BAA4B,KAAK,KAAK,KAAK;AAAA,EAC1D;AAAA,EAEA,MAAgB,MAAM;AACpB,QAAI,qBAAqB;AAEzB,UAAM,eAAe,YAAY;AAC/B,uBAAiB,SAAS,KAAK,OAAO;AACpC,YAAI,KAAK,gBAAgB,OAAO,QAAS;AAEzC,YAAI,UAAU,iBAAiB,gBAAgB;AAC7C,eAAK,gBAAgB,MAAM;AAAA,QAC7B,OAAO;AACL,eAAK,gBAAgB,SAAS,KAAK;AAAA,QACrC;AAAA,MACF;AACA,WAAK,gBAAgB,SAAS;AAC9B,WAAK,gBAAgB,MAAM;AAAA,IAC7B;AAEA,UAAM,2BAA2B,YAAY;AAC3C,UAAI;AACJ,YAAM,uBAAqC,CAAC;AAE5C,uBAAiB,MAAM,KAAK,iBAAiB;AAC3C,YAAI,KAAK,gBAAgB,OAAO,QAAS;AAGzC,eAAO,KAAK;AAAA,UACV,CAAC,eAAe,WAAW,GAAG,OAAO,MAAM,UAAU;AAAA,UACrD,KAAK;AAAA,QACP;AAEA,6BAAqB,KAAK,IAAI;AAAA,MAChC;AAEA,YAAM,QAAQ,IAAI,qBAAqB,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC;AAC3D,WAAK,MAAM,IAAI,iBAAiB,aAAa;AAAA,IAC/C;AAEA,UAAM,aAAa,OACjB,OACA,UACA,eACG;AACH,YAAM,cAAc,KAAK,KAAK,WAAW,OAAO,KAAK,aAAa,KAAK,WAAW;AAIlF,aAAM,qCAAU;AAChB,UAAI,WAAW,OAAO,QAAS;AAG/B,YAAM,cAAc,kBAAkB;AAAA,QACpC,MAAM;AAAA,QACN,WAAW;AAAA,MACb,CAAC;AAED,UAAI,eAAe;AACnB,uBAAiB,SAAS,aAAa;AACrC,YAAI,WAAW,OAAO,QAAS;AAG/B,YAAI,cAAc;AAChB,gBAAM,MAAM,SAAS,yBAAyB,IAAI,CAAC,WAAW;AAC9D,yBAAe;AAAA,QACjB;AAGA,cAAM,gBAAgB,MAAM,MAAM,oBAAoB,MAAM,MAAM;AAClE,8BAAsB;AAEtB,aAAK,MAAM,IAAI,KAAK;AAAA,MACtB;AAAA,IACF;AAEA,UAAM,QAAQ,IAAI,CAAC,aAAa,GAAG,yBAAyB,CAAC,CAAC;AAAA,EAChE;AACF;","names":[]}

package/dist/tts/tts.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/tts/tts.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { TypedEventEmitter as TypedEmitter } from '@livekit/typed-emitter';\nimport type { Span } from '@opentelemetry/api';\nimport { EventEmitter } from 'node:events';\nimport type { ReadableStream } from 'node:stream/web';\nimport { APIConnectionError, APIError } from '../_exceptions.js';\nimport { log } from '../log.js';\nimport type { TTSMetrics } from '../metrics/base.js';\nimport { DeferredReadableStream } from '../stream/deferred_stream.js';\nimport { recordException, traceTypes, tracer } from '../telemetry/index.js';\nimport { type APIConnectOptions, DEFAULT_API_CONNECT_OPTIONS, intervalForRetry } from '../types.js';\nimport { AsyncIterableQueue, delay, mergeFrames, startSoon, toError } from '../utils.js';\n\n/** SynthesizedAudio is a packet of speech synthesis as returned by the TTS. */\nexport interface SynthesizedAudio {\n  /** Request ID (one segment could be made up of multiple requests) */\n  requestId: string;\n  /** Segment ID, each segment is separated by a flush */\n  segmentId: string;\n  /** Synthesized audio frame */\n  frame: AudioFrame;\n  /** Current segment of the synthesized audio */\n  deltaText?: string;\n  /** Whether this is the last frame of the segment (streaming only) */\n  final: boolean;\n}\n\n/**\n * Describes the capabilities of the TTS provider.\n *\n * @remarks\n * At present, only `streaming` is supplied to this interface, and the framework only supports\n * providers that do have a streaming endpoint.\n */\nexport interface TTSCapabilities {\n  streaming: boolean;\n}\n\nexport interface TTSError {\n  type: 'tts_error';\n  timestamp: number;\n  label: string;\n  error: Error;\n  recoverable: boolean;\n}\n\nexport type TTSCallbacks = {\n  ['metrics_collected']: (metrics: TTSMetrics) => void;\n  ['error']: (error: TTSError) => void;\n};\n\n/**\n * An instance of a text-to-speech adapter.\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child TTS class, which inherits this class's methods.\n */\nexport abstract class TTS extends (EventEmitter as new () => TypedEmitter<TTSCallbacks>) {\n  #capabilities: TTSCapabilities;\n  #sampleRate: number;\n  #numChannels: number;\n  abstract label: string;\n\n  constructor(sampleRate: number, numChannels: number, capabilities: TTSCapabilities) {\n    super();\n    this.#capabilities = capabilities;\n    this.#sampleRate = sampleRate;\n    this.#numChannels = numChannels;\n  }\n\n  /** Returns this TTS's capabilities */\n  get capabilities(): TTSCapabilities {\n    return this.#capabilities;\n  }\n\n  /** Returns the sample rate of audio frames returned by this TTS */\n  get sampleRate(): number {\n    return this.#sampleRate;\n  }\n\n  /** Returns the channel count of audio frames returned by this TTS */\n  get numChannels(): number {\n    return this.#numChannels;\n  }\n\n  /**\n   * Receives text and returns synthesis in the form of a {@link ChunkedStream}\n   */\n  abstract synthesize(\n    text: string,\n    connOptions?: APIConnectOptions,\n    abortSignal?: AbortSignal,\n  ): ChunkedStream;\n\n  /**\n   * Returns a {@link SynthesizeStream} that can be used to push text and receive audio data\n   *\n   * @param options - Optional configuration including connection options\n   */\n  abstract stream(options?: { connOptions?: APIConnectOptions }): SynthesizeStream;\n\n  async close(): Promise<void> {\n    return;\n  }\n}\n\n/**\n * An instance of a text-to-speech stream, as an asynchronous iterable iterator.\n *\n * @example Looping through frames\n * ```ts\n * for await (const event of stream) {\n *   await source.captureFrame(event.frame);\n * }\n * ```\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child SynthesizeStream class, which inherits this class's methods.\n */\nexport abstract class SynthesizeStream\n  implements AsyncIterableIterator<SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM>\n{\n  protected static readonly FLUSH_SENTINEL = Symbol('FLUSH_SENTINEL');\n  static readonly END_OF_STREAM = Symbol('END_OF_STREAM');\n  protected input = new AsyncIterableQueue<string | typeof SynthesizeStream.FLUSH_SENTINEL>();\n  protected queue = new AsyncIterableQueue<\n    SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM\n  >();\n  protected output = new AsyncIterableQueue<\n    SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM\n  >();\n  protected closed = false;\n  protected connOptions: APIConnectOptions;\n  protected abortController = new AbortController();\n\n  private deferredInputStream: DeferredReadableStream<\n    string | typeof SynthesizeStream.FLUSH_SENTINEL\n  >;\n  private logger = log();\n\n  abstract label: string;\n\n  #tts: TTS;\n  #metricsPendingTexts: string[] = [];\n  #metricsText = '';\n  #monitorMetricsTask?: Promise<void>;\n  #ttsRequestSpan?: Span;\n\n  constructor(tts: TTS, connOptions: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS) {\n    this.#tts = tts;\n    this.connOptions = connOptions;\n    this.deferredInputStream = new DeferredReadableStream();\n    this.pumpInput();\n\n    this.abortController.signal.addEventListener('abort', () => {\n      this.deferredInputStream.detachSource();\n      // TODO (AJS-36) clean this up when we refactor with streams\n      if (!this.input.closed) this.input.close();\n      if (!this.output.closed) this.output.close();\n      this.closed = true;\n    });\n\n    // this is a hack to immitate asyncio.create_task so that mainTask\n    // is run **after** the constructor has finished. Otherwise we get\n    // runtime error when trying to access class variables in the\n    // `run` method.\n    startSoon(() => this.mainTask().finally(() => this.queue.close()));\n  }\n\n  private _mainTaskImpl = async (span: Span) => {\n    this.#ttsRequestSpan = span;\n    span.setAttributes({\n      [traceTypes.ATTR_TTS_STREAMING]: true,\n      [traceTypes.ATTR_TTS_LABEL]: this.#tts.label,\n    });\n\n    for (let i = 0; i < this.connOptions.maxRetry + 1; i++) {\n      try {\n        return await tracer.startActiveSpan(\n          async (attemptSpan) => {\n            attemptSpan.setAttribute(traceTypes.ATTR_RETRY_COUNT, i);\n            try {\n              return await this.run();\n            } catch (error) {\n              recordException(attemptSpan, toError(error));\n              throw error;\n            }\n          },\n          { name: 'tts_request_run' },\n        );\n      } catch (error) {\n        if (error instanceof APIError) {\n          const retryInterval = intervalForRetry(this.connOptions, i);\n\n          if (this.connOptions.maxRetry === 0 || !error.retryable) {\n            this.emitError({ error, recoverable: false });\n            throw error;\n          } else if (i === this.connOptions.maxRetry) {\n            this.emitError({ error, recoverable: false });\n            throw new APIConnectionError({\n              message: `failed to generate TTS completion after ${this.connOptions.maxRetry + 1} attempts`,\n              options: { retryable: false },\n            });\n          } else {\n            // Don't emit error event for recoverable errors during retry loop\n            // to avoid ERR_UNHANDLED_ERROR or premature session termination\n            this.logger.warn(\n              { tts: this.#tts.label, attempt: i + 1, error },\n              `failed to synthesize speech, retrying in  ${retryInterval}s`,\n            );\n          }\n\n          if (retryInterval > 0) {\n            await delay(retryInterval);\n          }\n        } else {\n          this.emitError({ error: toError(error), recoverable: false });\n          throw error;\n        }\n      }\n    }\n  };\n\n  private mainTask = async () =>\n    tracer.startActiveSpan(async (span) => this._mainTaskImpl(span), {\n      name: 'tts_request',\n      endOnExit: false,\n    });\n\n  private emitError({ error, recoverable }: { error: Error; recoverable: boolean }) {\n    this.#tts.emit('error', {\n      type: 'tts_error',\n      timestamp: Date.now(),\n      label: this.#tts.label,\n      error,\n      recoverable,\n    });\n  }\n\n  // NOTE(AJS-37): The implementation below uses an AsyncIterableQueue (`this.input`)\n  // bridged from a DeferredReadableStream (`this.deferredInputStream`) rather than\n  // consuming the stream directly.\n  //\n  // A full refactor to native Web Streams was considered but is currently deferred.\n  // The primary reason is to maintain architectural parity with the Python SDK,\n  // which is a key design goal for the project. This ensures a consistent developer\n  // experience across both platforms.\n  //\n  // For more context, see the discussion in GitHub issue # 844.\n  protected async pumpInput() {\n    const reader = this.deferredInputStream.stream.getReader();\n    try {\n      while (true) {\n        const { done, value } = await reader.read();\n        if (done || value === SynthesizeStream.FLUSH_SENTINEL) {\n          break;\n        }\n        this.pushText(value);\n      }\n      this.endInput();\n    } catch (error) {\n      this.logger.error(error, 'Error reading deferred input stream');\n    } finally {\n      reader.releaseLock();\n      // Ensure output is closed when the stream ends\n      if (!this.#monitorMetricsTask) {\n        // No text was received, close the output directly\n        this.output.close();\n      }\n    }\n  }\n\n  protected async monitorMetrics() {\n    const startTime = process.hrtime.bigint();\n    let audioDurationMs = 0;\n    let ttfb: bigint = BigInt(-1);\n    let requestId = '';\n\n    const emit = () => {\n      if (this.#metricsPendingTexts.length) {\n        const text = this.#metricsPendingTexts.shift()!;\n        const duration = process.hrtime.bigint() - startTime;\n        const roundedAudioDurationMs = Math.round(audioDurationMs);\n        const metrics: TTSMetrics = {\n          type: 'tts_metrics',\n          timestamp: Date.now(),\n          requestId,\n          ttfbMs: ttfb === BigInt(-1) ? -1 : Math.trunc(Number(ttfb / BigInt(1000000))),\n          durationMs: Math.trunc(Number(duration / BigInt(1000000))),\n          charactersCount: text.length,\n          audioDurationMs: roundedAudioDurationMs,\n          cancelled: this.abortController.signal.aborted,\n          label: this.#tts.label,\n          streamed: false,\n        };\n        if (this.#ttsRequestSpan) {\n          this.#ttsRequestSpan.setAttribute(traceTypes.ATTR_TTS_METRICS, JSON.stringify(metrics));\n        }\n        this.#tts.emit('metrics_collected', metrics);\n      }\n    };\n\n    for await (const audio of this.queue) {\n      if (this.abortController.signal.aborted) {\n        break;\n      }\n      this.output.put(audio);\n      if (audio === SynthesizeStream.END_OF_STREAM) continue;\n      requestId = audio.requestId;\n      if (ttfb === BigInt(-1)) {\n        ttfb = process.hrtime.bigint() - startTime;\n      }\n      // TODO(AJS-102): use frame.durationMs once available in rtc-node\n      audioDurationMs += (audio.frame.samplesPerChannel / audio.frame.sampleRate) * 1000;\n      if (audio.final) {\n        emit();\n      }\n    }\n\n    if (requestId) {\n      emit();\n    }\n\n    if (this.#ttsRequestSpan) {\n      this.#ttsRequestSpan.end();\n      this.#ttsRequestSpan = undefined;\n    }\n  }\n\n  protected abstract run(): Promise<void>;\n\n  updateInputStream(text: ReadableStream<string>) {\n    this.deferredInputStream.setSource(text);\n  }\n\n  /** Push a string of text to the TTS */\n  /** @deprecated Use `updateInputStream` instead */\n  pushText(text: string) {\n    if (!this.#monitorMetricsTask) {\n      this.#monitorMetricsTask = this.monitorMetrics();\n      // Close output when metrics task completes\n      this.#monitorMetricsTask.finally(() => this.output.close());\n    }\n    this.#metricsText += text;\n\n    if (this.input.closed || this.closed) {\n      // Stream was aborted/closed, silently skip\n      return;\n    }\n\n    this.input.put(text);\n  }\n\n  /** Flush the TTS, causing it to process all pending text */\n  flush() {\n    if (this.#metricsText) {\n      this.#metricsPendingTexts.push(this.#metricsText);\n      this.#metricsText = '';\n    }\n\n    if (this.input.closed || this.closed) {\n      // Stream was aborted/closed, silently skip\n      return;\n    }\n\n    this.input.put(SynthesizeStream.FLUSH_SENTINEL);\n  }\n\n  /** Mark the input as ended and forbid additional pushes */\n  endInput() {\n    this.flush();\n\n    if (this.input.closed || this.closed) {\n      // Stream was aborted/closed, silently skip\n      return;\n    }\n\n    this.input.close();\n  }\n\n  next(): Promise<IteratorResult<SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM>> {\n    return this.output.next();\n  }\n\n  get abortSignal(): AbortSignal {\n    return this.abortController.signal;\n  }\n\n  /** Close both the input and output of the TTS stream */\n  close() {\n    this.abortController.abort();\n  }\n\n  [Symbol.asyncIterator](): SynthesizeStream {\n    return this;\n  }\n}\n\n/**\n * An instance of a text-to-speech response, as an asynchronous iterable iterator.\n *\n * @example Looping through frames\n * ```ts\n * for await (const event of stream) {\n *   await source.captureFrame(event.frame);\n * }\n * ```\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child ChunkedStream class, which inherits this class's methods.\n */\nexport abstract class ChunkedStream implements AsyncIterableIterator<SynthesizedAudio> {\n  protected queue = new AsyncIterableQueue<SynthesizedAudio>();\n  protected output = new AsyncIterableQueue<SynthesizedAudio>();\n  protected closed = false;\n  abstract label: string;\n  #text: string;\n  #tts: TTS;\n  #ttsRequestSpan?: Span;\n  private _connOptions: APIConnectOptions;\n  private logger = log();\n\n  protected abortController = new AbortController();\n\n  constructor(\n    text: string,\n    tts: TTS,\n    connOptions: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,\n    abortSignal?: AbortSignal,\n  ) {\n    this.#text = text;\n    this.#tts = tts;\n    this._connOptions = connOptions;\n\n    if (abortSignal) {\n      abortSignal.addEventListener('abort', () => this.abortController.abort(), { once: true });\n    }\n\n    this.monitorMetrics();\n\n    // this is a hack to immitate asyncio.create_task so that mainTask\n    // is run **after** the constructor has finished. Otherwise we get\n    // runtime error when trying to access class variables in the\n    // `run` method.\n    Promise.resolve().then(() => this.mainTask().finally(() => this.queue.close()));\n  }\n\n  private _mainTaskImpl = async (span: Span) => {\n    this.#ttsRequestSpan = span;\n    span.setAttributes({\n      [traceTypes.ATTR_TTS_STREAMING]: false,\n      [traceTypes.ATTR_TTS_LABEL]: this.#tts.label,\n    });\n\n    for (let i = 0; i < this._connOptions.maxRetry + 1; i++) {\n      try {\n        return await tracer.startActiveSpan(\n          async (attemptSpan) => {\n            attemptSpan.setAttribute(traceTypes.ATTR_RETRY_COUNT, i);\n            try {\n              return await this.run();\n            } catch (error) {\n              recordException(attemptSpan, toError(error));\n              throw error;\n            }\n          },\n          { name: 'tts_request_run' },\n        );\n      } catch (error) {\n        if (error instanceof APIError) {\n          const retryInterval = intervalForRetry(this._connOptions, i);\n\n          if (this._connOptions.maxRetry === 0 || !error.retryable) {\n            this.emitError({ error, recoverable: false });\n            throw error;\n          } else if (i === this._connOptions.maxRetry) {\n            this.emitError({ error, recoverable: false });\n            throw new APIConnectionError({\n              message: `failed to generate TTS completion after ${this._connOptions.maxRetry + 1} attempts`,\n              options: { retryable: false },\n            });\n          } else {\n            // Don't emit error event for recoverable errors during retry loop\n            // to avoid ERR_UNHANDLED_ERROR or premature session termination\n            this.logger.warn(\n              { tts: this.#tts.label, attempt: i + 1, error },\n              `failed to generate TTS completion, retrying in ${retryInterval}s`,\n            );\n          }\n\n          if (retryInterval > 0) {\n            await delay(retryInterval);\n          }\n        } else {\n          this.emitError({ error: toError(error), recoverable: false });\n          throw error;\n        }\n      }\n    }\n  };\n\n  private async mainTask() {\n    return tracer.startActiveSpan(async (span) => this._mainTaskImpl(span), {\n      name: 'tts_request',\n      endOnExit: false,\n    });\n  }\n\n  private emitError({ error, recoverable }: { error: Error; recoverable: boolean }) {\n    this.#tts.emit('error', {\n      type: 'tts_error',\n      timestamp: Date.now(),\n      label: this.#tts.label,\n      error,\n      recoverable,\n    });\n  }\n\n  protected abstract run(): Promise<void>;\n\n  get inputText(): string {\n    return this.#text;\n  }\n\n  get abortSignal(): AbortSignal {\n    return this.abortController.signal;\n  }\n\n  protected async monitorMetrics() {\n    const startTime = process.hrtime.bigint();\n    let audioDurationMs = 0;\n    let ttfb: bigint = BigInt(-1);\n    let requestId = '';\n\n    for await (const audio of this.queue) {\n      this.output.put(audio);\n      requestId = audio.requestId;\n      if (ttfb === BigInt(-1)) {\n        ttfb = process.hrtime.bigint() - startTime;\n      }\n      audioDurationMs += (audio.frame.samplesPerChannel / audio.frame.sampleRate) * 1000;\n    }\n    this.output.close();\n\n    const duration = process.hrtime.bigint() - startTime;\n    const metrics: TTSMetrics = {\n      type: 'tts_metrics',\n      timestamp: Date.now(),\n      requestId,\n      ttfbMs: ttfb === BigInt(-1) ? -1 : Math.trunc(Number(ttfb / BigInt(1000000))),\n      durationMs: Math.trunc(Number(duration / BigInt(1000000))),\n      charactersCount: this.#text.length,\n      audioDurationMs: Math.round(audioDurationMs),\n      cancelled: false, // TODO(AJS-186): support ChunkedStream with 1.0 - add this.abortController.signal.aborted here\n      label: this.#tts.label,\n      streamed: false,\n    };\n\n    if (this.#ttsRequestSpan) {\n      this.#ttsRequestSpan.setAttribute(traceTypes.ATTR_TTS_METRICS, JSON.stringify(metrics));\n      this.#ttsRequestSpan.end();\n      this.#ttsRequestSpan = undefined;\n    }\n\n    this.#tts.emit('metrics_collected', metrics);\n  }\n\n  /** Collect every frame into one in a single call */\n  async collect(): Promise<AudioFrame> {\n    const frames = [];\n    for await (const event of this) {\n      frames.push(event.frame);\n    }\n    return mergeFrames(frames);\n  }\n\n  next(): Promise<IteratorResult<SynthesizedAudio>> {\n    return this.output.next();\n  }\n\n  /** Close both the input and output of the TTS stream */\n  close() {\n    if (!this.queue.closed) this.queue.close();\n    if (!this.output.closed) this.output.close();\n    if (!this.abortController.signal.aborted) this.abortController.abort();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): ChunkedStream {\n    return this;\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAMA,yBAA6B;AAE7B,wBAA6C;AAC7C,iBAAoB;AAEpB,6BAAuC;AACvC,uBAAoD;AACpD,mBAAsF;AACtF,mBAA2E;AA+CpE,MAAe,YAAa,gCAAsD;AAAA,EACvF;AAAA,EACA;AAAA,EACA;AAAA,EAGA,YAAY,YAAoB,aAAqB,cAA+B;AAClF,UAAM;AACN,SAAK,gBAAgB;AACrB,SAAK,cAAc;AACnB,SAAK,eAAe;AAAA,EACtB;AAAA;AAAA,EAGA,IAAI,eAAgC;AAClC,WAAO,KAAK;AAAA,EACd;AAAA;AAAA,EAGA,IAAI,aAAqB;AACvB,WAAO,KAAK;AAAA,EACd;AAAA;AAAA,EAGA,IAAI,cAAsB;AACxB,WAAO,KAAK;AAAA,EACd;AAAA,EAkBA,MAAM,QAAuB;AAC3B;AAAA,EACF;AACF;AAgBO,MAAe,iBAEtB;AAAA,EACE,OAA0B,iBAAiB,OAAO,gBAAgB;AAAA,EAClE,OAAgB,gBAAgB,OAAO,eAAe;AAAA,EAC5C,QAAQ,IAAI,gCAAoE;AAAA,EAChF,QAAQ,IAAI,gCAEpB;AAAA,EACQ,SAAS,IAAI,gCAErB;AAAA,EACQ,SAAS;AAAA,EACT;AAAA,EACA,kBAAkB,IAAI,gBAAgB;AAAA,EAExC;AAAA,EAGA,aAAS,gBAAI;AAAA,EAIrB;AAAA,EACA,uBAAiC,CAAC;AAAA,EAClC,eAAe;AAAA,EACf;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,cAAiC,0CAA6B;AAClF,SAAK,OAAO;AACZ,SAAK,cAAc;AACnB,SAAK,sBAAsB,IAAI,8CAAuB;AACtD,SAAK,UAAU;AAEf,SAAK,gBAAgB,OAAO,iBAAiB,SAAS,MAAM;AAC1D,WAAK,oBAAoB,aAAa;AAEtC,UAAI,CAAC,KAAK,MAAM,OAAQ,MAAK,MAAM,MAAM;AACzC,UAAI,CAAC,KAAK,OAAO,OAAQ,MAAK,OAAO,MAAM;AAC3C,WAAK,SAAS;AAAA,IAChB,CAAC;AAMD,gCAAU,MAAM,KAAK,SAAS,EAAE,QAAQ,MAAM,KAAK,MAAM,MAAM,CAAC,CAAC;AAAA,EACnE;AAAA,EAEQ,gBAAgB,OAAO,SAAe;AAC5C,SAAK,kBAAkB;AACvB,SAAK,cAAc;AAAA,MACjB,CAAC,4BAAW,kBAAkB,GAAG;AAAA,MACjC,CAAC,4BAAW,cAAc,GAAG,KAAK,KAAK;AAAA,IACzC,CAAC;AAED,aAAS,IAAI,GAAG,IAAI,KAAK,YAAY,WAAW,GAAG,KAAK;AACtD,UAAI;AACF,eAAO,MAAM,wBAAO;AAAA,UAClB,OAAO,gBAAgB;AACrB,wBAAY,aAAa,4BAAW,kBAAkB,CAAC;AACvD,gBAAI;AACF,qBAAO,MAAM,KAAK,IAAI;AAAA,YACxB,SAAS,OAAO;AACd,oDAAgB,iBAAa,sBAAQ,KAAK,CAAC;AAC3C,oBAAM;AAAA,YACR;AAAA,UACF;AAAA,UACA,EAAE,MAAM,kBAAkB;AAAA,QAC5B;AAAA,MACF,SAAS,OAAO;AACd,YAAI,iBAAiB,4BAAU;AAC7B,gBAAM,oBAAgB,+BAAiB,KAAK,aAAa,CAAC;AAE1D,cAAI,KAAK,YAAY,aAAa,KAAK,CAAC,MAAM,WAAW;AACvD,iBAAK,UAAU,EAAE,OAAO,aAAa,MAAM,CAAC;AAC5C,kBAAM;AAAA,UACR,WAAW,MAAM,KAAK,YAAY,UAAU;AAC1C,iBAAK,UAAU,EAAE,OAAO,aAAa,MAAM,CAAC;AAC5C,kBAAM,IAAI,qCAAmB;AAAA,cAC3B,SAAS,2CAA2C,KAAK,YAAY,WAAW,CAAC;AAAA,cACjF,SAAS,EAAE,WAAW,MAAM;AAAA,YAC9B,CAAC;AAAA,UACH,OAAO;AAGL,iBAAK,OAAO;AAAA,cACV,EAAE,KAAK,KAAK,KAAK,OAAO,SAAS,IAAI,GAAG,MAAM;AAAA,cAC9C,6CAA6C,aAAa;AAAA,YAC5D;AAAA,UACF;AAEA,cAAI,gBAAgB,GAAG;AACrB,sBAAM,oBAAM,aAAa;AAAA,UAC3B;AAAA,QACF,OAAO;AACL,eAAK,UAAU,EAAE,WAAO,sBAAQ,KAAK,GAAG,aAAa,MAAM,CAAC;AAC5D,gBAAM;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA,EAEQ,WAAW,YACjB,wBAAO,gBAAgB,OAAO,SAAS,KAAK,cAAc,IAAI,GAAG;AAAA,IAC/D,MAAM;AAAA,IACN,WAAW;AAAA,EACb,CAAC;AAAA,EAEK,UAAU,EAAE,OAAO,YAAY,GAA2C;AAChF,SAAK,KAAK,KAAK,SAAS;AAAA,MACtB,MAAM;AAAA,MACN,WAAW,KAAK,IAAI;AAAA,MACpB,OAAO,KAAK,KAAK;AAAA,MACjB;AAAA,MACA;AAAA,IACF,CAAC;AAAA,EACH;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAYA,MAAgB,YAAY;AAC1B,UAAM,SAAS,KAAK,oBAAoB,OAAO,UAAU;AACzD,QAAI;AACF,aAAO,MAAM;AACX,cAAM,EAAE,MAAM,MAAM,IAAI,MAAM,OAAO,KAAK;AAC1C,YAAI,QAAQ,UAAU,iBAAiB,gBAAgB;AACrD;AAAA,QACF;AACA,aAAK,SAAS,KAAK;AAAA,MACrB;AACA,WAAK,SAAS;AAAA,IAChB,SAAS,OAAO;AACd,WAAK,OAAO,MAAM,OAAO,qCAAqC;AAAA,IAChE,UAAE;AACA,aAAO,YAAY;AAEnB,UAAI,CAAC,KAAK,qBAAqB;AAE7B,aAAK,OAAO,MAAM;AAAA,MACpB;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAgB,iBAAiB;AAC/B,UAAM,YAAY,QAAQ,OAAO,OAAO;AACxC,QAAI,kBAAkB;AACtB,QAAI,OAAe,OAAO,EAAE;AAC5B,QAAI,YAAY;AAEhB,UAAM,OAAO,MAAM;AACjB,UAAI,KAAK,qBAAqB,QAAQ;AACpC,cAAM,OAAO,KAAK,qBAAqB,MAAM;AAC7C,cAAM,WAAW,QAAQ,OAAO,OAAO,IAAI;AAC3C,cAAM,yBAAyB,KAAK,MAAM,eAAe;AACzD,cAAM,UAAsB;AAAA,UAC1B,MAAM;AAAA,UACN,WAAW,KAAK,IAAI;AAAA,UACpB;AAAA,UACA,QAAQ,SAAS,OAAO,EAAE,IAAI,KAAK,KAAK,MAAM,OAAO,OAAO,OAAO,GAAO,CAAC,CAAC;AAAA,UAC5E,YAAY,KAAK,MAAM,OAAO,WAAW,OAAO,GAAO,CAAC,CAAC;AAAA,UACzD,iBAAiB,KAAK;AAAA,UACtB,iBAAiB;AAAA,UACjB,WAAW,KAAK,gBAAgB,OAAO;AAAA,UACvC,OAAO,KAAK,KAAK;AAAA,UACjB,UAAU;AAAA,QACZ;AACA,YAAI,KAAK,iBAAiB;AACxB,eAAK,gBAAgB,aAAa,4BAAW,kBAAkB,KAAK,UAAU,OAAO,CAAC;AAAA,QACxF;AACA,aAAK,KAAK,KAAK,qBAAqB,OAAO;AAAA,MAC7C;AAAA,IACF;AAEA,qBAAiB,SAAS,KAAK,OAAO;AACpC,UAAI,KAAK,gBAAgB,OAAO,SAAS;AACvC;AAAA,MACF;AACA,WAAK,OAAO,IAAI,KAAK;AACrB,UAAI,UAAU,iBAAiB,cAAe;AAC9C,kBAAY,MAAM;AAClB,UAAI,SAAS,OAAO,EAAE,GAAG;AACvB,eAAO,QAAQ,OAAO,OAAO,IAAI;AAAA,MACnC;AAEA,yBAAoB,MAAM,MAAM,oBAAoB,MAAM,MAAM,aAAc;AAC9E,UAAI,MAAM,OAAO;AACf,aAAK;AAAA,MACP;AAAA,IACF;AAEA,QAAI,WAAW;AACb,WAAK;AAAA,IACP;AAEA,QAAI,KAAK,iBAAiB;AACxB,WAAK,gBAAgB,IAAI;AACzB,WAAK,kBAAkB;AAAA,IACzB;AAAA,EACF;AAAA,EAIA,kBAAkB,MAA8B;AAC9C,SAAK,oBAAoB,UAAU,IAAI;AAAA,EACzC;AAAA;AAAA;AAAA,EAIA,SAAS,MAAc;AACrB,QAAI,CAAC,KAAK,qBAAqB;AAC7B,WAAK,sBAAsB,KAAK,eAAe;AAE/C,WAAK,oBAAoB,QAAQ,MAAM,KAAK,OAAO,MAAM,CAAC;AAAA,IAC5D;AACA,SAAK,gBAAgB;AAErB,QAAI,KAAK,MAAM,UAAU,KAAK,QAAQ;AAEpC;AAAA,IACF;AAEA,SAAK,MAAM,IAAI,IAAI;AAAA,EACrB;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,KAAK,cAAc;AACrB,WAAK,qBAAqB,KAAK,KAAK,YAAY;AAChD,WAAK,eAAe;AAAA,IACtB;AAEA,QAAI,KAAK,MAAM,UAAU,KAAK,QAAQ;AAEpC;AAAA,IACF;AAEA,SAAK,MAAM,IAAI,iBAAiB,cAAc;AAAA,EAChD;AAAA;AAAA,EAGA,WAAW;AACT,SAAK,MAAM;AAEX,QAAI,KAAK,MAAM,UAAU,KAAK,QAAQ;AAEpC;AAAA,IACF;AAEA,SAAK,MAAM,MAAM;AAAA,EACnB;AAAA,EAEA,OAA0F;AACxF,WAAO,KAAK,OAAO,KAAK;AAAA,EAC1B;AAAA,EAEA,IAAI,cAA2B;AAC7B,WAAO,KAAK,gBAAgB;AAAA,EAC9B;AAAA;AAAA,EAGA,QAAQ;AACN,SAAK,gBAAgB,MAAM;AAAA,EAC7B;AAAA,EAEA,CAAC,OAAO,aAAa,IAAsB;AACzC,WAAO;AAAA,EACT;AACF;AAgBO,MAAe,cAAiE;AAAA,EAC3E,QAAQ,IAAI,gCAAqC;AAAA,EACjD,SAAS,IAAI,gCAAqC;AAAA,EAClD,SAAS;AAAA,EAEnB;AAAA,EACA;AAAA,EACA;AAAA,EACQ;AAAA,EACA,aAAS,gBAAI;AAAA,EAEX,kBAAkB,IAAI,gBAAgB;AAAA,EAEhD,YACE,MACA,KACA,cAAiC,0CACjC,aACA;AACA,SAAK,QAAQ;AACb,SAAK,OAAO;AACZ,SAAK,eAAe;AAEpB,QAAI,aAAa;AACf,kBAAY,iBAAiB,SAAS,MAAM,KAAK,gBAAgB,MAAM,GAAG,EAAE,MAAM,KAAK,CAAC;AAAA,IAC1F;AAEA,SAAK,eAAe;AAMpB,YAAQ,QAAQ,EAAE,KAAK,MAAM,KAAK,SAAS,EAAE,QAAQ,MAAM,KAAK,MAAM,MAAM,CAAC,CAAC;AAAA,EAChF;AAAA,EAEQ,gBAAgB,OAAO,SAAe;AAC5C,SAAK,kBAAkB;AACvB,SAAK,cAAc;AAAA,MACjB,CAAC,4BAAW,kBAAkB,GAAG;AAAA,MACjC,CAAC,4BAAW,cAAc,GAAG,KAAK,KAAK;AAAA,IACzC,CAAC;AAED,aAAS,IAAI,GAAG,IAAI,KAAK,aAAa,WAAW,GAAG,KAAK;AACvD,UAAI;AACF,eAAO,MAAM,wBAAO;AAAA,UAClB,OAAO,gBAAgB;AACrB,wBAAY,aAAa,4BAAW,kBAAkB,CAAC;AACvD,gBAAI;AACF,qBAAO,MAAM,KAAK,IAAI;AAAA,YACxB,SAAS,OAAO;AACd,oDAAgB,iBAAa,sBAAQ,KAAK,CAAC;AAC3C,oBAAM;AAAA,YACR;AAAA,UACF;AAAA,UACA,EAAE,MAAM,kBAAkB;AAAA,QAC5B;AAAA,MACF,SAAS,OAAO;AACd,YAAI,iBAAiB,4BAAU;AAC7B,gBAAM,oBAAgB,+BAAiB,KAAK,cAAc,CAAC;AAE3D,cAAI,KAAK,aAAa,aAAa,KAAK,CAAC,MAAM,WAAW;AACxD,iBAAK,UAAU,EAAE,OAAO,aAAa,MAAM,CAAC;AAC5C,kBAAM;AAAA,UACR,WAAW,MAAM,KAAK,aAAa,UAAU;AAC3C,iBAAK,UAAU,EAAE,OAAO,aAAa,MAAM,CAAC;AAC5C,kBAAM,IAAI,qCAAmB;AAAA,cAC3B,SAAS,2CAA2C,KAAK,aAAa,WAAW,CAAC;AAAA,cAClF,SAAS,EAAE,WAAW,MAAM;AAAA,YAC9B,CAAC;AAAA,UACH,OAAO;AAGL,iBAAK,OAAO;AAAA,cACV,EAAE,KAAK,KAAK,KAAK,OAAO,SAAS,IAAI,GAAG,MAAM;AAAA,cAC9C,kDAAkD,aAAa;AAAA,YACjE;AAAA,UACF;AAEA,cAAI,gBAAgB,GAAG;AACrB,sBAAM,oBAAM,aAAa;AAAA,UAC3B;AAAA,QACF,OAAO;AACL,eAAK,UAAU,EAAE,WAAO,sBAAQ,KAAK,GAAG,aAAa,MAAM,CAAC;AAC5D,gBAAM;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAc,WAAW;AACvB,WAAO,wBAAO,gBAAgB,OAAO,SAAS,KAAK,cAAc,IAAI,GAAG;AAAA,MACtE,MAAM;AAAA,MACN,WAAW;AAAA,IACb,CAAC;AAAA,EACH;AAAA,EAEQ,UAAU,EAAE,OAAO,YAAY,GAA2C;AAChF,SAAK,KAAK,KAAK,SAAS;AAAA,MACtB,MAAM;AAAA,MACN,WAAW,KAAK,IAAI;AAAA,MACpB,OAAO,KAAK,KAAK;AAAA,MACjB;AAAA,MACA;AAAA,IACF,CAAC;AAAA,EACH;AAAA,EAIA,IAAI,YAAoB;AACtB,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,IAAI,cAA2B;AAC7B,WAAO,KAAK,gBAAgB;AAAA,EAC9B;AAAA,EAEA,MAAgB,iBAAiB;AAC/B,UAAM,YAAY,QAAQ,OAAO,OAAO;AACxC,QAAI,kBAAkB;AACtB,QAAI,OAAe,OAAO,EAAE;AAC5B,QAAI,YAAY;AAEhB,qBAAiB,SAAS,KAAK,OAAO;AACpC,WAAK,OAAO,IAAI,KAAK;AACrB,kBAAY,MAAM;AAClB,UAAI,SAAS,OAAO,EAAE,GAAG;AACvB,eAAO,QAAQ,OAAO,OAAO,IAAI;AAAA,MACnC;AACA,yBAAoB,MAAM,MAAM,oBAAoB,MAAM,MAAM,aAAc;AAAA,IAChF;AACA,SAAK,OAAO,MAAM;AAElB,UAAM,WAAW,QAAQ,OAAO,OAAO,IAAI;AAC3C,UAAM,UAAsB;AAAA,MAC1B,MAAM;AAAA,MACN,WAAW,KAAK,IAAI;AAAA,MACpB;AAAA,MACA,QAAQ,SAAS,OAAO,EAAE,IAAI,KAAK,KAAK,MAAM,OAAO,OAAO,OAAO,GAAO,CAAC,CAAC;AAAA,MAC5E,YAAY,KAAK,MAAM,OAAO,WAAW,OAAO,GAAO,CAAC,CAAC;AAAA,MACzD,iBAAiB,KAAK,MAAM;AAAA,MAC5B,iBAAiB,KAAK,MAAM,eAAe;AAAA,MAC3C,WAAW;AAAA;AAAA,MACX,OAAO,KAAK,KAAK;AAAA,MACjB,UAAU;AAAA,IACZ;AAEA,QAAI,KAAK,iBAAiB;AACxB,WAAK,gBAAgB,aAAa,4BAAW,kBAAkB,KAAK,UAAU,OAAO,CAAC;AACtF,WAAK,gBAAgB,IAAI;AACzB,WAAK,kBAAkB;AAAA,IACzB;AAEA,SAAK,KAAK,KAAK,qBAAqB,OAAO;AAAA,EAC7C;AAAA;AAAA,EAGA,MAAM,UAA+B;AACnC,UAAM,SAAS,CAAC;AAChB,qBAAiB,SAAS,MAAM;AAC9B,aAAO,KAAK,MAAM,KAAK;AAAA,IACzB;AACA,eAAO,0BAAY,MAAM;AAAA,EAC3B;AAAA,EAEA,OAAkD;AAChD,WAAO,KAAK,OAAO,KAAK;AAAA,EAC1B;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,CAAC,KAAK,MAAM,OAAQ,MAAK,MAAM,MAAM;AACzC,QAAI,CAAC,KAAK,OAAO,OAAQ,MAAK,OAAO,MAAM;AAC3C,QAAI,CAAC,KAAK,gBAAgB,OAAO,QAAS,MAAK,gBAAgB,MAAM;AACrE,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAmB;AACtC,WAAO;AAAA,EACT;AACF;","names":[]}
+{"version":3,"sources":["../../src/tts/tts.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { TypedEventEmitter as TypedEmitter } from '@livekit/typed-emitter';\nimport type { Span } from '@opentelemetry/api';\nimport { EventEmitter } from 'node:events';\nimport type { ReadableStream } from 'node:stream/web';\nimport { APIConnectionError, APIError } from '../_exceptions.js';\nimport { log } from '../log.js';\nimport type { TTSMetrics } from '../metrics/base.js';\nimport { DeferredReadableStream } from '../stream/deferred_stream.js';\nimport { recordException, traceTypes, tracer } from '../telemetry/index.js';\nimport { type APIConnectOptions, DEFAULT_API_CONNECT_OPTIONS, intervalForRetry } from '../types.js';\nimport { AsyncIterableQueue, delay, mergeFrames, startSoon, toError } from '../utils.js';\nimport type { TimedString } from '../voice/io.js';\n\n/**\n * SynthesizedAudio is a packet of speech synthesis as returned by the TTS.\n */\nexport interface SynthesizedAudio {\n  /** Request ID (one segment could be made up of multiple requests) */\n  requestId: string;\n  /** Segment ID, each segment is separated by a flush */\n  segmentId: string;\n  /** Synthesized audio frame */\n  frame: AudioFrame;\n  /** Current segment of the synthesized audio */\n  deltaText?: string;\n  /** Whether this is the last frame of the segment (streaming only) */\n  final: boolean;\n  /**\n   * Timed transcripts associated with this audio packet (word-level timestamps).\n   */\n  timedTranscripts?: TimedString[];\n}\n\n/**\n * Describes the capabilities of the TTS provider.\n *\n * @remarks\n * At present, only `streaming` is supplied to this interface, and the framework only supports\n * providers that do have a streaming endpoint.\n */\nexport interface TTSCapabilities {\n  streaming: boolean;\n  // Whether this TTS supports aligned transcripts (word-level timestamps).\n  alignedTranscript?: boolean;\n}\n\nexport interface TTSError {\n  type: 'tts_error';\n  timestamp: number;\n  label: string;\n  error: Error;\n  recoverable: boolean;\n}\n\nexport type TTSCallbacks = {\n  ['metrics_collected']: (metrics: TTSMetrics) => void;\n  ['error']: (error: TTSError) => void;\n};\n\n/**\n * An instance of a text-to-speech adapter.\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child TTS class, which inherits this class's methods.\n */\nexport abstract class TTS extends (EventEmitter as new () => TypedEmitter<TTSCallbacks>) {\n  #capabilities: TTSCapabilities;\n  #sampleRate: number;\n  #numChannels: number;\n  abstract label: string;\n\n  constructor(sampleRate: number, numChannels: number, capabilities: TTSCapabilities) {\n    super();\n    this.#capabilities = capabilities;\n    this.#sampleRate = sampleRate;\n    this.#numChannels = numChannels;\n  }\n\n  /** Returns this TTS's capabilities */\n  get capabilities(): TTSCapabilities {\n    return this.#capabilities;\n  }\n\n  /** Returns the sample rate of audio frames returned by this TTS */\n  get sampleRate(): number {\n    return this.#sampleRate;\n  }\n\n  /** Returns the channel count of audio frames returned by this TTS */\n  get numChannels(): number {\n    return this.#numChannels;\n  }\n\n  /**\n   * Receives text and returns synthesis in the form of a {@link ChunkedStream}\n   */\n  abstract synthesize(\n    text: string,\n    connOptions?: APIConnectOptions,\n    abortSignal?: AbortSignal,\n  ): ChunkedStream;\n\n  /**\n   * Returns a {@link SynthesizeStream} that can be used to push text and receive audio data\n   *\n   * @param options - Optional configuration including connection options\n   */\n  abstract stream(options?: { connOptions?: APIConnectOptions }): SynthesizeStream;\n\n  async close(): Promise<void> {\n    return;\n  }\n}\n\n/**\n * An instance of a text-to-speech stream, as an asynchronous iterable iterator.\n *\n * @example Looping through frames\n * ```ts\n * for await (const event of stream) {\n *   await source.captureFrame(event.frame);\n * }\n * ```\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child SynthesizeStream class, which inherits this class's methods.\n */\nexport abstract class SynthesizeStream\n  implements AsyncIterableIterator<SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM>\n{\n  protected static readonly FLUSH_SENTINEL = Symbol('FLUSH_SENTINEL');\n  static readonly END_OF_STREAM = Symbol('END_OF_STREAM');\n  protected input = new AsyncIterableQueue<string | typeof SynthesizeStream.FLUSH_SENTINEL>();\n  protected queue = new AsyncIterableQueue<\n    SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM\n  >();\n  protected output = new AsyncIterableQueue<\n    SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM\n  >();\n  protected closed = false;\n  protected connOptions: APIConnectOptions;\n  protected abortController = new AbortController();\n\n  private deferredInputStream: DeferredReadableStream<\n    string | typeof SynthesizeStream.FLUSH_SENTINEL\n  >;\n  private logger = log();\n\n  abstract label: string;\n\n  #tts: TTS;\n  #metricsPendingTexts: string[] = [];\n  #metricsText = '';\n  #monitorMetricsTask?: Promise<void>;\n  #ttsRequestSpan?: Span;\n\n  constructor(tts: TTS, connOptions: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS) {\n    this.#tts = tts;\n    this.connOptions = connOptions;\n    this.deferredInputStream = new DeferredReadableStream();\n    this.pumpInput();\n\n    this.abortController.signal.addEventListener('abort', () => {\n      this.deferredInputStream.detachSource();\n      // TODO (AJS-36) clean this up when we refactor with streams\n      if (!this.input.closed) this.input.close();\n      if (!this.output.closed) this.output.close();\n      this.closed = true;\n    });\n\n    // this is a hack to immitate asyncio.create_task so that mainTask\n    // is run **after** the constructor has finished. Otherwise we get\n    // runtime error when trying to access class variables in the\n    // `run` method.\n    startSoon(() => this.mainTask().finally(() => this.queue.close()));\n  }\n\n  private _mainTaskImpl = async (span: Span) => {\n    this.#ttsRequestSpan = span;\n    span.setAttributes({\n      [traceTypes.ATTR_TTS_STREAMING]: true,\n      [traceTypes.ATTR_TTS_LABEL]: this.#tts.label,\n    });\n\n    for (let i = 0; i < this.connOptions.maxRetry + 1; i++) {\n      try {\n        return await tracer.startActiveSpan(\n          async (attemptSpan) => {\n            attemptSpan.setAttribute(traceTypes.ATTR_RETRY_COUNT, i);\n            try {\n              return await this.run();\n            } catch (error) {\n              recordException(attemptSpan, toError(error));\n              throw error;\n            }\n          },\n          { name: 'tts_request_run' },\n        );\n      } catch (error) {\n        if (error instanceof APIError) {\n          const retryInterval = intervalForRetry(this.connOptions, i);\n\n          if (this.connOptions.maxRetry === 0 || !error.retryable) {\n            this.emitError({ error, recoverable: false });\n            throw error;\n          } else if (i === this.connOptions.maxRetry) {\n            this.emitError({ error, recoverable: false });\n            throw new APIConnectionError({\n              message: `failed to generate TTS completion after ${this.connOptions.maxRetry + 1} attempts`,\n              options: { retryable: false },\n            });\n          } else {\n            // Don't emit error event for recoverable errors during retry loop\n            // to avoid ERR_UNHANDLED_ERROR or premature session termination\n            this.logger.warn(\n              { tts: this.#tts.label, attempt: i + 1, error },\n              `failed to synthesize speech, retrying in  ${retryInterval}s`,\n            );\n          }\n\n          if (retryInterval > 0) {\n            await delay(retryInterval);\n          }\n        } else {\n          this.emitError({ error: toError(error), recoverable: false });\n          throw error;\n        }\n      }\n    }\n  };\n\n  private mainTask = async () =>\n    tracer.startActiveSpan(async (span) => this._mainTaskImpl(span), {\n      name: 'tts_request',\n      endOnExit: false,\n    });\n\n  private emitError({ error, recoverable }: { error: Error; recoverable: boolean }) {\n    this.#tts.emit('error', {\n      type: 'tts_error',\n      timestamp: Date.now(),\n      label: this.#tts.label,\n      error,\n      recoverable,\n    });\n  }\n\n  // NOTE(AJS-37): The implementation below uses an AsyncIterableQueue (`this.input`)\n  // bridged from a DeferredReadableStream (`this.deferredInputStream`) rather than\n  // consuming the stream directly.\n  //\n  // A full refactor to native Web Streams was considered but is currently deferred.\n  // The primary reason is to maintain architectural parity with the Python SDK,\n  // which is a key design goal for the project. This ensures a consistent developer\n  // experience across both platforms.\n  //\n  // For more context, see the discussion in GitHub issue # 844.\n  protected async pumpInput() {\n    const reader = this.deferredInputStream.stream.getReader();\n    try {\n      while (true) {\n        const { done, value } = await reader.read();\n        if (done || value === SynthesizeStream.FLUSH_SENTINEL) {\n          break;\n        }\n        this.pushText(value);\n      }\n      this.endInput();\n    } catch (error) {\n      this.logger.error(error, 'Error reading deferred input stream');\n    } finally {\n      reader.releaseLock();\n      // Ensure output is closed when the stream ends\n      if (!this.#monitorMetricsTask) {\n        // No text was received, close the output directly\n        this.output.close();\n      }\n    }\n  }\n\n  protected async monitorMetrics() {\n    const startTime = process.hrtime.bigint();\n    let audioDurationMs = 0;\n    let ttfb: bigint = BigInt(-1);\n    let requestId = '';\n\n    const emit = () => {\n      if (this.#metricsPendingTexts.length) {\n        const text = this.#metricsPendingTexts.shift()!;\n        const duration = process.hrtime.bigint() - startTime;\n        const roundedAudioDurationMs = Math.round(audioDurationMs);\n        const metrics: TTSMetrics = {\n          type: 'tts_metrics',\n          timestamp: Date.now(),\n          requestId,\n          ttfbMs: ttfb === BigInt(-1) ? -1 : Math.trunc(Number(ttfb / BigInt(1000000))),\n          durationMs: Math.trunc(Number(duration / BigInt(1000000))),\n          charactersCount: text.length,\n          audioDurationMs: roundedAudioDurationMs,\n          cancelled: this.abortController.signal.aborted,\n          label: this.#tts.label,\n          streamed: false,\n        };\n        if (this.#ttsRequestSpan) {\n          this.#ttsRequestSpan.setAttribute(traceTypes.ATTR_TTS_METRICS, JSON.stringify(metrics));\n        }\n        this.#tts.emit('metrics_collected', metrics);\n      }\n    };\n\n    for await (const audio of this.queue) {\n      if (this.abortController.signal.aborted) {\n        break;\n      }\n      this.output.put(audio);\n      if (audio === SynthesizeStream.END_OF_STREAM) continue;\n      requestId = audio.requestId;\n      if (ttfb === BigInt(-1)) {\n        ttfb = process.hrtime.bigint() - startTime;\n      }\n      // TODO(AJS-102): use frame.durationMs once available in rtc-node\n      audioDurationMs += (audio.frame.samplesPerChannel / audio.frame.sampleRate) * 1000;\n      if (audio.final) {\n        emit();\n      }\n    }\n\n    if (requestId) {\n      emit();\n    }\n\n    if (this.#ttsRequestSpan) {\n      this.#ttsRequestSpan.end();\n      this.#ttsRequestSpan = undefined;\n    }\n  }\n\n  protected abstract run(): Promise<void>;\n\n  updateInputStream(text: ReadableStream<string>) {\n    this.deferredInputStream.setSource(text);\n  }\n\n  /** Push a string of text to the TTS */\n  /** @deprecated Use `updateInputStream` instead */\n  pushText(text: string) {\n    if (!this.#monitorMetricsTask) {\n      this.#monitorMetricsTask = this.monitorMetrics();\n      // Close output when metrics task completes\n      this.#monitorMetricsTask.finally(() => this.output.close());\n    }\n    this.#metricsText += text;\n\n    if (this.input.closed || this.closed) {\n      // Stream was aborted/closed, silently skip\n      return;\n    }\n\n    this.input.put(text);\n  }\n\n  /** Flush the TTS, causing it to process all pending text */\n  flush() {\n    if (this.#metricsText) {\n      this.#metricsPendingTexts.push(this.#metricsText);\n      this.#metricsText = '';\n    }\n\n    if (this.input.closed || this.closed) {\n      // Stream was aborted/closed, silently skip\n      return;\n    }\n\n    this.input.put(SynthesizeStream.FLUSH_SENTINEL);\n  }\n\n  /** Mark the input as ended and forbid additional pushes */\n  endInput() {\n    this.flush();\n\n    if (this.input.closed || this.closed) {\n      // Stream was aborted/closed, silently skip\n      return;\n    }\n\n    this.input.close();\n  }\n\n  next(): Promise<IteratorResult<SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM>> {\n    return this.output.next();\n  }\n\n  get abortSignal(): AbortSignal {\n    return this.abortController.signal;\n  }\n\n  /** Close both the input and output of the TTS stream */\n  close() {\n    this.abortController.abort();\n  }\n\n  [Symbol.asyncIterator](): SynthesizeStream {\n    return this;\n  }\n}\n\n/**\n * An instance of a text-to-speech response, as an asynchronous iterable iterator.\n *\n * @example Looping through frames\n * ```ts\n * for await (const event of stream) {\n *   await source.captureFrame(event.frame);\n * }\n * ```\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child ChunkedStream class, which inherits this class's methods.\n */\nexport abstract class ChunkedStream implements AsyncIterableIterator<SynthesizedAudio> {\n  protected queue = new AsyncIterableQueue<SynthesizedAudio>();\n  protected output = new AsyncIterableQueue<SynthesizedAudio>();\n  protected closed = false;\n  abstract label: string;\n  #text: string;\n  #tts: TTS;\n  #ttsRequestSpan?: Span;\n  private _connOptions: APIConnectOptions;\n  private logger = log();\n\n  protected abortController = new AbortController();\n\n  constructor(\n    text: string,\n    tts: TTS,\n    connOptions: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,\n    abortSignal?: AbortSignal,\n  ) {\n    this.#text = text;\n    this.#tts = tts;\n    this._connOptions = connOptions;\n\n    if (abortSignal) {\n      abortSignal.addEventListener('abort', () => this.abortController.abort(), { once: true });\n    }\n\n    this.monitorMetrics();\n\n    // this is a hack to immitate asyncio.create_task so that mainTask\n    // is run **after** the constructor has finished. Otherwise we get\n    // runtime error when trying to access class variables in the\n    // `run` method.\n    Promise.resolve().then(() => this.mainTask().finally(() => this.queue.close()));\n  }\n\n  private _mainTaskImpl = async (span: Span) => {\n    this.#ttsRequestSpan = span;\n    span.setAttributes({\n      [traceTypes.ATTR_TTS_STREAMING]: false,\n      [traceTypes.ATTR_TTS_LABEL]: this.#tts.label,\n    });\n\n    for (let i = 0; i < this._connOptions.maxRetry + 1; i++) {\n      try {\n        return await tracer.startActiveSpan(\n          async (attemptSpan) => {\n            attemptSpan.setAttribute(traceTypes.ATTR_RETRY_COUNT, i);\n            try {\n              return await this.run();\n            } catch (error) {\n              recordException(attemptSpan, toError(error));\n              throw error;\n            }\n          },\n          { name: 'tts_request_run' },\n        );\n      } catch (error) {\n        if (error instanceof APIError) {\n          const retryInterval = intervalForRetry(this._connOptions, i);\n\n          if (this._connOptions.maxRetry === 0 || !error.retryable) {\n            this.emitError({ error, recoverable: false });\n            throw error;\n          } else if (i === this._connOptions.maxRetry) {\n            this.emitError({ error, recoverable: false });\n            throw new APIConnectionError({\n              message: `failed to generate TTS completion after ${this._connOptions.maxRetry + 1} attempts`,\n              options: { retryable: false },\n            });\n          } else {\n            // Don't emit error event for recoverable errors during retry loop\n            // to avoid ERR_UNHANDLED_ERROR or premature session termination\n            this.logger.warn(\n              { tts: this.#tts.label, attempt: i + 1, error },\n              `failed to generate TTS completion, retrying in ${retryInterval}s`,\n            );\n          }\n\n          if (retryInterval > 0) {\n            await delay(retryInterval);\n          }\n        } else {\n          this.emitError({ error: toError(error), recoverable: false });\n          throw error;\n        }\n      }\n    }\n  };\n\n  private async mainTask() {\n    return tracer.startActiveSpan(async (span) => this._mainTaskImpl(span), {\n      name: 'tts_request',\n      endOnExit: false,\n    });\n  }\n\n  private emitError({ error, recoverable }: { error: Error; recoverable: boolean }) {\n    this.#tts.emit('error', {\n      type: 'tts_error',\n      timestamp: Date.now(),\n      label: this.#tts.label,\n      error,\n      recoverable,\n    });\n  }\n\n  protected abstract run(): Promise<void>;\n\n  get inputText(): string {\n    return this.#text;\n  }\n\n  get abortSignal(): AbortSignal {\n    return this.abortController.signal;\n  }\n\n  protected async monitorMetrics() {\n    const startTime = process.hrtime.bigint();\n    let audioDurationMs = 0;\n    let ttfb: bigint = BigInt(-1);\n    let requestId = '';\n\n    for await (const audio of this.queue) {\n      this.output.put(audio);\n      requestId = audio.requestId;\n      if (ttfb === BigInt(-1)) {\n        ttfb = process.hrtime.bigint() - startTime;\n      }\n      audioDurationMs += (audio.frame.samplesPerChannel / audio.frame.sampleRate) * 1000;\n    }\n    this.output.close();\n\n    const duration = process.hrtime.bigint() - startTime;\n    const metrics: TTSMetrics = {\n      type: 'tts_metrics',\n      timestamp: Date.now(),\n      requestId,\n      ttfbMs: ttfb === BigInt(-1) ? -1 : Math.trunc(Number(ttfb / BigInt(1000000))),\n      durationMs: Math.trunc(Number(duration / BigInt(1000000))),\n      charactersCount: this.#text.length,\n      audioDurationMs: Math.round(audioDurationMs),\n      cancelled: false, // TODO(AJS-186): support ChunkedStream with 1.0 - add this.abortController.signal.aborted here\n      label: this.#tts.label,\n      streamed: false,\n    };\n\n    if (this.#ttsRequestSpan) {\n      this.#ttsRequestSpan.setAttribute(traceTypes.ATTR_TTS_METRICS, JSON.stringify(metrics));\n      this.#ttsRequestSpan.end();\n      this.#ttsRequestSpan = undefined;\n    }\n\n    this.#tts.emit('metrics_collected', metrics);\n  }\n\n  /** Collect every frame into one in a single call */\n  async collect(): Promise<AudioFrame> {\n    const frames = [];\n    for await (const event of this) {\n      frames.push(event.frame);\n    }\n    return mergeFrames(frames);\n  }\n\n  next(): Promise<IteratorResult<SynthesizedAudio>> {\n    return this.output.next();\n  }\n\n  /** Close both the input and output of the TTS stream */\n  close() {\n    if (!this.queue.closed) this.queue.close();\n    if (!this.output.closed) this.output.close();\n    if (!this.abortController.signal.aborted) this.abortController.abort();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): ChunkedStream {\n    return this;\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAMA,yBAA6B;AAE7B,wBAA6C;AAC7C,iBAAoB;AAEpB,6BAAuC;AACvC,uBAAoD;AACpD,mBAAsF;AACtF,mBAA2E;AAwDpE,MAAe,YAAa,gCAAsD;AAAA,EACvF;AAAA,EACA;AAAA,EACA;AAAA,EAGA,YAAY,YAAoB,aAAqB,cAA+B;AAClF,UAAM;AACN,SAAK,gBAAgB;AACrB,SAAK,cAAc;AACnB,SAAK,eAAe;AAAA,EACtB;AAAA;AAAA,EAGA,IAAI,eAAgC;AAClC,WAAO,KAAK;AAAA,EACd;AAAA;AAAA,EAGA,IAAI,aAAqB;AACvB,WAAO,KAAK;AAAA,EACd;AAAA;AAAA,EAGA,IAAI,cAAsB;AACxB,WAAO,KAAK;AAAA,EACd;AAAA,EAkBA,MAAM,QAAuB;AAC3B;AAAA,EACF;AACF;AAgBO,MAAe,iBAEtB;AAAA,EACE,OAA0B,iBAAiB,OAAO,gBAAgB;AAAA,EAClE,OAAgB,gBAAgB,OAAO,eAAe;AAAA,EAC5C,QAAQ,IAAI,gCAAoE;AAAA,EAChF,QAAQ,IAAI,gCAEpB;AAAA,EACQ,SAAS,IAAI,gCAErB;AAAA,EACQ,SAAS;AAAA,EACT;AAAA,EACA,kBAAkB,IAAI,gBAAgB;AAAA,EAExC;AAAA,EAGA,aAAS,gBAAI;AAAA,EAIrB;AAAA,EACA,uBAAiC,CAAC;AAAA,EAClC,eAAe;AAAA,EACf;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,cAAiC,0CAA6B;AAClF,SAAK,OAAO;AACZ,SAAK,cAAc;AACnB,SAAK,sBAAsB,IAAI,8CAAuB;AACtD,SAAK,UAAU;AAEf,SAAK,gBAAgB,OAAO,iBAAiB,SAAS,MAAM;AAC1D,WAAK,oBAAoB,aAAa;AAEtC,UAAI,CAAC,KAAK,MAAM,OAAQ,MAAK,MAAM,MAAM;AACzC,UAAI,CAAC,KAAK,OAAO,OAAQ,MAAK,OAAO,MAAM;AAC3C,WAAK,SAAS;AAAA,IAChB,CAAC;AAMD,gCAAU,MAAM,KAAK,SAAS,EAAE,QAAQ,MAAM,KAAK,MAAM,MAAM,CAAC,CAAC;AAAA,EACnE;AAAA,EAEQ,gBAAgB,OAAO,SAAe;AAC5C,SAAK,kBAAkB;AACvB,SAAK,cAAc;AAAA,MACjB,CAAC,4BAAW,kBAAkB,GAAG;AAAA,MACjC,CAAC,4BAAW,cAAc,GAAG,KAAK,KAAK;AAAA,IACzC,CAAC;AAED,aAAS,IAAI,GAAG,IAAI,KAAK,YAAY,WAAW,GAAG,KAAK;AACtD,UAAI;AACF,eAAO,MAAM,wBAAO;AAAA,UAClB,OAAO,gBAAgB;AACrB,wBAAY,aAAa,4BAAW,kBAAkB,CAAC;AACvD,gBAAI;AACF,qBAAO,MAAM,KAAK,IAAI;AAAA,YACxB,SAAS,OAAO;AACd,oDAAgB,iBAAa,sBAAQ,KAAK,CAAC;AAC3C,oBAAM;AAAA,YACR;AAAA,UACF;AAAA,UACA,EAAE,MAAM,kBAAkB;AAAA,QAC5B;AAAA,MACF,SAAS,OAAO;AACd,YAAI,iBAAiB,4BAAU;AAC7B,gBAAM,oBAAgB,+BAAiB,KAAK,aAAa,CAAC;AAE1D,cAAI,KAAK,YAAY,aAAa,KAAK,CAAC,MAAM,WAAW;AACvD,iBAAK,UAAU,EAAE,OAAO,aAAa,MAAM,CAAC;AAC5C,kBAAM;AAAA,UACR,WAAW,MAAM,KAAK,YAAY,UAAU;AAC1C,iBAAK,UAAU,EAAE,OAAO,aAAa,MAAM,CAAC;AAC5C,kBAAM,IAAI,qCAAmB;AAAA,cAC3B,SAAS,2CAA2C,KAAK,YAAY,WAAW,CAAC;AAAA,cACjF,SAAS,EAAE,WAAW,MAAM;AAAA,YAC9B,CAAC;AAAA,UACH,OAAO;AAGL,iBAAK,OAAO;AAAA,cACV,EAAE,KAAK,KAAK,KAAK,OAAO,SAAS,IAAI,GAAG,MAAM;AAAA,cAC9C,6CAA6C,aAAa;AAAA,YAC5D;AAAA,UACF;AAEA,cAAI,gBAAgB,GAAG;AACrB,sBAAM,oBAAM,aAAa;AAAA,UAC3B;AAAA,QACF,OAAO;AACL,eAAK,UAAU,EAAE,WAAO,sBAAQ,KAAK,GAAG,aAAa,MAAM,CAAC;AAC5D,gBAAM;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA,EAEQ,WAAW,YACjB,wBAAO,gBAAgB,OAAO,SAAS,KAAK,cAAc,IAAI,GAAG;AAAA,IAC/D,MAAM;AAAA,IACN,WAAW;AAAA,EACb,CAAC;AAAA,EAEK,UAAU,EAAE,OAAO,YAAY,GAA2C;AAChF,SAAK,KAAK,KAAK,SAAS;AAAA,MACtB,MAAM;AAAA,MACN,WAAW,KAAK,IAAI;AAAA,MACpB,OAAO,KAAK,KAAK;AAAA,MACjB;AAAA,MACA;AAAA,IACF,CAAC;AAAA,EACH;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAYA,MAAgB,YAAY;AAC1B,UAAM,SAAS,KAAK,oBAAoB,OAAO,UAAU;AACzD,QAAI;AACF,aAAO,MAAM;AACX,cAAM,EAAE,MAAM,MAAM,IAAI,MAAM,OAAO,KAAK;AAC1C,YAAI,QAAQ,UAAU,iBAAiB,gBAAgB;AACrD;AAAA,QACF;AACA,aAAK,SAAS,KAAK;AAAA,MACrB;AACA,WAAK,SAAS;AAAA,IAChB,SAAS,OAAO;AACd,WAAK,OAAO,MAAM,OAAO,qCAAqC;AAAA,IAChE,UAAE;AACA,aAAO,YAAY;AAEnB,UAAI,CAAC,KAAK,qBAAqB;AAE7B,aAAK,OAAO,MAAM;AAAA,MACpB;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAgB,iBAAiB;AAC/B,UAAM,YAAY,QAAQ,OAAO,OAAO;AACxC,QAAI,kBAAkB;AACtB,QAAI,OAAe,OAAO,EAAE;AAC5B,QAAI,YAAY;AAEhB,UAAM,OAAO,MAAM;AACjB,UAAI,KAAK,qBAAqB,QAAQ;AACpC,cAAM,OAAO,KAAK,qBAAqB,MAAM;AAC7C,cAAM,WAAW,QAAQ,OAAO,OAAO,IAAI;AAC3C,cAAM,yBAAyB,KAAK,MAAM,eAAe;AACzD,cAAM,UAAsB;AAAA,UAC1B,MAAM;AAAA,UACN,WAAW,KAAK,IAAI;AAAA,UACpB;AAAA,UACA,QAAQ,SAAS,OAAO,EAAE,IAAI,KAAK,KAAK,MAAM,OAAO,OAAO,OAAO,GAAO,CAAC,CAAC;AAAA,UAC5E,YAAY,KAAK,MAAM,OAAO,WAAW,OAAO,GAAO,CAAC,CAAC;AAAA,UACzD,iBAAiB,KAAK;AAAA,UACtB,iBAAiB;AAAA,UACjB,WAAW,KAAK,gBAAgB,OAAO;AAAA,UACvC,OAAO,KAAK,KAAK;AAAA,UACjB,UAAU;AAAA,QACZ;AACA,YAAI,KAAK,iBAAiB;AACxB,eAAK,gBAAgB,aAAa,4BAAW,kBAAkB,KAAK,UAAU,OAAO,CAAC;AAAA,QACxF;AACA,aAAK,KAAK,KAAK,qBAAqB,OAAO;AAAA,MAC7C;AAAA,IACF;AAEA,qBAAiB,SAAS,KAAK,OAAO;AACpC,UAAI,KAAK,gBAAgB,OAAO,SAAS;AACvC;AAAA,MACF;AACA,WAAK,OAAO,IAAI,KAAK;AACrB,UAAI,UAAU,iBAAiB,cAAe;AAC9C,kBAAY,MAAM;AAClB,UAAI,SAAS,OAAO,EAAE,GAAG;AACvB,eAAO,QAAQ,OAAO,OAAO,IAAI;AAAA,MACnC;AAEA,yBAAoB,MAAM,MAAM,oBAAoB,MAAM,MAAM,aAAc;AAC9E,UAAI,MAAM,OAAO;AACf,aAAK;AAAA,MACP;AAAA,IACF;AAEA,QAAI,WAAW;AACb,WAAK;AAAA,IACP;AAEA,QAAI,KAAK,iBAAiB;AACxB,WAAK,gBAAgB,IAAI;AACzB,WAAK,kBAAkB;AAAA,IACzB;AAAA,EACF;AAAA,EAIA,kBAAkB,MAA8B;AAC9C,SAAK,oBAAoB,UAAU,IAAI;AAAA,EACzC;AAAA;AAAA;AAAA,EAIA,SAAS,MAAc;AACrB,QAAI,CAAC,KAAK,qBAAqB;AAC7B,WAAK,sBAAsB,KAAK,eAAe;AAE/C,WAAK,oBAAoB,QAAQ,MAAM,KAAK,OAAO,MAAM,CAAC;AAAA,IAC5D;AACA,SAAK,gBAAgB;AAErB,QAAI,KAAK,MAAM,UAAU,KAAK,QAAQ;AAEpC;AAAA,IACF;AAEA,SAAK,MAAM,IAAI,IAAI;AAAA,EACrB;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,KAAK,cAAc;AACrB,WAAK,qBAAqB,KAAK,KAAK,YAAY;AAChD,WAAK,eAAe;AAAA,IACtB;AAEA,QAAI,KAAK,MAAM,UAAU,KAAK,QAAQ;AAEpC;AAAA,IACF;AAEA,SAAK,MAAM,IAAI,iBAAiB,cAAc;AAAA,EAChD;AAAA;AAAA,EAGA,WAAW;AACT,SAAK,MAAM;AAEX,QAAI,KAAK,MAAM,UAAU,KAAK,QAAQ;AAEpC;AAAA,IACF;AAEA,SAAK,MAAM,MAAM;AAAA,EACnB;AAAA,EAEA,OAA0F;AACxF,WAAO,KAAK,OAAO,KAAK;AAAA,EAC1B;AAAA,EAEA,IAAI,cAA2B;AAC7B,WAAO,KAAK,gBAAgB;AAAA,EAC9B;AAAA;AAAA,EAGA,QAAQ;AACN,SAAK,gBAAgB,MAAM;AAAA,EAC7B;AAAA,EAEA,CAAC,OAAO,aAAa,IAAsB;AACzC,WAAO;AAAA,EACT;AACF;AAgBO,MAAe,cAAiE;AAAA,EAC3E,QAAQ,IAAI,gCAAqC;AAAA,EACjD,SAAS,IAAI,gCAAqC;AAAA,EAClD,SAAS;AAAA,EAEnB;AAAA,EACA;AAAA,EACA;AAAA,EACQ;AAAA,EACA,aAAS,gBAAI;AAAA,EAEX,kBAAkB,IAAI,gBAAgB;AAAA,EAEhD,YACE,MACA,KACA,cAAiC,0CACjC,aACA;AACA,SAAK,QAAQ;AACb,SAAK,OAAO;AACZ,SAAK,eAAe;AAEpB,QAAI,aAAa;AACf,kBAAY,iBAAiB,SAAS,MAAM,KAAK,gBAAgB,MAAM,GAAG,EAAE,MAAM,KAAK,CAAC;AAAA,IAC1F;AAEA,SAAK,eAAe;AAMpB,YAAQ,QAAQ,EAAE,KAAK,MAAM,KAAK,SAAS,EAAE,QAAQ,MAAM,KAAK,MAAM,MAAM,CAAC,CAAC;AAAA,EAChF;AAAA,EAEQ,gBAAgB,OAAO,SAAe;AAC5C,SAAK,kBAAkB;AACvB,SAAK,cAAc;AAAA,MACjB,CAAC,4BAAW,kBAAkB,GAAG;AAAA,MACjC,CAAC,4BAAW,cAAc,GAAG,KAAK,KAAK;AAAA,IACzC,CAAC;AAED,aAAS,IAAI,GAAG,IAAI,KAAK,aAAa,WAAW,GAAG,KAAK;AACvD,UAAI;AACF,eAAO,MAAM,wBAAO;AAAA,UAClB,OAAO,gBAAgB;AACrB,wBAAY,aAAa,4BAAW,kBAAkB,CAAC;AACvD,gBAAI;AACF,qBAAO,MAAM,KAAK,IAAI;AAAA,YACxB,SAAS,OAAO;AACd,oDAAgB,iBAAa,sBAAQ,KAAK,CAAC;AAC3C,oBAAM;AAAA,YACR;AAAA,UACF;AAAA,UACA,EAAE,MAAM,kBAAkB;AAAA,QAC5B;AAAA,MACF,SAAS,OAAO;AACd,YAAI,iBAAiB,4BAAU;AAC7B,gBAAM,oBAAgB,+BAAiB,KAAK,cAAc,CAAC;AAE3D,cAAI,KAAK,aAAa,aAAa,KAAK,CAAC,MAAM,WAAW;AACxD,iBAAK,UAAU,EAAE,OAAO,aAAa,MAAM,CAAC;AAC5C,kBAAM;AAAA,UACR,WAAW,MAAM,KAAK,aAAa,UAAU;AAC3C,iBAAK,UAAU,EAAE,OAAO,aAAa,MAAM,CAAC;AAC5C,kBAAM,IAAI,qCAAmB;AAAA,cAC3B,SAAS,2CAA2C,KAAK,aAAa,WAAW,CAAC;AAAA,cAClF,SAAS,EAAE,WAAW,MAAM;AAAA,YAC9B,CAAC;AAAA,UACH,OAAO;AAGL,iBAAK,OAAO;AAAA,cACV,EAAE,KAAK,KAAK,KAAK,OAAO,SAAS,IAAI,GAAG,MAAM;AAAA,cAC9C,kDAAkD,aAAa;AAAA,YACjE;AAAA,UACF;AAEA,cAAI,gBAAgB,GAAG;AACrB,sBAAM,oBAAM,aAAa;AAAA,UAC3B;AAAA,QACF,OAAO;AACL,eAAK,UAAU,EAAE,WAAO,sBAAQ,KAAK,GAAG,aAAa,MAAM,CAAC;AAC5D,gBAAM;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAc,WAAW;AACvB,WAAO,wBAAO,gBAAgB,OAAO,SAAS,KAAK,cAAc,IAAI,GAAG;AAAA,MACtE,MAAM;AAAA,MACN,WAAW;AAAA,IACb,CAAC;AAAA,EACH;AAAA,EAEQ,UAAU,EAAE,OAAO,YAAY,GAA2C;AAChF,SAAK,KAAK,KAAK,SAAS;AAAA,MACtB,MAAM;AAAA,MACN,WAAW,KAAK,IAAI;AAAA,MACpB,OAAO,KAAK,KAAK;AAAA,MACjB;AAAA,MACA;AAAA,IACF,CAAC;AAAA,EACH;AAAA,EAIA,IAAI,YAAoB;AACtB,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,IAAI,cAA2B;AAC7B,WAAO,KAAK,gBAAgB;AAAA,EAC9B;AAAA,EAEA,MAAgB,iBAAiB;AAC/B,UAAM,YAAY,QAAQ,OAAO,OAAO;AACxC,QAAI,kBAAkB;AACtB,QAAI,OAAe,OAAO,EAAE;AAC5B,QAAI,YAAY;AAEhB,qBAAiB,SAAS,KAAK,OAAO;AACpC,WAAK,OAAO,IAAI,KAAK;AACrB,kBAAY,MAAM;AAClB,UAAI,SAAS,OAAO,EAAE,GAAG;AACvB,eAAO,QAAQ,OAAO,OAAO,IAAI;AAAA,MACnC;AACA,yBAAoB,MAAM,MAAM,oBAAoB,MAAM,MAAM,aAAc;AAAA,IAChF;AACA,SAAK,OAAO,MAAM;AAElB,UAAM,WAAW,QAAQ,OAAO,OAAO,IAAI;AAC3C,UAAM,UAAsB;AAAA,MAC1B,MAAM;AAAA,MACN,WAAW,KAAK,IAAI;AAAA,MACpB;AAAA,MACA,QAAQ,SAAS,OAAO,EAAE,IAAI,KAAK,KAAK,MAAM,OAAO,OAAO,OAAO,GAAO,CAAC,CAAC;AAAA,MAC5E,YAAY,KAAK,MAAM,OAAO,WAAW,OAAO,GAAO,CAAC,CAAC;AAAA,MACzD,iBAAiB,KAAK,MAAM;AAAA,MAC5B,iBAAiB,KAAK,MAAM,eAAe;AAAA,MAC3C,WAAW;AAAA;AAAA,MACX,OAAO,KAAK,KAAK;AAAA,MACjB,UAAU;AAAA,IACZ;AAEA,QAAI,KAAK,iBAAiB;AACxB,WAAK,gBAAgB,aAAa,4BAAW,kBAAkB,KAAK,UAAU,OAAO,CAAC;AACtF,WAAK,gBAAgB,IAAI;AACzB,WAAK,kBAAkB;AAAA,IACzB;AAEA,SAAK,KAAK,KAAK,qBAAqB,OAAO;AAAA,EAC7C;AAAA;AAAA,EAGA,MAAM,UAA+B;AACnC,UAAM,SAAS,CAAC;AAChB,qBAAiB,SAAS,MAAM;AAC9B,aAAO,KAAK,MAAM,KAAK;AAAA,IACzB;AACA,eAAO,0BAAY,MAAM;AAAA,EAC3B;AAAA,EAEA,OAAkD;AAChD,WAAO,KAAK,OAAO,KAAK;AAAA,EAC1B;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,CAAC,KAAK,MAAM,OAAQ,MAAK,MAAM,MAAM;AACzC,QAAI,CAAC,KAAK,OAAO,OAAQ,MAAK,OAAO,MAAM;AAC3C,QAAI,CAAC,KAAK,gBAAgB,OAAO,QAAS,MAAK,gBAAgB,MAAM;AACrE,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAmB;AACtC,WAAO;AAAA,EACT;AACF;","names":[]}