@livekit/agents 0.6.4 → 0.7.0

package/dist/tokenize/basic/basic.d.ts

@@ -1,4 +1,5 @@
 import * as tokenizer from '../tokenizer.js';
+import { splitWords } from './word.js';
 export declare class SentenceTokenizer extends tokenizer.SentenceTokenizer {
     #private;
     constructor(language?: string, minSentenceLength?: number, streamContextLength?: number);
@@ -12,5 +13,6 @@ export declare class WordTokenizer extends tokenizer.WordTokenizer {
     stream(language?: string): tokenizer.WordStream;
 }
 export declare const hyphenateWord: (word: string) => string[];
+export { splitWords };
 export declare const tokenizeParagraphs: (text: string) => string[];
 //# sourceMappingURL=basic.d.ts.map

package/dist/tokenize/basic/basic.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"basic.d.ts","sourceRoot":"","sources":["../../../src/tokenize/basic/basic.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,SAAS,MAAM,iBAAiB,CAAC;AAY7C,qBAAa,iBAAkB,SAAQ,SAAS,CAAC,iBAAiB;;gBAGpD,QAAQ,SAAU,EAAE,iBAAiB,SAAK,EAAE,mBAAmB,SAAK;IAUhF,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,MAAM,EAAE;IAKnD,MAAM,CAAC,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC,cAAc;CAOpD;AAED,qBAAa,aAAc,SAAQ,SAAS,CAAC,aAAa;;gBAG5C,iBAAiB,UAAO;IAMpC,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,MAAM,EAAE;IAKnD,MAAM,CAAC,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC,UAAU;CAOhD;AAED,eAAO,MAAM,aAAa,SAAU,MAAM,KAAG,MAAM,EAElD,CAAC;AAEF,eAAO,MAAM,kBAAkB,SAAU,MAAM,KAAG,MAAM,EAEvD,CAAC"}
+{"version":3,"file":"basic.d.ts","sourceRoot":"","sources":["../../../src/tokenize/basic/basic.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,SAAS,MAAM,iBAAiB,CAAC;AAI7C,OAAO,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AAQvC,qBAAa,iBAAkB,SAAQ,SAAS,CAAC,iBAAiB;;gBAGpD,QAAQ,SAAU,EAAE,iBAAiB,SAAK,EAAE,mBAAmB,SAAK;IAUhF,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,MAAM,EAAE;IAKnD,MAAM,CAAC,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC,cAAc;CAOpD;AAED,qBAAa,aAAc,SAAQ,SAAS,CAAC,aAAa;;gBAG5C,iBAAiB,UAAO;IAMpC,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,MAAM,EAAE;IAKnD,MAAM,CAAC,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC,UAAU;CAOhD;AAED,eAAO,MAAM,aAAa,SAAU,MAAM,KAAG,MAAM,EAElD,CAAC;AAEF,OAAO,EAAE,UAAU,EAAE,CAAC;AAEtB,eAAO,MAAM,kBAAkB,SAAU,MAAM,KAAG,MAAM,EAEvD,CAAC"}

package/dist/tokenize/basic/basic.js

@@ -56,6 +56,7 @@ export {
   SentenceTokenizer,
   WordTokenizer,
   hyphenateWord,
+  splitWords,
   tokenizeParagraphs
 };
 //# sourceMappingURL=basic.js.map

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

@@ -1 +1 @@
-{"version":3,"sources":["../../../src/tokenize/basic/basic.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { BufferedSentenceStream, BufferedWordStream } from '../token_stream.js';\nimport * as tokenizer from '../tokenizer.js';\nimport { hyphenator } from './hyphenator.js';\nimport { splitParagraphs } from './paragraph.js';\nimport { splitSentences } from './sentence.js';\nimport { splitWords } from './word.js';\n\ninterface TokenizerOptions {\n  language: string;\n  minSentenceLength: number;\n  streamContextLength: number;\n}\n\nexport class SentenceTokenizer extends tokenizer.SentenceTokenizer {\n  #config: TokenizerOptions;\n\n  constructor(language = 'en-US', minSentenceLength = 20, streamContextLength = 10) {\n    super();\n    this.#config = {\n      language,\n      minSentenceLength,\n      streamContextLength,\n    };\n  }\n\n  // eslint-disable-next-line @typescript-eslint/no-unused-vars\n  tokenize(text: string, language?: string): string[] {\n    return splitSentences(text, this.#config.minSentenceLength).map((tok) => tok[0]);\n  }\n\n  // eslint-disable-next-line @typescript-eslint/no-unused-vars\n  stream(language?: string): tokenizer.SentenceStream {\n    return new BufferedSentenceStream(\n      (text: string) => splitSentences(text, this.#config.minSentenceLength),\n      this.#config.minSentenceLength,\n      this.#config.streamContextLength,\n    );\n  }\n}\n\nexport class WordTokenizer extends tokenizer.WordTokenizer {\n  #ignorePunctuation: boolean;\n\n  constructor(ignorePunctuation = true) {\n    super();\n    this.#ignorePunctuation = ignorePunctuation;\n  }\n\n  // eslint-disable-next-line @typescript-eslint/no-unused-vars\n  tokenize(text: string, language?: string): string[] {\n    return splitWords(text, this.#ignorePunctuation).map((tok) => tok[0]);\n  }\n\n  // eslint-disable-next-line @typescript-eslint/no-unused-vars\n  stream(language?: string): tokenizer.WordStream {\n    return new BufferedWordStream(\n      (text: string) => splitWords(text, this.#ignorePunctuation),\n      1,\n      1,\n    );\n  }\n}\n\nexport const hyphenateWord = (word: string): string[] => {\n  return hyphenator.hyphenateWord(word);\n};\n\nexport const tokenizeParagraphs = (text: string): string[] => {\n  return splitParagraphs(text).map((tok) => tok[0]);\n};\n"],"mappings":"AAGA,SAAS,wBAAwB,0BAA0B;AAC3D,YAAY,eAAe;AAC3B,SAAS,kBAAkB;AAC3B,SAAS,uBAAuB;AAChC,SAAS,sBAAsB;AAC/B,SAAS,kBAAkB;AAQpB,MAAM,0BAA0B,UAAU,kBAAkB;AAAA,EACjE;AAAA,EAEA,YAAY,WAAW,SAAS,oBAAoB,IAAI,sBAAsB,IAAI;AAChF,UAAM;AACN,SAAK,UAAU;AAAA,MACb;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA,EAGA,SAAS,MAAc,UAA6B;AAClD,WAAO,eAAe,MAAM,KAAK,QAAQ,iBAAiB,EAAE,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;AAAA,EACjF;AAAA;AAAA,EAGA,OAAO,UAA6C;AAClD,WAAO,IAAI;AAAA,MACT,CAAC,SAAiB,eAAe,MAAM,KAAK,QAAQ,iBAAiB;AAAA,MACrE,KAAK,QAAQ;AAAA,MACb,KAAK,QAAQ;AAAA,IACf;AAAA,EACF;AACF;AAEO,MAAM,sBAAsB,UAAU,cAAc;AAAA,EACzD;AAAA,EAEA,YAAY,oBAAoB,MAAM;AACpC,UAAM;AACN,SAAK,qBAAqB;AAAA,EAC5B;AAAA;AAAA,EAGA,SAAS,MAAc,UAA6B;AAClD,WAAO,WAAW,MAAM,KAAK,kBAAkB,EAAE,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;AAAA,EACtE;AAAA;AAAA,EAGA,OAAO,UAAyC;AAC9C,WAAO,IAAI;AAAA,MACT,CAAC,SAAiB,WAAW,MAAM,KAAK,kBAAkB;AAAA,MAC1D;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACF;AAEO,MAAM,gBAAgB,CAAC,SAA2B;AACvD,SAAO,WAAW,cAAc,IAAI;AACtC;AAEO,MAAM,qBAAqB,CAAC,SAA2B;AAC5D,SAAO,gBAAgB,IAAI,EAAE,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;AAClD;","names":[]}
+{"version":3,"sources":["../../../src/tokenize/basic/basic.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { BufferedSentenceStream, BufferedWordStream } from '../token_stream.js';\nimport * as tokenizer from '../tokenizer.js';\nimport { hyphenator } from './hyphenator.js';\nimport { splitParagraphs } from './paragraph.js';\nimport { splitSentences } from './sentence.js';\nimport { splitWords } from './word.js';\n\ninterface TokenizerOptions {\n  language: string;\n  minSentenceLength: number;\n  streamContextLength: number;\n}\n\nexport class SentenceTokenizer extends tokenizer.SentenceTokenizer {\n  #config: TokenizerOptions;\n\n  constructor(language = 'en-US', minSentenceLength = 20, streamContextLength = 10) {\n    super();\n    this.#config = {\n      language,\n      minSentenceLength,\n      streamContextLength,\n    };\n  }\n\n  // eslint-disable-next-line @typescript-eslint/no-unused-vars\n  tokenize(text: string, language?: string): string[] {\n    return splitSentences(text, this.#config.minSentenceLength).map((tok) => tok[0]);\n  }\n\n  // eslint-disable-next-line @typescript-eslint/no-unused-vars\n  stream(language?: string): tokenizer.SentenceStream {\n    return new BufferedSentenceStream(\n      (text: string) => splitSentences(text, this.#config.minSentenceLength),\n      this.#config.minSentenceLength,\n      this.#config.streamContextLength,\n    );\n  }\n}\n\nexport class WordTokenizer extends tokenizer.WordTokenizer {\n  #ignorePunctuation: boolean;\n\n  constructor(ignorePunctuation = true) {\n    super();\n    this.#ignorePunctuation = ignorePunctuation;\n  }\n\n  // eslint-disable-next-line @typescript-eslint/no-unused-vars\n  tokenize(text: string, language?: string): string[] {\n    return splitWords(text, this.#ignorePunctuation).map((tok) => tok[0]);\n  }\n\n  // eslint-disable-next-line @typescript-eslint/no-unused-vars\n  stream(language?: string): tokenizer.WordStream {\n    return new BufferedWordStream(\n      (text: string) => splitWords(text, this.#ignorePunctuation),\n      1,\n      1,\n    );\n  }\n}\n\nexport const hyphenateWord = (word: string): string[] => {\n  return hyphenator.hyphenateWord(word);\n};\n\nexport { splitWords };\n\nexport const tokenizeParagraphs = (text: string): string[] => {\n  return splitParagraphs(text).map((tok) => tok[0]);\n};\n"],"mappings":"AAGA,SAAS,wBAAwB,0BAA0B;AAC3D,YAAY,eAAe;AAC3B,SAAS,kBAAkB;AAC3B,SAAS,uBAAuB;AAChC,SAAS,sBAAsB;AAC/B,SAAS,kBAAkB;AAQpB,MAAM,0BAA0B,UAAU,kBAAkB;AAAA,EACjE;AAAA,EAEA,YAAY,WAAW,SAAS,oBAAoB,IAAI,sBAAsB,IAAI;AAChF,UAAM;AACN,SAAK,UAAU;AAAA,MACb;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA,EAGA,SAAS,MAAc,UAA6B;AAClD,WAAO,eAAe,MAAM,KAAK,QAAQ,iBAAiB,EAAE,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;AAAA,EACjF;AAAA;AAAA,EAGA,OAAO,UAA6C;AAClD,WAAO,IAAI;AAAA,MACT,CAAC,SAAiB,eAAe,MAAM,KAAK,QAAQ,iBAAiB;AAAA,MACrE,KAAK,QAAQ;AAAA,MACb,KAAK,QAAQ;AAAA,IACf;AAAA,EACF;AACF;AAEO,MAAM,sBAAsB,UAAU,cAAc;AAAA,EACzD;AAAA,EAEA,YAAY,oBAAoB,MAAM;AACpC,UAAM;AACN,SAAK,qBAAqB;AAAA,EAC5B;AAAA;AAAA,EAGA,SAAS,MAAc,UAA6B;AAClD,WAAO,WAAW,MAAM,KAAK,kBAAkB,EAAE,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;AAAA,EACtE;AAAA;AAAA,EAGA,OAAO,UAAyC;AAC9C,WAAO,IAAI;AAAA,MACT,CAAC,SAAiB,WAAW,MAAM,KAAK,kBAAkB;AAAA,MAC1D;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACF;AAEO,MAAM,gBAAgB,CAAC,SAA2B;AACvD,SAAO,WAAW,cAAc,IAAI;AACtC;AAIO,MAAM,qBAAqB,CAAC,SAA2B;AAC5D,SAAO,gBAAgB,IAAI,EAAE,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;AAClD;","names":[]}

package/dist/tokenize/basic/index.cjs

@@ -21,6 +21,7 @@ __export(basic_exports, {
   SentenceTokenizer: () => import_basic.SentenceTokenizer,
   WordTokenizer: () => import_basic.WordTokenizer,
   hyphenateWord: () => import_basic.hyphenateWord,
+  splitWords: () => import_basic.splitWords,
   tokenizeParagraphs: () => import_basic.tokenizeParagraphs
 });
 module.exports = __toCommonJS(basic_exports);
@@ -30,6 +31,7 @@ var import_basic = require("./basic.cjs");
   SentenceTokenizer,
   WordTokenizer,
   hyphenateWord,
+  splitWords,
   tokenizeParagraphs
 });
 //# sourceMappingURL=index.cjs.map

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

@@ -1 +1 @@
-{"version":3,"sources":["../../../src/tokenize/basic/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport { SentenceTokenizer, WordTokenizer, tokenizeParagraphs, hyphenateWord } from './basic.js';\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAIA,mBAAoF;","names":[]}
+{"version":3,"sources":["../../../src/tokenize/basic/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport {\n  SentenceTokenizer,\n  WordTokenizer,\n  tokenizeParagraphs,\n  hyphenateWord,\n  splitWords,\n} from './basic.js';\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAIA,mBAMO;","names":[]}

package/dist/tokenize/basic/index.d.ts

@@ -1,2 +1,2 @@
-export { SentenceTokenizer, WordTokenizer, tokenizeParagraphs, hyphenateWord } from './basic.js';
+export { SentenceTokenizer, WordTokenizer, tokenizeParagraphs, hyphenateWord, splitWords, } from './basic.js';
 //# sourceMappingURL=index.d.ts.map

package/dist/tokenize/basic/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/tokenize/basic/index.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,kBAAkB,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/tokenize/basic/index.ts"],"names":[],"mappings":"AAIA,OAAO,EACL,iBAAiB,EACjB,aAAa,EACb,kBAAkB,EAClB,aAAa,EACb,UAAU,GACX,MAAM,YAAY,CAAC"}

package/dist/tokenize/basic/index.js

@@ -1,8 +1,15 @@
-import { SentenceTokenizer, WordTokenizer, tokenizeParagraphs, hyphenateWord } from "./basic.js";
+import {
+  SentenceTokenizer,
+  WordTokenizer,
+  tokenizeParagraphs,
+  hyphenateWord,
+  splitWords
+} from "./basic.js";
 export {
   SentenceTokenizer,
   WordTokenizer,
   hyphenateWord,
+  splitWords,
   tokenizeParagraphs
 };
 //# sourceMappingURL=index.js.map

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

@@ -1 +1 @@
-{"version":3,"sources":["../../../src/tokenize/basic/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport { SentenceTokenizer, WordTokenizer, tokenizeParagraphs, hyphenateWord } from './basic.js';\n"],"mappings":"AAIA,SAAS,mBAAmB,eAAe,oBAAoB,qBAAqB;","names":[]}
+{"version":3,"sources":["../../../src/tokenize/basic/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport {\n  SentenceTokenizer,\n  WordTokenizer,\n  tokenizeParagraphs,\n  hyphenateWord,\n  splitWords,\n} from './basic.js';\n"],"mappings":"AAIA;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;","names":[]}

package/dist/tokenize/token_stream.cjs

@@ -54,9 +54,11 @@ class BufferedTokenStream {
       if (tokens.length <= 1) break;
       if (this.#outBuf) this.#outBuf += " ";
       const tok = tokens.shift();
-      let tokText = tok;
-      if (tok.length > 1 && typeof tok[1] === "number") {
+      let tokText;
+      if (Array.isArray(tok)) {
         tokText = tok[0];
+      } else {
+        tokText = tok;
       }
       this.#outBuf += tokText;
       if (this.#outBuf.length >= this.#minTokenLength) {
@@ -79,7 +81,7 @@ class BufferedTokenStream {
       const tokens = this.#func(this.#inBuf);
       if (tokens) {
         if (this.#outBuf) this.#outBuf += " ";
-        if (typeof tokens[0] !== "string") {
+        if (Array.isArray(tokens[0])) {
           this.#outBuf += tokens.map((tok) => tok[0]).join(" ");
         } else {
           this.#outBuf += tokens.join(" ");

package/dist/tokenize/token_stream.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/tokenize/token_stream.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { randomUUID } from 'node:crypto';\nimport { AsyncIterableQueue } from '../utils.js';\nimport type { TokenData } from './tokenizer.js';\nimport { SentenceStream, WordStream } from './tokenizer.js';\n\ntype TokenizeFunc = (x: string) => string[] | [string, number, number][];\n\nexport class BufferedTokenStream implements AsyncIterableIterator<TokenData> {\n  protected queue = new AsyncIterableQueue<TokenData>();\n  protected closed = false;\n\n  #func: TokenizeFunc;\n  #minTokenLength: number;\n  #minContextLength: number;\n  #bufTokens: string[] = [];\n  #inBuf = '';\n  #outBuf = '';\n  #currentSegmentId: string;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    this.#func = func;\n    this.#minTokenLength = minTokenLength;\n    this.#minContextLength = minContextLength;\n\n    this.#currentSegmentId = randomUUID();\n  }\n\n  /** Push a string of text into the token stream */\n  pushText(text: string) {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n\n    this.#inBuf += text;\n    if (this.#inBuf.length < this.#minContextLength) return;\n\n    while (true) {\n      const tokens = this.#func(this.#inBuf);\n      if (tokens.length <= 1) break;\n\n      if (this.#outBuf) this.#outBuf += ' ';\n\n      const tok = tokens.shift()!;\n      let tokText = tok as string;\n      if (tok.length > 1 && typeof tok[1] === 'number') {\n        tokText = tok[0];\n      }\n\n      this.#outBuf += tokText;\n      if (this.#outBuf.length >= this.#minTokenLength) {\n        this.queue.put({ token: this.#outBuf, segmentId: this.#currentSegmentId });\n        this.#outBuf = '';\n      }\n\n      if (typeof tok! !== 'string') {\n        this.#inBuf = this.#inBuf.slice(tok![2]);\n      } else {\n        this.#inBuf = this.#inBuf\n          .slice(Math.max(0, this.#inBuf.indexOf(tok)) + tok.length)\n          .trimStart();\n      }\n    }\n  }\n\n  /** Flush the stream, causing it to process all pending text */\n  flush() {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n\n    if (this.#inBuf || this.#outBuf) {\n      const tokens = this.#func(this.#inBuf);\n      if (tokens) {\n        if (this.#outBuf) this.#outBuf += ' ';\n\n        if (typeof tokens[0] !== 'string') {\n          this.#outBuf += tokens.map((tok) => tok[0]).join(' ');\n        } else {\n          this.#outBuf += tokens.join(' ');\n        }\n      }\n\n      if (this.#outBuf) {\n        this.queue.put({ token: this.#outBuf, segmentId: this.#currentSegmentId });\n      }\n\n      this.#currentSegmentId = randomUUID();\n    }\n\n    this.#inBuf = '';\n    this.#outBuf = '';\n  }\n\n  /** Mark the input as ended and forbid additional pushes */\n  endInput() {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.flush();\n    this.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.queue.next();\n  }\n\n  /** Close both the input and output of the token stream */\n  close() {\n    this.queue.close();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): BufferedTokenStream {\n    return this;\n  }\n}\n\nexport class BufferedSentenceStream extends SentenceStream {\n  #stream: BufferedTokenStream;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    super();\n    this.#stream = new BufferedTokenStream(func, minTokenLength, minContextLength);\n  }\n\n  pushText(text: string) {\n    this.#stream.pushText(text);\n  }\n\n  flush() {\n    this.#stream.flush();\n  }\n\n  close() {\n    super.close();\n    this.#stream.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.#stream.next();\n  }\n}\n\nexport class BufferedWordStream extends WordStream {\n  #stream: BufferedTokenStream;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    super();\n    this.#stream = new BufferedTokenStream(func, minTokenLength, minContextLength);\n  }\n\n  pushText(text: string) {\n    this.#stream.pushText(text);\n  }\n\n  flush() {\n    this.#stream.flush();\n  }\n\n  endInput() {\n    this.#stream.endInput();\n  }\n\n  close() {\n    this.#stream.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.#stream.next();\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,yBAA2B;AAC3B,mBAAmC;AAEnC,uBAA2C;AAIpC,MAAM,oBAAgE;AAAA,EACjE,QAAQ,IAAI,gCAA8B;AAAA,EAC1C,SAAS;AAAA,EAEnB;AAAA,EACA;AAAA,EACA;AAAA,EACA,aAAuB,CAAC;AAAA,EACxB,SAAS;AAAA,EACT,UAAU;AAAA,EACV;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,SAAK,QAAQ;AACb,SAAK,kBAAkB;AACvB,SAAK,oBAAoB;AAEzB,SAAK,wBAAoB,+BAAW;AAAA,EACtC;AAAA;AAAA,EAGA,SAAS,MAAc;AACrB,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AAEA,SAAK,UAAU;AACf,QAAI,KAAK,OAAO,SAAS,KAAK,kBAAmB;AAEjD,WAAO,MAAM;AACX,YAAM,SAAS,KAAK,MAAM,KAAK,MAAM;AACrC,UAAI,OAAO,UAAU,EAAG;AAExB,UAAI,KAAK,QAAS,MAAK,WAAW;AAElC,YAAM,MAAM,OAAO,MAAM;AACzB,UAAI,UAAU;AACd,UAAI,IAAI,SAAS,KAAK,OAAO,IAAI,CAAC,MAAM,UAAU;AAChD,kBAAU,IAAI,CAAC;AAAA,MACjB;AAEA,WAAK,WAAW;AAChB,UAAI,KAAK,QAAQ,UAAU,KAAK,iBAAiB;AAC/C,aAAK,MAAM,IAAI,EAAE,OAAO,KAAK,SAAS,WAAW,KAAK,kBAAkB,CAAC;AACzE,aAAK,UAAU;AAAA,MACjB;AAEA,UAAI,OAAO,QAAS,UAAU;AAC5B,aAAK,SAAS,KAAK,OAAO,MAAM,IAAK,CAAC,CAAC;AAAA,MACzC,OAAO;AACL,aAAK,SAAS,KAAK,OAChB,MAAM,KAAK,IAAI,GAAG,KAAK,OAAO,QAAQ,GAAG,CAAC,IAAI,IAAI,MAAM,EACxD,UAAU;AAAA,MACf;AAAA,IACF;AAAA,EACF;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AAEA,QAAI,KAAK,UAAU,KAAK,SAAS;AAC/B,YAAM,SAAS,KAAK,MAAM,KAAK,MAAM;AACrC,UAAI,QAAQ;AACV,YAAI,KAAK,QAAS,MAAK,WAAW;AAElC,YAAI,OAAO,OAAO,CAAC,MAAM,UAAU;AACjC,eAAK,WAAW,OAAO,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC,EAAE,KAAK,GAAG;AAAA,QACtD,OAAO;AACL,eAAK,WAAW,OAAO,KAAK,GAAG;AAAA,QACjC;AAAA,MACF;AAEA,UAAI,KAAK,SAAS;AAChB,aAAK,MAAM,IAAI,EAAE,OAAO,KAAK,SAAS,WAAW,KAAK,kBAAkB,CAAC;AAAA,MAC3E;AAEA,WAAK,wBAAoB,+BAAW;AAAA,IACtC;AAEA,SAAK,SAAS;AACd,SAAK,UAAU;AAAA,EACjB;AAAA;AAAA,EAGA,WAAW;AACT,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM;AACX,SAAK,MAAM;AAAA,EACb;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,MAAM,KAAK;AAAA,EACzB;AAAA;AAAA,EAGA,QAAQ;AACN,SAAK,MAAM,MAAM;AACjB,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAyB;AAC5C,WAAO;AAAA,EACT;AACF;AAEO,MAAM,+BAA+B,gCAAe;AAAA,EACzD;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,UAAM;AACN,SAAK,UAAU,IAAI,oBAAoB,MAAM,gBAAgB,gBAAgB;AAAA,EAC/E;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,QAAQ,SAAS,IAAI;AAAA,EAC5B;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,QAAQ;AACN,UAAM,MAAM;AACZ,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,QAAQ,KAAK;AAAA,EAC3B;AACF;AAEO,MAAM,2BAA2B,4BAAW;AAAA,EACjD;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,UAAM;AACN,SAAK,UAAU,IAAI,oBAAoB,MAAM,gBAAgB,gBAAgB;AAAA,EAC/E;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,QAAQ,SAAS,IAAI;AAAA,EAC5B;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,WAAW;AACT,SAAK,QAAQ,SAAS;AAAA,EACxB;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,QAAQ,KAAK;AAAA,EAC3B;AACF;","names":[]}
+{"version":3,"sources":["../../src/tokenize/token_stream.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { randomUUID } from 'node:crypto';\nimport { AsyncIterableQueue } from '../utils.js';\nimport type { TokenData } from './tokenizer.js';\nimport { SentenceStream, WordStream } from './tokenizer.js';\n\ntype TokenizeFunc = (x: string) => string[] | [string, number, number][];\n\nexport class BufferedTokenStream implements AsyncIterableIterator<TokenData> {\n  protected queue = new AsyncIterableQueue<TokenData>();\n  protected closed = false;\n\n  #func: TokenizeFunc;\n  #minTokenLength: number;\n  #minContextLength: number;\n  #bufTokens: string[] = [];\n  #inBuf = '';\n  #outBuf = '';\n  #currentSegmentId: string;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    this.#func = func;\n    this.#minTokenLength = minTokenLength;\n    this.#minContextLength = minContextLength;\n\n    this.#currentSegmentId = randomUUID();\n  }\n\n  /** Push a string of text into the token stream */\n  pushText(text: string) {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n\n    this.#inBuf += text;\n    if (this.#inBuf.length < this.#minContextLength) return;\n\n    while (true) {\n      const tokens = this.#func(this.#inBuf);\n      if (tokens.length <= 1) break;\n\n      if (this.#outBuf) this.#outBuf += ' ';\n\n      const tok = tokens.shift()!;\n      let tokText: string;\n      if (Array.isArray(tok)) {\n        tokText = tok[0];\n      } else {\n        tokText = tok;\n      }\n\n      this.#outBuf += tokText;\n\n      if (this.#outBuf.length >= this.#minTokenLength) {\n        this.queue.put({ token: this.#outBuf, segmentId: this.#currentSegmentId });\n        this.#outBuf = '';\n      }\n\n      if (typeof tok! !== 'string') {\n        this.#inBuf = this.#inBuf.slice(tok![2]);\n      } else {\n        this.#inBuf = this.#inBuf\n          .slice(Math.max(0, this.#inBuf.indexOf(tok)) + tok.length)\n          .trimStart();\n      }\n    }\n  }\n\n  /** Flush the stream, causing it to process all pending text */\n  flush() {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n\n    if (this.#inBuf || this.#outBuf) {\n      const tokens = this.#func(this.#inBuf);\n      if (tokens) {\n        if (this.#outBuf) this.#outBuf += ' ';\n\n        if (Array.isArray(tokens[0])) {\n          this.#outBuf += tokens.map((tok) => tok[0]).join(' ');\n        } else {\n          this.#outBuf += tokens.join(' ');\n        }\n      }\n\n      if (this.#outBuf) {\n        this.queue.put({ token: this.#outBuf, segmentId: this.#currentSegmentId });\n      }\n\n      this.#currentSegmentId = randomUUID();\n    }\n\n    this.#inBuf = '';\n    this.#outBuf = '';\n  }\n\n  /** Mark the input as ended and forbid additional pushes */\n  endInput() {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.flush();\n    this.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.queue.next();\n  }\n\n  /** Close both the input and output of the token stream */\n  close() {\n    this.queue.close();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): BufferedTokenStream {\n    return this;\n  }\n}\n\nexport class BufferedSentenceStream extends SentenceStream {\n  #stream: BufferedTokenStream;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    super();\n    this.#stream = new BufferedTokenStream(func, minTokenLength, minContextLength);\n  }\n\n  pushText(text: string) {\n    this.#stream.pushText(text);\n  }\n\n  flush() {\n    this.#stream.flush();\n  }\n\n  close() {\n    super.close();\n    this.#stream.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.#stream.next();\n  }\n}\n\nexport class BufferedWordStream extends WordStream {\n  #stream: BufferedTokenStream;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    super();\n    this.#stream = new BufferedTokenStream(func, minTokenLength, minContextLength);\n  }\n\n  pushText(text: string) {\n    this.#stream.pushText(text);\n  }\n\n  flush() {\n    this.#stream.flush();\n  }\n\n  endInput() {\n    this.#stream.endInput();\n  }\n\n  close() {\n    this.#stream.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.#stream.next();\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,yBAA2B;AAC3B,mBAAmC;AAEnC,uBAA2C;AAIpC,MAAM,oBAAgE;AAAA,EACjE,QAAQ,IAAI,gCAA8B;AAAA,EAC1C,SAAS;AAAA,EAEnB;AAAA,EACA;AAAA,EACA;AAAA,EACA,aAAuB,CAAC;AAAA,EACxB,SAAS;AAAA,EACT,UAAU;AAAA,EACV;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,SAAK,QAAQ;AACb,SAAK,kBAAkB;AACvB,SAAK,oBAAoB;AAEzB,SAAK,wBAAoB,+BAAW;AAAA,EACtC;AAAA;AAAA,EAGA,SAAS,MAAc;AACrB,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AAEA,SAAK,UAAU;AACf,QAAI,KAAK,OAAO,SAAS,KAAK,kBAAmB;AAEjD,WAAO,MAAM;AACX,YAAM,SAAS,KAAK,MAAM,KAAK,MAAM;AACrC,UAAI,OAAO,UAAU,EAAG;AAExB,UAAI,KAAK,QAAS,MAAK,WAAW;AAElC,YAAM,MAAM,OAAO,MAAM;AACzB,UAAI;AACJ,UAAI,MAAM,QAAQ,GAAG,GAAG;AACtB,kBAAU,IAAI,CAAC;AAAA,MACjB,OAAO;AACL,kBAAU;AAAA,MACZ;AAEA,WAAK,WAAW;AAEhB,UAAI,KAAK,QAAQ,UAAU,KAAK,iBAAiB;AAC/C,aAAK,MAAM,IAAI,EAAE,OAAO,KAAK,SAAS,WAAW,KAAK,kBAAkB,CAAC;AACzE,aAAK,UAAU;AAAA,MACjB;AAEA,UAAI,OAAO,QAAS,UAAU;AAC5B,aAAK,SAAS,KAAK,OAAO,MAAM,IAAK,CAAC,CAAC;AAAA,MACzC,OAAO;AACL,aAAK,SAAS,KAAK,OAChB,MAAM,KAAK,IAAI,GAAG,KAAK,OAAO,QAAQ,GAAG,CAAC,IAAI,IAAI,MAAM,EACxD,UAAU;AAAA,MACf;AAAA,IACF;AAAA,EACF;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AAEA,QAAI,KAAK,UAAU,KAAK,SAAS;AAC/B,YAAM,SAAS,KAAK,MAAM,KAAK,MAAM;AACrC,UAAI,QAAQ;AACV,YAAI,KAAK,QAAS,MAAK,WAAW;AAElC,YAAI,MAAM,QAAQ,OAAO,CAAC,CAAC,GAAG;AAC5B,eAAK,WAAW,OAAO,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC,EAAE,KAAK,GAAG;AAAA,QACtD,OAAO;AACL,eAAK,WAAW,OAAO,KAAK,GAAG;AAAA,QACjC;AAAA,MACF;AAEA,UAAI,KAAK,SAAS;AAChB,aAAK,MAAM,IAAI,EAAE,OAAO,KAAK,SAAS,WAAW,KAAK,kBAAkB,CAAC;AAAA,MAC3E;AAEA,WAAK,wBAAoB,+BAAW;AAAA,IACtC;AAEA,SAAK,SAAS;AACd,SAAK,UAAU;AAAA,EACjB;AAAA;AAAA,EAGA,WAAW;AACT,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM;AACX,SAAK,MAAM;AAAA,EACb;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,MAAM,KAAK;AAAA,EACzB;AAAA;AAAA,EAGA,QAAQ;AACN,SAAK,MAAM,MAAM;AACjB,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAyB;AAC5C,WAAO;AAAA,EACT;AACF;AAEO,MAAM,+BAA+B,gCAAe;AAAA,EACzD;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,UAAM;AACN,SAAK,UAAU,IAAI,oBAAoB,MAAM,gBAAgB,gBAAgB;AAAA,EAC/E;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,QAAQ,SAAS,IAAI;AAAA,EAC5B;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,QAAQ;AACN,UAAM,MAAM;AACZ,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,QAAQ,KAAK;AAAA,EAC3B;AACF;AAEO,MAAM,2BAA2B,4BAAW;AAAA,EACjD;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,UAAM;AACN,SAAK,UAAU,IAAI,oBAAoB,MAAM,gBAAgB,gBAAgB;AAAA,EAC/E;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,QAAQ,SAAS,IAAI;AAAA,EAC5B;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,WAAW;AACT,SAAK,QAAQ,SAAS;AAAA,EACxB;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,QAAQ,KAAK;AAAA,EAC3B;AACF;","names":[]}

package/dist/tokenize/token_stream.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"token_stream.d.ts","sourceRoot":"","sources":["../../src/tokenize/token_stream.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAC;AACjD,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAChD,OAAO,EAAE,cAAc,EAAE,UAAU,EAAE,MAAM,gBAAgB,CAAC;AAE5D,KAAK,YAAY,GAAG,CAAC,CAAC,EAAE,MAAM,KAAK,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CAAC;AAEzE,qBAAa,mBAAoB,YAAW,qBAAqB,CAAC,SAAS,CAAC;;IAC1E,SAAS,CAAC,KAAK,gCAAuC;IACtD,SAAS,CAAC,MAAM,UAAS;gBAUb,IAAI,EAAE,YAAY,EAAE,cAAc,EAAE,MAAM,EAAE,gBAAgB,EAAE,MAAM;IAQhF,kDAAkD;IAClD,QAAQ,CAAC,IAAI,EAAE,MAAM;IAoCrB,+DAA+D;IAC/D,KAAK;IA4BL,2DAA2D;IAC3D,QAAQ;IAQR,IAAI,IAAI,OAAO,CAAC,cAAc,CAAC,SAAS,CAAC,CAAC;IAI1C,0DAA0D;IAC1D,KAAK;IAKL,CAAC,MAAM,CAAC,aAAa,CAAC,IAAI,mBAAmB;CAG9C;AAED,qBAAa,sBAAuB,SAAQ,cAAc;;gBAG5C,IAAI,EAAE,YAAY,EAAE,cAAc,EAAE,MAAM,EAAE,gBAAgB,EAAE,MAAM;IAKhF,QAAQ,CAAC,IAAI,EAAE,MAAM;IAIrB,KAAK;IAIL,KAAK;IAKL,IAAI,IAAI,OAAO,CAAC,cAAc,CAAC,SAAS,CAAC,CAAC;CAG3C;AAED,qBAAa,kBAAmB,SAAQ,UAAU;;gBAGpC,IAAI,EAAE,YAAY,EAAE,cAAc,EAAE,MAAM,EAAE,gBAAgB,EAAE,MAAM;IAKhF,QAAQ,CAAC,IAAI,EAAE,MAAM;IAIrB,KAAK;IAIL,QAAQ;IAIR,KAAK;IAIL,IAAI,IAAI,OAAO,CAAC,cAAc,CAAC,SAAS,CAAC,CAAC;CAG3C"}
+{"version":3,"file":"token_stream.d.ts","sourceRoot":"","sources":["../../src/tokenize/token_stream.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAC;AACjD,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAChD,OAAO,EAAE,cAAc,EAAE,UAAU,EAAE,MAAM,gBAAgB,CAAC;AAE5D,KAAK,YAAY,GAAG,CAAC,CAAC,EAAE,MAAM,KAAK,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CAAC;AAEzE,qBAAa,mBAAoB,YAAW,qBAAqB,CAAC,SAAS,CAAC;;IAC1E,SAAS,CAAC,KAAK,gCAAuC;IACtD,SAAS,CAAC,MAAM,UAAS;gBAUb,IAAI,EAAE,YAAY,EAAE,cAAc,EAAE,MAAM,EAAE,gBAAgB,EAAE,MAAM;IAQhF,kDAAkD;IAClD,QAAQ,CAAC,IAAI,EAAE,MAAM;IAuCrB,+DAA+D;IAC/D,KAAK;IA4BL,2DAA2D;IAC3D,QAAQ;IAQR,IAAI,IAAI,OAAO,CAAC,cAAc,CAAC,SAAS,CAAC,CAAC;IAI1C,0DAA0D;IAC1D,KAAK;IAKL,CAAC,MAAM,CAAC,aAAa,CAAC,IAAI,mBAAmB;CAG9C;AAED,qBAAa,sBAAuB,SAAQ,cAAc;;gBAG5C,IAAI,EAAE,YAAY,EAAE,cAAc,EAAE,MAAM,EAAE,gBAAgB,EAAE,MAAM;IAKhF,QAAQ,CAAC,IAAI,EAAE,MAAM;IAIrB,KAAK;IAIL,KAAK;IAKL,IAAI,IAAI,OAAO,CAAC,cAAc,CAAC,SAAS,CAAC,CAAC;CAG3C;AAED,qBAAa,kBAAmB,SAAQ,UAAU;;gBAGpC,IAAI,EAAE,YAAY,EAAE,cAAc,EAAE,MAAM,EAAE,gBAAgB,EAAE,MAAM;IAKhF,QAAQ,CAAC,IAAI,EAAE,MAAM;IAIrB,KAAK;IAIL,QAAQ;IAIR,KAAK;IAIL,IAAI,IAAI,OAAO,CAAC,cAAc,CAAC,SAAS,CAAC,CAAC;CAG3C"}

package/dist/tokenize/token_stream.js

@@ -29,9 +29,11 @@ class BufferedTokenStream {
       if (tokens.length <= 1) break;
       if (this.#outBuf) this.#outBuf += " ";
       const tok = tokens.shift();
-      let tokText = tok;
-      if (tok.length > 1 && typeof tok[1] === "number") {
+      let tokText;
+      if (Array.isArray(tok)) {
         tokText = tok[0];
+      } else {
+        tokText = tok;
       }
       this.#outBuf += tokText;
       if (this.#outBuf.length >= this.#minTokenLength) {
@@ -54,7 +56,7 @@ class BufferedTokenStream {
       const tokens = this.#func(this.#inBuf);
       if (tokens) {
         if (this.#outBuf) this.#outBuf += " ";
-        if (typeof tokens[0] !== "string") {
+        if (Array.isArray(tokens[0])) {
           this.#outBuf += tokens.map((tok) => tok[0]).join(" ");
         } else {
           this.#outBuf += tokens.join(" ");

package/dist/tokenize/token_stream.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/tokenize/token_stream.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { randomUUID } from 'node:crypto';\nimport { AsyncIterableQueue } from '../utils.js';\nimport type { TokenData } from './tokenizer.js';\nimport { SentenceStream, WordStream } from './tokenizer.js';\n\ntype TokenizeFunc = (x: string) => string[] | [string, number, number][];\n\nexport class BufferedTokenStream implements AsyncIterableIterator<TokenData> {\n  protected queue = new AsyncIterableQueue<TokenData>();\n  protected closed = false;\n\n  #func: TokenizeFunc;\n  #minTokenLength: number;\n  #minContextLength: number;\n  #bufTokens: string[] = [];\n  #inBuf = '';\n  #outBuf = '';\n  #currentSegmentId: string;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    this.#func = func;\n    this.#minTokenLength = minTokenLength;\n    this.#minContextLength = minContextLength;\n\n    this.#currentSegmentId = randomUUID();\n  }\n\n  /** Push a string of text into the token stream */\n  pushText(text: string) {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n\n    this.#inBuf += text;\n    if (this.#inBuf.length < this.#minContextLength) return;\n\n    while (true) {\n      const tokens = this.#func(this.#inBuf);\n      if (tokens.length <= 1) break;\n\n      if (this.#outBuf) this.#outBuf += ' ';\n\n      const tok = tokens.shift()!;\n      let tokText = tok as string;\n      if (tok.length > 1 && typeof tok[1] === 'number') {\n        tokText = tok[0];\n      }\n\n      this.#outBuf += tokText;\n      if (this.#outBuf.length >= this.#minTokenLength) {\n        this.queue.put({ token: this.#outBuf, segmentId: this.#currentSegmentId });\n        this.#outBuf = '';\n      }\n\n      if (typeof tok! !== 'string') {\n        this.#inBuf = this.#inBuf.slice(tok![2]);\n      } else {\n        this.#inBuf = this.#inBuf\n          .slice(Math.max(0, this.#inBuf.indexOf(tok)) + tok.length)\n          .trimStart();\n      }\n    }\n  }\n\n  /** Flush the stream, causing it to process all pending text */\n  flush() {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n\n    if (this.#inBuf || this.#outBuf) {\n      const tokens = this.#func(this.#inBuf);\n      if (tokens) {\n        if (this.#outBuf) this.#outBuf += ' ';\n\n        if (typeof tokens[0] !== 'string') {\n          this.#outBuf += tokens.map((tok) => tok[0]).join(' ');\n        } else {\n          this.#outBuf += tokens.join(' ');\n        }\n      }\n\n      if (this.#outBuf) {\n        this.queue.put({ token: this.#outBuf, segmentId: this.#currentSegmentId });\n      }\n\n      this.#currentSegmentId = randomUUID();\n    }\n\n    this.#inBuf = '';\n    this.#outBuf = '';\n  }\n\n  /** Mark the input as ended and forbid additional pushes */\n  endInput() {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.flush();\n    this.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.queue.next();\n  }\n\n  /** Close both the input and output of the token stream */\n  close() {\n    this.queue.close();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): BufferedTokenStream {\n    return this;\n  }\n}\n\nexport class BufferedSentenceStream extends SentenceStream {\n  #stream: BufferedTokenStream;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    super();\n    this.#stream = new BufferedTokenStream(func, minTokenLength, minContextLength);\n  }\n\n  pushText(text: string) {\n    this.#stream.pushText(text);\n  }\n\n  flush() {\n    this.#stream.flush();\n  }\n\n  close() {\n    super.close();\n    this.#stream.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.#stream.next();\n  }\n}\n\nexport class BufferedWordStream extends WordStream {\n  #stream: BufferedTokenStream;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    super();\n    this.#stream = new BufferedTokenStream(func, minTokenLength, minContextLength);\n  }\n\n  pushText(text: string) {\n    this.#stream.pushText(text);\n  }\n\n  flush() {\n    this.#stream.flush();\n  }\n\n  endInput() {\n    this.#stream.endInput();\n  }\n\n  close() {\n    this.#stream.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.#stream.next();\n  }\n}\n"],"mappings":"AAGA,SAAS,kBAAkB;AAC3B,SAAS,0BAA0B;AAEnC,SAAS,gBAAgB,kBAAkB;AAIpC,MAAM,oBAAgE;AAAA,EACjE,QAAQ,IAAI,mBAA8B;AAAA,EAC1C,SAAS;AAAA,EAEnB;AAAA,EACA;AAAA,EACA;AAAA,EACA,aAAuB,CAAC;AAAA,EACxB,SAAS;AAAA,EACT,UAAU;AAAA,EACV;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,SAAK,QAAQ;AACb,SAAK,kBAAkB;AACvB,SAAK,oBAAoB;AAEzB,SAAK,oBAAoB,WAAW;AAAA,EACtC;AAAA;AAAA,EAGA,SAAS,MAAc;AACrB,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AAEA,SAAK,UAAU;AACf,QAAI,KAAK,OAAO,SAAS,KAAK,kBAAmB;AAEjD,WAAO,MAAM;AACX,YAAM,SAAS,KAAK,MAAM,KAAK,MAAM;AACrC,UAAI,OAAO,UAAU,EAAG;AAExB,UAAI,KAAK,QAAS,MAAK,WAAW;AAElC,YAAM,MAAM,OAAO,MAAM;AACzB,UAAI,UAAU;AACd,UAAI,IAAI,SAAS,KAAK,OAAO,IAAI,CAAC,MAAM,UAAU;AAChD,kBAAU,IAAI,CAAC;AAAA,MACjB;AAEA,WAAK,WAAW;AAChB,UAAI,KAAK,QAAQ,UAAU,KAAK,iBAAiB;AAC/C,aAAK,MAAM,IAAI,EAAE,OAAO,KAAK,SAAS,WAAW,KAAK,kBAAkB,CAAC;AACzE,aAAK,UAAU;AAAA,MACjB;AAEA,UAAI,OAAO,QAAS,UAAU;AAC5B,aAAK,SAAS,KAAK,OAAO,MAAM,IAAK,CAAC,CAAC;AAAA,MACzC,OAAO;AACL,aAAK,SAAS,KAAK,OAChB,MAAM,KAAK,IAAI,GAAG,KAAK,OAAO,QAAQ,GAAG,CAAC,IAAI,IAAI,MAAM,EACxD,UAAU;AAAA,MACf;AAAA,IACF;AAAA,EACF;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AAEA,QAAI,KAAK,UAAU,KAAK,SAAS;AAC/B,YAAM,SAAS,KAAK,MAAM,KAAK,MAAM;AACrC,UAAI,QAAQ;AACV,YAAI,KAAK,QAAS,MAAK,WAAW;AAElC,YAAI,OAAO,OAAO,CAAC,MAAM,UAAU;AACjC,eAAK,WAAW,OAAO,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC,EAAE,KAAK,GAAG;AAAA,QACtD,OAAO;AACL,eAAK,WAAW,OAAO,KAAK,GAAG;AAAA,QACjC;AAAA,MACF;AAEA,UAAI,KAAK,SAAS;AAChB,aAAK,MAAM,IAAI,EAAE,OAAO,KAAK,SAAS,WAAW,KAAK,kBAAkB,CAAC;AAAA,MAC3E;AAEA,WAAK,oBAAoB,WAAW;AAAA,IACtC;AAEA,SAAK,SAAS;AACd,SAAK,UAAU;AAAA,EACjB;AAAA;AAAA,EAGA,WAAW;AACT,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM;AACX,SAAK,MAAM;AAAA,EACb;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,MAAM,KAAK;AAAA,EACzB;AAAA;AAAA,EAGA,QAAQ;AACN,SAAK,MAAM,MAAM;AACjB,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAyB;AAC5C,WAAO;AAAA,EACT;AACF;AAEO,MAAM,+BAA+B,eAAe;AAAA,EACzD;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,UAAM;AACN,SAAK,UAAU,IAAI,oBAAoB,MAAM,gBAAgB,gBAAgB;AAAA,EAC/E;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,QAAQ,SAAS,IAAI;AAAA,EAC5B;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,QAAQ;AACN,UAAM,MAAM;AACZ,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,QAAQ,KAAK;AAAA,EAC3B;AACF;AAEO,MAAM,2BAA2B,WAAW;AAAA,EACjD;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,UAAM;AACN,SAAK,UAAU,IAAI,oBAAoB,MAAM,gBAAgB,gBAAgB;AAAA,EAC/E;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,QAAQ,SAAS,IAAI;AAAA,EAC5B;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,WAAW;AACT,SAAK,QAAQ,SAAS;AAAA,EACxB;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,QAAQ,KAAK;AAAA,EAC3B;AACF;","names":[]}
+{"version":3,"sources":["../../src/tokenize/token_stream.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { randomUUID } from 'node:crypto';\nimport { AsyncIterableQueue } from '../utils.js';\nimport type { TokenData } from './tokenizer.js';\nimport { SentenceStream, WordStream } from './tokenizer.js';\n\ntype TokenizeFunc = (x: string) => string[] | [string, number, number][];\n\nexport class BufferedTokenStream implements AsyncIterableIterator<TokenData> {\n  protected queue = new AsyncIterableQueue<TokenData>();\n  protected closed = false;\n\n  #func: TokenizeFunc;\n  #minTokenLength: number;\n  #minContextLength: number;\n  #bufTokens: string[] = [];\n  #inBuf = '';\n  #outBuf = '';\n  #currentSegmentId: string;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    this.#func = func;\n    this.#minTokenLength = minTokenLength;\n    this.#minContextLength = minContextLength;\n\n    this.#currentSegmentId = randomUUID();\n  }\n\n  /** Push a string of text into the token stream */\n  pushText(text: string) {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n\n    this.#inBuf += text;\n    if (this.#inBuf.length < this.#minContextLength) return;\n\n    while (true) {\n      const tokens = this.#func(this.#inBuf);\n      if (tokens.length <= 1) break;\n\n      if (this.#outBuf) this.#outBuf += ' ';\n\n      const tok = tokens.shift()!;\n      let tokText: string;\n      if (Array.isArray(tok)) {\n        tokText = tok[0];\n      } else {\n        tokText = tok;\n      }\n\n      this.#outBuf += tokText;\n\n      if (this.#outBuf.length >= this.#minTokenLength) {\n        this.queue.put({ token: this.#outBuf, segmentId: this.#currentSegmentId });\n        this.#outBuf = '';\n      }\n\n      if (typeof tok! !== 'string') {\n        this.#inBuf = this.#inBuf.slice(tok![2]);\n      } else {\n        this.#inBuf = this.#inBuf\n          .slice(Math.max(0, this.#inBuf.indexOf(tok)) + tok.length)\n          .trimStart();\n      }\n    }\n  }\n\n  /** Flush the stream, causing it to process all pending text */\n  flush() {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n\n    if (this.#inBuf || this.#outBuf) {\n      const tokens = this.#func(this.#inBuf);\n      if (tokens) {\n        if (this.#outBuf) this.#outBuf += ' ';\n\n        if (Array.isArray(tokens[0])) {\n          this.#outBuf += tokens.map((tok) => tok[0]).join(' ');\n        } else {\n          this.#outBuf += tokens.join(' ');\n        }\n      }\n\n      if (this.#outBuf) {\n        this.queue.put({ token: this.#outBuf, segmentId: this.#currentSegmentId });\n      }\n\n      this.#currentSegmentId = randomUUID();\n    }\n\n    this.#inBuf = '';\n    this.#outBuf = '';\n  }\n\n  /** Mark the input as ended and forbid additional pushes */\n  endInput() {\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.flush();\n    this.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.queue.next();\n  }\n\n  /** Close both the input and output of the token stream */\n  close() {\n    this.queue.close();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): BufferedTokenStream {\n    return this;\n  }\n}\n\nexport class BufferedSentenceStream extends SentenceStream {\n  #stream: BufferedTokenStream;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    super();\n    this.#stream = new BufferedTokenStream(func, minTokenLength, minContextLength);\n  }\n\n  pushText(text: string) {\n    this.#stream.pushText(text);\n  }\n\n  flush() {\n    this.#stream.flush();\n  }\n\n  close() {\n    super.close();\n    this.#stream.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.#stream.next();\n  }\n}\n\nexport class BufferedWordStream extends WordStream {\n  #stream: BufferedTokenStream;\n\n  constructor(func: TokenizeFunc, minTokenLength: number, minContextLength: number) {\n    super();\n    this.#stream = new BufferedTokenStream(func, minTokenLength, minContextLength);\n  }\n\n  pushText(text: string) {\n    this.#stream.pushText(text);\n  }\n\n  flush() {\n    this.#stream.flush();\n  }\n\n  endInput() {\n    this.#stream.endInput();\n  }\n\n  close() {\n    this.#stream.close();\n  }\n\n  next(): Promise<IteratorResult<TokenData>> {\n    return this.#stream.next();\n  }\n}\n"],"mappings":"AAGA,SAAS,kBAAkB;AAC3B,SAAS,0BAA0B;AAEnC,SAAS,gBAAgB,kBAAkB;AAIpC,MAAM,oBAAgE;AAAA,EACjE,QAAQ,IAAI,mBAA8B;AAAA,EAC1C,SAAS;AAAA,EAEnB;AAAA,EACA;AAAA,EACA;AAAA,EACA,aAAuB,CAAC;AAAA,EACxB,SAAS;AAAA,EACT,UAAU;AAAA,EACV;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,SAAK,QAAQ;AACb,SAAK,kBAAkB;AACvB,SAAK,oBAAoB;AAEzB,SAAK,oBAAoB,WAAW;AAAA,EACtC;AAAA;AAAA,EAGA,SAAS,MAAc;AACrB,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AAEA,SAAK,UAAU;AACf,QAAI,KAAK,OAAO,SAAS,KAAK,kBAAmB;AAEjD,WAAO,MAAM;AACX,YAAM,SAAS,KAAK,MAAM,KAAK,MAAM;AACrC,UAAI,OAAO,UAAU,EAAG;AAExB,UAAI,KAAK,QAAS,MAAK,WAAW;AAElC,YAAM,MAAM,OAAO,MAAM;AACzB,UAAI;AACJ,UAAI,MAAM,QAAQ,GAAG,GAAG;AACtB,kBAAU,IAAI,CAAC;AAAA,MACjB,OAAO;AACL,kBAAU;AAAA,MACZ;AAEA,WAAK,WAAW;AAEhB,UAAI,KAAK,QAAQ,UAAU,KAAK,iBAAiB;AAC/C,aAAK,MAAM,IAAI,EAAE,OAAO,KAAK,SAAS,WAAW,KAAK,kBAAkB,CAAC;AACzE,aAAK,UAAU;AAAA,MACjB;AAEA,UAAI,OAAO,QAAS,UAAU;AAC5B,aAAK,SAAS,KAAK,OAAO,MAAM,IAAK,CAAC,CAAC;AAAA,MACzC,OAAO;AACL,aAAK,SAAS,KAAK,OAChB,MAAM,KAAK,IAAI,GAAG,KAAK,OAAO,QAAQ,GAAG,CAAC,IAAI,IAAI,MAAM,EACxD,UAAU;AAAA,MACf;AAAA,IACF;AAAA,EACF;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AAEA,QAAI,KAAK,UAAU,KAAK,SAAS;AAC/B,YAAM,SAAS,KAAK,MAAM,KAAK,MAAM;AACrC,UAAI,QAAQ;AACV,YAAI,KAAK,QAAS,MAAK,WAAW;AAElC,YAAI,MAAM,QAAQ,OAAO,CAAC,CAAC,GAAG;AAC5B,eAAK,WAAW,OAAO,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC,EAAE,KAAK,GAAG;AAAA,QACtD,OAAO;AACL,eAAK,WAAW,OAAO,KAAK,GAAG;AAAA,QACjC;AAAA,MACF;AAEA,UAAI,KAAK,SAAS;AAChB,aAAK,MAAM,IAAI,EAAE,OAAO,KAAK,SAAS,WAAW,KAAK,kBAAkB,CAAC;AAAA,MAC3E;AAEA,WAAK,oBAAoB,WAAW;AAAA,IACtC;AAEA,SAAK,SAAS;AACd,SAAK,UAAU;AAAA,EACjB;AAAA;AAAA,EAGA,WAAW;AACT,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM;AACX,SAAK,MAAM;AAAA,EACb;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,MAAM,KAAK;AAAA,EACzB;AAAA;AAAA,EAGA,QAAQ;AACN,SAAK,MAAM,MAAM;AACjB,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAyB;AAC5C,WAAO;AAAA,EACT;AACF;AAEO,MAAM,+BAA+B,eAAe;AAAA,EACzD;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,UAAM;AACN,SAAK,UAAU,IAAI,oBAAoB,MAAM,gBAAgB,gBAAgB;AAAA,EAC/E;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,QAAQ,SAAS,IAAI;AAAA,EAC5B;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,QAAQ;AACN,UAAM,MAAM;AACZ,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,QAAQ,KAAK;AAAA,EAC3B;AACF;AAEO,MAAM,2BAA2B,WAAW;AAAA,EACjD;AAAA,EAEA,YAAY,MAAoB,gBAAwB,kBAA0B;AAChF,UAAM;AACN,SAAK,UAAU,IAAI,oBAAoB,MAAM,gBAAgB,gBAAgB;AAAA,EAC/E;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,QAAQ,SAAS,IAAI;AAAA,EAC5B;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,WAAW;AACT,SAAK,QAAQ,SAAS;AAAA,EACxB;AAAA,EAEA,QAAQ;AACN,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,OAA2C;AACzC,WAAO,KAAK,QAAQ,KAAK;AAAA,EAC3B;AACF;","names":[]}

package/dist/transcription.cjs

@@ -18,114 +18,231 @@ var __copyProps = (to, from, except, desc) => {
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 var transcription_exports = {};
 __export(transcription_exports, {
-  BasicTranscriptionForwarder: () => BasicTranscriptionForwarder
+  TextAudioSynchronizer: () => TextAudioSynchronizer,
+  defaultTextSyncOptions: () => defaultTextSyncOptions
 });
 module.exports = __toCommonJS(transcription_exports);
-var import_log = require("./log.cjs");
-class BasicTranscriptionForwarder {
-  #room;
-  #participantIdentity;
-  #trackSid;
-  #currentText = "";
-  #totalAudioDuration = 0;
-  #currentPlayoutTime = 0;
-  #DEFAULT_CHARS_PER_SECOND = 16;
-  #charsPerSecond = this.#DEFAULT_CHARS_PER_SECOND;
-  #messageId;
-  #isRunning = false;
-  #logger = (0, import_log.log)();
-  currentCharacterIndex = 0;
-  constructor(room, participantIdentity, trackSid, messageId) {
-    this.#room = room;
-    this.#participantIdentity = participantIdentity;
-    this.#trackSid = trackSid;
-    this.#messageId = messageId;
-  }
-  get text() {
-    return this.#currentText;
-  }
-  start() {
-    if (!this.#isRunning) {
-      this.#isRunning = true;
-      this.#startPublishingLoop().catch((error) => {
-        this.#logger.error("Error in publishing loop:", error);
-        this.#isRunning = false;
-      });
-    }
+var import_protocol = require("@livekit/protocol");
+var import_rtc_node = require("@livekit/rtc-node");
+var import_node_crypto = require("node:crypto");
+var import_node_events = require("node:events");
+var import_tokenize = require("./tokenize/index.cjs");
+var import_utils = require("./utils.cjs");
+const STANDARD_SPEECH_RATE = 3830;
+const defaultTextSyncOptions = {
+  language: "",
+  speed: 1,
+  newSentenceDelay: 400,
+  sentenceTokenizer: new import_tokenize.basic.SentenceTokenizer(),
+  hyphenateWord: import_tokenize.basic.hyphenateWord,
+  splitWords: import_tokenize.basic.splitWords
+};
+class TextAudioSynchronizer extends import_node_events.EventEmitter {
+  #opts;
+  #speed;
+  #closed = false;
+  #interrupted = false;
+  #closeFut = new import_utils.Future();
+  #playingSegIndex = -1;
+  #finishedSegIndex = -1;
+  #textQChanged = new import_utils.AsyncIterableQueue();
+  #textQ = [];
+  #audioQChanged = new import_utils.AsyncIterableQueue();
+  #audioQ = [];
+  #playedText = "";
+  #task;
+  #audioData;
+  #textData;
+  constructor(opts) {
+    super();
+    this.#opts = opts;
+    this.#speed = opts.speed * STANDARD_SPEECH_RATE;
   }
   pushAudio(frame) {
-    this.#totalAudioDuration += frame.samplesPerChannel / frame.sampleRate;
+    this.#checkNotClosed();
+    if (!this.#audioData) {
+      this.#audioData = { pushedDuration: 0, done: false };
+      this.#audioQ.push(this.#audioData);
+      this.#audioQChanged.put(1);
+    }
+    this.#audioData.pushedDuration += frame.samplesPerChannel / frame.sampleRate;
   }
   pushText(text) {
-    this.#currentText += text;
+    this.#checkNotClosed();
+    if (!this.#textData) {
+      this.#textData = {
+        sentenceStream: this.#opts.sentenceTokenizer.stream(),
+        pushedText: "",
+        done: false,
+        forwardedHyphens: 0,
+        forwardedSentences: 0
+      };
+      this.#textQ.push(this.#textData);
+      this.#textQChanged.put(1);
+    }
+    this.#textData.pushedText += text;
+    this.#textData.sentenceStream.pushText(text);
   }
-  #textIsComplete = false;
-  #audioIsComplete = false;
-  markTextComplete() {
-    this.#textIsComplete = true;
-    this.#adjustTimingIfBothFinished();
+  markAudioSegmentEnd() {
+    this.#checkNotClosed();
+    if (!this.#audioData) {
+      this.pushAudio(new import_rtc_node.AudioFrame(new Int16Array(), 24e3, 1, 0));
+    }
+    this.#audioData.done = true;
+    this.#audioData = void 0;
   }
-  markAudioComplete() {
-    this.#audioIsComplete = true;
-    this.#adjustTimingIfBothFinished();
+  markTextSegmentEnd() {
+    var _a, _b;
+    this.#checkNotClosed();
+    if (!this.#textData) {
+      this.pushText("");
+    }
+    this.#textData.done = true;
+    (_a = this.#textData) == null ? void 0 : _a.sentenceStream.flush();
+    (_b = this.#textData) == null ? void 0 : _b.sentenceStream.close();
+    this.#textData = void 0;
   }
-  #adjustTimingIfBothFinished() {
-    if (this.#textIsComplete && this.#audioIsComplete) {
-      const actualDuration = this.#totalAudioDuration;
-      if (actualDuration > 0 && this.#currentText.length > 0) {
-        this.#charsPerSecond = this.#currentText.length / actualDuration;
-      }
+  segmentPlayoutStarted() {
+    this.#checkNotClosed();
+    this.#playingSegIndex++;
+    if (!this.#task) {
+      this.#task = this.#mainLoop();
     }
   }
-  #computeSleepInterval() {
-    return Math.min(Math.max(1 / this.#charsPerSecond, 0.0625), 0.5);
+  segmentPlayoutFinished() {
+    this.#checkNotClosed();
+    this.#finishedSegIndex++;
   }
-  async #startPublishingLoop() {
-    this.#isRunning = true;
-    let sleepInterval = this.#computeSleepInterval();
-    let isComplete = false;
-    while (this.#isRunning && !isComplete) {
-      this.#currentPlayoutTime += sleepInterval;
-      this.currentCharacterIndex = Math.floor(this.#currentPlayoutTime * this.#charsPerSecond);
-      isComplete = this.#textIsComplete && this.currentCharacterIndex >= this.#currentText.length;
-      await this.#publishTranscription(false);
-      if (this.#isRunning && !isComplete) {
-        sleepInterval = this.#computeSleepInterval();
-        await new Promise((resolve) => setTimeout(resolve, sleepInterval * 1e3));
-      }
+  get playedText() {
+    return this.#playedText;
+  }
+  async close(interrupt) {
+    if (this.#closed) {
+      return;
+    }
+    this.#closed = true;
+    this.#interrupted = interrupt;
+    this.#closeFut.resolve();
+    for (const textData of this.#textQ) {
+      textData == null ? void 0 : textData.sentenceStream.close();
     }
-    if (this.#isRunning) {
-      this.close(false);
+    this.#textQ.push(void 0);
+    this.#audioQ.push(void 0);
+    this.#textQChanged.put(1);
+    this.#audioQChanged.put(1);
+    await this.#task;
+  }
+  async #mainLoop() {
+    let segIndex = 0;
+    let qDone = false;
+    while (!qDone) {
+      await this.#textQChanged.next();
+      await this.#audioQChanged.next();
+      while (this.#textQ.length && this.#audioQ.length) {
+        const textData = this.#textQ.pop();
+        const audioData = this.#audioQ.pop();
+        if (!(textData && audioData)) {
+          qDone = true;
+          break;
+        }
+        while (!this.#closed) {
+          if (this.#playingSegIndex >= segIndex) break;
+          await this.#sleepIfNotClosed(125);
+        }
+        const sentenceStream = textData.sentenceStream;
+        const forwardStartTime = Date.now();
+        for await (const ev of sentenceStream) {
+          await this.#syncSentence(segIndex, forwardStartTime, textData, audioData, ev.token);
+        }
+        segIndex++;
+      }
     }
   }
-  async #publishTranscription(final) {
-    var _a;
-    const textToPublish = this.#currentText.slice(0, this.currentCharacterIndex);
-    await ((_a = this.#room.localParticipant) == null ? void 0 : _a.publishTranscription({
-      participantIdentity: this.#participantIdentity,
-      trackSid: this.#trackSid,
-      segments: [
-        {
-          text: textToPublish,
-          final,
-          id: this.#messageId,
+  async #syncSentence(segIndex, segStartTime, textData, audioData, sentence) {
+    let realSpeed;
+    if (audioData.pushedDuration > 0 && audioData.done) {
+      realSpeed = this.#calcHyphens(textData.pushedText).length / audioData.pushedDuration;
+    }
+    const segId = "SG_" + (0, import_node_crypto.randomUUID)();
+    const words = this.#opts.splitWords(sentence);
+    const processedWords = [];
+    const ogText = this.#playedText;
+    for (const [word, _, end] of words) {
+      if (segIndex <= this.#finishedSegIndex) break;
+      if (this.#interrupted) return;
+      const wordHyphens = this.#opts.hyphenateWord(word).length;
+      processedWords.push(word);
+      const elapsed = Date.now() - segStartTime;
+      const text = sentence.slice(0, end);
+      let speed = this.#speed;
+      let delay;
+      if (realSpeed) {
+        speed = realSpeed;
+        const estimatedPausesMs = textData.forwardedSentences * this.#opts.newSentenceDelay;
+        const hyphPauses = estimatedPausesMs * speed;
+        const targetHyphens = Math.round(speed * elapsed);
+        const dt = targetHyphens - textData.forwardedHyphens - hyphPauses;
+        const toWaitHyphens = Math.max(0, wordHyphens - dt);
+        delay = toWaitHyphens / speed;
+      } else {
+        delay = wordHyphens / speed;
+      }
+      const firstDelay = Math.min(delay / 2, 2 / speed);
+      await this.#sleepIfNotClosed(firstDelay * 1e6);
+      this.emit(
+        "textUpdated",
+        new import_protocol.TranscriptionSegment({
+          id: segId,
+          text,
           startTime: BigInt(0),
           endTime: BigInt(0),
-          language: ""
-        }
-      ]
-    }));
+          final: false,
+          language: this.#opts.language
+        })
+      );
+      this.#playedText = `${ogText} ${text}`;
+      await this.#sleepIfNotClosed((delay - firstDelay) * 1e6);
+      textData.forwardedHyphens += wordHyphens;
+    }
+    this.emit(
+      "textUpdated",
+      new import_protocol.TranscriptionSegment({
+        id: segId,
+        text: sentence,
+        startTime: BigInt(0),
+        endTime: BigInt(0),
+        final: true,
+        language: this.#opts.language
+      })
+    );
+    this.#playedText = `${ogText} ${sentence}`;
+    await this.#sleepIfNotClosed(this.#opts.newSentenceDelay);
+    textData.forwardedSentences++;
   }
-  async close(interrupt) {
-    this.#isRunning = false;
-    if (!interrupt) {
-      this.currentCharacterIndex = this.#currentText.length;
+  async #sleepIfNotClosed(delay) {
+    await Promise.race([
+      this.#closeFut.await,
+      new Promise((resolve) => setTimeout(resolve, delay))
+    ]);
+  }
+  #calcHyphens(text) {
+    const hyphens = [];
+    const words = this.#opts.splitWords(text);
+    for (const word of words) {
+      const n = this.#opts.hyphenateWord(word[0]);
+      hyphens.push(...n);
+    }
+    return hyphens;
+  }
+  #checkNotClosed() {
+    if (this.#closed) {
+      throw new Error("TextAudioSynchronizer is closed");
     }
-    await this.#publishTranscription(true);
   }
 }
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
-  BasicTranscriptionForwarder
+  TextAudioSynchronizer,
+  defaultTextSyncOptions
 });
 //# sourceMappingURL=transcription.cjs.map

package/dist/transcription.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/transcription.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame, Room } from '@livekit/rtc-node';\nimport { log } from './log.js';\n\nexport interface TranscriptionForwarder {\n  start(): void;\n  pushAudio(frame: AudioFrame): void;\n  pushText(text: string): void;\n  markTextComplete(): void;\n  markAudioComplete(): void;\n  close(interrupt: boolean): Promise<void>;\n  currentCharacterIndex: number;\n  text: string;\n}\n\nexport class BasicTranscriptionForwarder implements TranscriptionForwarder {\n  #room: Room;\n  #participantIdentity: string;\n  #trackSid: string;\n  #currentText: string = '';\n  #totalAudioDuration: number = 0;\n  #currentPlayoutTime: number = 0;\n  #DEFAULT_CHARS_PER_SECOND = 16;\n  #charsPerSecond: number = this.#DEFAULT_CHARS_PER_SECOND;\n  #messageId: string;\n  #isRunning: boolean = false;\n  #logger = log();\n  currentCharacterIndex: number = 0;\n\n  constructor(room: Room, participantIdentity: string, trackSid: string, messageId: string) {\n    this.#room = room;\n    this.#participantIdentity = participantIdentity;\n    this.#trackSid = trackSid;\n    this.#messageId = messageId;\n  }\n\n  get text(): string {\n    return this.#currentText;\n  }\n\n  start(): void {\n    if (!this.#isRunning) {\n      this.#isRunning = true;\n      this.#startPublishingLoop().catch((error) => {\n        this.#logger.error('Error in publishing loop:', error);\n        this.#isRunning = false;\n      });\n    }\n  }\n\n  pushAudio(frame: AudioFrame): void {\n    this.#totalAudioDuration += frame.samplesPerChannel / frame.sampleRate;\n  }\n\n  pushText(text: string): void {\n    this.#currentText += text;\n  }\n\n  #textIsComplete: boolean = false;\n  #audioIsComplete: boolean = false;\n\n  markTextComplete(): void {\n    this.#textIsComplete = true;\n    this.#adjustTimingIfBothFinished();\n  }\n\n  markAudioComplete(): void {\n    this.#audioIsComplete = true;\n    this.#adjustTimingIfBothFinished();\n  }\n\n  #adjustTimingIfBothFinished(): void {\n    if (this.#textIsComplete && this.#audioIsComplete) {\n      const actualDuration = this.#totalAudioDuration;\n      if (actualDuration > 0 && this.#currentText.length > 0) {\n        this.#charsPerSecond = this.#currentText.length / actualDuration;\n      }\n    }\n  }\n\n  #computeSleepInterval(): number {\n    return Math.min(Math.max(1 / this.#charsPerSecond, 0.0625), 0.5);\n  }\n\n  async #startPublishingLoop(): Promise<void> {\n    this.#isRunning = true;\n    let sleepInterval = this.#computeSleepInterval();\n    let isComplete = false;\n    while (this.#isRunning && !isComplete) {\n      this.#currentPlayoutTime += sleepInterval;\n      this.currentCharacterIndex = Math.floor(this.#currentPlayoutTime * this.#charsPerSecond);\n      isComplete = this.#textIsComplete && this.currentCharacterIndex >= this.#currentText.length;\n      await this.#publishTranscription(false);\n      if (this.#isRunning && !isComplete) {\n        sleepInterval = this.#computeSleepInterval();\n        await new Promise((resolve) => setTimeout(resolve, sleepInterval * 1000));\n      }\n    }\n\n    if (this.#isRunning) {\n      this.close(false);\n    }\n  }\n\n  async #publishTranscription(final: boolean): Promise<void> {\n    const textToPublish = this.#currentText.slice(0, this.currentCharacterIndex);\n    await this.#room.localParticipant?.publishTranscription({\n      participantIdentity: this.#participantIdentity,\n      trackSid: this.#trackSid,\n      segments: [\n        {\n          text: textToPublish,\n          final: final,\n          id: this.#messageId,\n          startTime: BigInt(0),\n          endTime: BigInt(0),\n          language: '',\n        },\n      ],\n    });\n  }\n\n  async close(interrupt: boolean): Promise<void> {\n    this.#isRunning = false;\n\n    // Publish whatever we had as final\n    if (!interrupt) {\n      this.currentCharacterIndex = this.#currentText.length;\n    }\n    await this.#publishTranscription(true);\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAIA,iBAAoB;AAab,MAAM,4BAA8D;AAAA,EACzE;AAAA,EACA;AAAA,EACA;AAAA,EACA,eAAuB;AAAA,EACvB,sBAA8B;AAAA,EAC9B,sBAA8B;AAAA,EAC9B,4BAA4B;AAAA,EAC5B,kBAA0B,KAAK;AAAA,EAC/B;AAAA,EACA,aAAsB;AAAA,EACtB,cAAU,gBAAI;AAAA,EACd,wBAAgC;AAAA,EAEhC,YAAY,MAAY,qBAA6B,UAAkB,WAAmB;AACxF,SAAK,QAAQ;AACb,SAAK,uBAAuB;AAC5B,SAAK,YAAY;AACjB,SAAK,aAAa;AAAA,EACpB;AAAA,EAEA,IAAI,OAAe;AACjB,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,QAAc;AACZ,QAAI,CAAC,KAAK,YAAY;AACpB,WAAK,aAAa;AAClB,WAAK,qBAAqB,EAAE,MAAM,CAAC,UAAU;AAC3C,aAAK,QAAQ,MAAM,6BAA6B,KAAK;AACrD,aAAK,aAAa;AAAA,MACpB,CAAC;AAAA,IACH;AAAA,EACF;AAAA,EAEA,UAAU,OAAyB;AACjC,SAAK,uBAAuB,MAAM,oBAAoB,MAAM;AAAA,EAC9D;AAAA,EAEA,SAAS,MAAoB;AAC3B,SAAK,gBAAgB;AAAA,EACvB;AAAA,EAEA,kBAA2B;AAAA,EAC3B,mBAA4B;AAAA,EAE5B,mBAAyB;AACvB,SAAK,kBAAkB;AACvB,SAAK,4BAA4B;AAAA,EACnC;AAAA,EAEA,oBAA0B;AACxB,SAAK,mBAAmB;AACxB,SAAK,4BAA4B;AAAA,EACnC;AAAA,EAEA,8BAAoC;AAClC,QAAI,KAAK,mBAAmB,KAAK,kBAAkB;AACjD,YAAM,iBAAiB,KAAK;AAC5B,UAAI,iBAAiB,KAAK,KAAK,aAAa,SAAS,GAAG;AACtD,aAAK,kBAAkB,KAAK,aAAa,SAAS;AAAA,MACpD;AAAA,IACF;AAAA,EACF;AAAA,EAEA,wBAAgC;AAC9B,WAAO,KAAK,IAAI,KAAK,IAAI,IAAI,KAAK,iBAAiB,MAAM,GAAG,GAAG;AAAA,EACjE;AAAA,EAEA,MAAM,uBAAsC;AAC1C,SAAK,aAAa;AAClB,QAAI,gBAAgB,KAAK,sBAAsB;AAC/C,QAAI,aAAa;AACjB,WAAO,KAAK,cAAc,CAAC,YAAY;AACrC,WAAK,uBAAuB;AAC5B,WAAK,wBAAwB,KAAK,MAAM,KAAK,sBAAsB,KAAK,eAAe;AACvF,mBAAa,KAAK,mBAAmB,KAAK,yBAAyB,KAAK,aAAa;AACrF,YAAM,KAAK,sBAAsB,KAAK;AACtC,UAAI,KAAK,cAAc,CAAC,YAAY;AAClC,wBAAgB,KAAK,sBAAsB;AAC3C,cAAM,IAAI,QAAQ,CAAC,YAAY,WAAW,SAAS,gBAAgB,GAAI,CAAC;AAAA,MAC1E;AAAA,IACF;AAEA,QAAI,KAAK,YAAY;AACnB,WAAK,MAAM,KAAK;AAAA,IAClB;AAAA,EACF;AAAA,EAEA,MAAM,sBAAsB,OAA+B;AA1G7D;AA2GI,UAAM,gBAAgB,KAAK,aAAa,MAAM,GAAG,KAAK,qBAAqB;AAC3E,YAAM,UAAK,MAAM,qBAAX,mBAA6B,qBAAqB;AAAA,MACtD,qBAAqB,KAAK;AAAA,MAC1B,UAAU,KAAK;AAAA,MACf,UAAU;AAAA,QACR;AAAA,UACE,MAAM;AAAA,UACN;AAAA,UACA,IAAI,KAAK;AAAA,UACT,WAAW,OAAO,CAAC;AAAA,UACnB,SAAS,OAAO,CAAC;AAAA,UACjB,UAAU;AAAA,QACZ;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAM,MAAM,WAAmC;AAC7C,SAAK,aAAa;AAGlB,QAAI,CAAC,WAAW;AACd,WAAK,wBAAwB,KAAK,aAAa;AAAA,IACjD;AACA,UAAM,KAAK,sBAAsB,IAAI;AAAA,EACvC;AACF;","names":[]}
+{"version":3,"sources":["../src/transcription.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { TranscriptionSegment } from '@livekit/protocol';\nimport { AudioFrame } from '@livekit/rtc-node';\nimport type { TypedEventEmitter as TypedEmitter } from '@livekit/typed-emitter';\nimport { randomUUID } from 'node:crypto';\nimport { EventEmitter } from 'node:events';\nimport { basic } from './tokenize/index.js';\nimport type { SentenceStream, SentenceTokenizer } from './tokenize/tokenizer.js';\nimport { AsyncIterableQueue, Future } from './utils.js';\n\n// standard speech rate in hyphens/ms\nconst STANDARD_SPEECH_RATE = 3830;\n\nexport interface TextSyncOptions {\n  language: string;\n  speed: number;\n  newSentenceDelay: number;\n  sentenceTokenizer: SentenceTokenizer;\n  hyphenateWord: (word: string) => string[];\n  splitWords: (words: string) => [string, number, number][];\n}\n\nexport const defaultTextSyncOptions: TextSyncOptions = {\n  language: '',\n  speed: 1,\n  newSentenceDelay: 400,\n  sentenceTokenizer: new basic.SentenceTokenizer(),\n  hyphenateWord: basic.hyphenateWord,\n  splitWords: basic.splitWords,\n};\n\ninterface AudioData {\n  pushedDuration: number;\n  done: boolean;\n}\n\ninterface TextData {\n  sentenceStream: SentenceStream;\n  pushedText: string;\n  done: boolean;\n  forwardedHyphens: number;\n  forwardedSentences: number;\n}\n\ntype SyncCallbacks = {\n  textUpdated: (text: TranscriptionSegment) => void;\n};\n\nexport class TextAudioSynchronizer extends (EventEmitter as new () => TypedEmitter<SyncCallbacks>) {\n  #opts: TextSyncOptions;\n  #speed: number;\n\n  #closed = false;\n  #interrupted = false;\n  #closeFut = new Future();\n\n  #playingSegIndex = -1;\n  #finishedSegIndex = -1;\n\n  #textQChanged = new AsyncIterableQueue<number>();\n  #textQ: (TextData | undefined)[] = [];\n  #audioQChanged = new AsyncIterableQueue<number>();\n  #audioQ: (AudioData | undefined)[] = [];\n\n  #playedText = '';\n  #task?: Promise<void>;\n\n  #audioData?: AudioData;\n  #textData?: TextData;\n\n  constructor(opts: TextSyncOptions) {\n    super();\n\n    this.#opts = opts;\n    this.#speed = opts.speed * STANDARD_SPEECH_RATE;\n  }\n\n  pushAudio(frame: AudioFrame) {\n    this.#checkNotClosed();\n    if (!this.#audioData) {\n      this.#audioData = { pushedDuration: 0, done: false };\n      this.#audioQ.push(this.#audioData);\n      this.#audioQChanged.put(1);\n    }\n    this.#audioData.pushedDuration += frame.samplesPerChannel / frame.sampleRate;\n  }\n\n  pushText(text: string) {\n    this.#checkNotClosed();\n    if (!this.#textData) {\n      this.#textData = {\n        sentenceStream: this.#opts.sentenceTokenizer.stream(),\n        pushedText: '',\n        done: false,\n        forwardedHyphens: 0,\n        forwardedSentences: 0,\n      };\n      this.#textQ.push(this.#textData);\n      this.#textQChanged.put(1);\n    }\n\n    this.#textData.pushedText += text;\n    this.#textData.sentenceStream.pushText(text);\n  }\n\n  markAudioSegmentEnd() {\n    this.#checkNotClosed();\n\n    if (!this.#audioData) {\n      // create empty audio data if none exists\n      this.pushAudio(new AudioFrame(new Int16Array(), 24000, 1, 0));\n    }\n\n    this.#audioData!.done = true;\n    this.#audioData = undefined;\n  }\n\n  markTextSegmentEnd() {\n    this.#checkNotClosed();\n\n    if (!this.#textData) {\n      this.pushText('');\n    }\n\n    this.#textData!.done = true;\n    this.#textData?.sentenceStream.flush();\n    this.#textData?.sentenceStream.close();\n    this.#textData = undefined;\n  }\n\n  segmentPlayoutStarted() {\n    this.#checkNotClosed();\n    this.#playingSegIndex++;\n\n    if (!this.#task) {\n      this.#task = this.#mainLoop();\n    }\n  }\n\n  segmentPlayoutFinished() {\n    this.#checkNotClosed();\n    this.#finishedSegIndex++;\n  }\n\n  get playedText(): string {\n    return this.#playedText;\n  }\n\n  async close(interrupt: boolean) {\n    if (this.#closed) {\n      return;\n    }\n    this.#closed = true;\n    this.#interrupted = interrupt;\n    this.#closeFut.resolve();\n\n    for (const textData of this.#textQ) {\n      textData?.sentenceStream.close();\n    }\n\n    this.#textQ.push(undefined);\n    this.#audioQ.push(undefined);\n    this.#textQChanged.put(1);\n    this.#audioQChanged.put(1);\n\n    await this.#task;\n  }\n\n  async #mainLoop() {\n    let segIndex = 0;\n    let qDone = false;\n\n    while (!qDone) {\n      await this.#textQChanged.next();\n      await this.#audioQChanged.next();\n\n      while (this.#textQ.length && this.#audioQ.length) {\n        const textData = this.#textQ.pop();\n        const audioData = this.#audioQ.pop();\n\n        if (!(textData && audioData)) {\n          qDone = true;\n          break;\n        }\n\n        // wait for segment to start playing\n        while (!this.#closed) {\n          if (this.#playingSegIndex >= segIndex) break;\n          await this.#sleepIfNotClosed(125);\n        }\n\n        const sentenceStream = textData.sentenceStream;\n        const forwardStartTime = Date.now();\n\n        for await (const ev of sentenceStream) {\n          await this.#syncSentence(segIndex, forwardStartTime, textData, audioData, ev.token);\n        }\n\n        segIndex++;\n      }\n    }\n  }\n\n  async #syncSentence(\n    segIndex: number,\n    segStartTime: number,\n    textData: TextData,\n    audioData: AudioData,\n    sentence: string,\n  ) {\n    let realSpeed: number | undefined;\n    if (audioData.pushedDuration > 0 && audioData.done) {\n      realSpeed = this.#calcHyphens(textData.pushedText).length / audioData.pushedDuration;\n    }\n\n    const segId = 'SG_' + randomUUID();\n    const words = this.#opts.splitWords(sentence);\n    const processedWords: string[] = [];\n\n    const ogText = this.#playedText;\n    // eslint-disable-next-line @typescript-eslint/no-unused-vars\n    for (const [word, _, end] of words) {\n      if (segIndex <= this.#finishedSegIndex) break;\n      if (this.#interrupted) return;\n\n      const wordHyphens = this.#opts.hyphenateWord(word).length;\n      processedWords.push(word);\n\n      const elapsed = Date.now() - segStartTime;\n      const text = sentence.slice(0, end); // TODO: rstrip punctuations\n\n      let speed = this.#speed;\n      let delay: number;\n      if (realSpeed) {\n        speed = realSpeed;\n        const estimatedPausesMs = textData.forwardedSentences * this.#opts.newSentenceDelay;\n        const hyphPauses = estimatedPausesMs * speed;\n        const targetHyphens = Math.round(speed * elapsed);\n        const dt = targetHyphens - textData.forwardedHyphens - hyphPauses;\n        const toWaitHyphens = Math.max(0, wordHyphens - dt);\n        delay = toWaitHyphens / speed;\n      } else {\n        delay = wordHyphens / speed;\n      }\n\n      const firstDelay = Math.min(delay / 2, 2 / speed);\n      await this.#sleepIfNotClosed(firstDelay * 1000000);\n\n      this.emit(\n        'textUpdated',\n        new TranscriptionSegment({\n          id: segId,\n          text: text,\n          startTime: BigInt(0),\n          endTime: BigInt(0),\n          final: false,\n          language: this.#opts.language,\n        }),\n      );\n\n      this.#playedText = `${ogText} ${text}`;\n      await this.#sleepIfNotClosed((delay - firstDelay) * 1000000);\n      textData.forwardedHyphens += wordHyphens;\n    }\n\n    this.emit(\n      'textUpdated',\n      new TranscriptionSegment({\n        id: segId,\n        text: sentence,\n        startTime: BigInt(0),\n        endTime: BigInt(0),\n        final: true,\n        language: this.#opts.language,\n      }),\n    );\n\n    this.#playedText = `${ogText} ${sentence}`;\n\n    await this.#sleepIfNotClosed(this.#opts.newSentenceDelay);\n    textData.forwardedSentences++;\n  }\n\n  async #sleepIfNotClosed(delay: number) {\n    await Promise.race([\n      this.#closeFut.await,\n      new Promise((resolve) => setTimeout(resolve, delay)),\n    ]);\n  }\n\n  #calcHyphens(text: string): string[] {\n    const hyphens: string[] = [];\n    const words = this.#opts.splitWords(text);\n    for (const word of words) {\n      const n = this.#opts.hyphenateWord(word[0]);\n      hyphens.push(...n);\n    }\n    return hyphens;\n  }\n\n  #checkNotClosed() {\n    if (this.#closed) {\n      throw new Error('TextAudioSynchronizer is closed');\n    }\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,sBAAqC;AACrC,sBAA2B;AAE3B,yBAA2B;AAC3B,yBAA6B;AAC7B,sBAAsB;AAEtB,mBAA2C;AAG3C,MAAM,uBAAuB;AAWtB,MAAM,yBAA0C;AAAA,EACrD,UAAU;AAAA,EACV,OAAO;AAAA,EACP,kBAAkB;AAAA,EAClB,mBAAmB,IAAI,sBAAM,kBAAkB;AAAA,EAC/C,eAAe,sBAAM;AAAA,EACrB,YAAY,sBAAM;AACpB;AAmBO,MAAM,8BAA+B,gCAAuD;AAAA,EACjG;AAAA,EACA;AAAA,EAEA,UAAU;AAAA,EACV,eAAe;AAAA,EACf,YAAY,IAAI,oBAAO;AAAA,EAEvB,mBAAmB;AAAA,EACnB,oBAAoB;AAAA,EAEpB,gBAAgB,IAAI,gCAA2B;AAAA,EAC/C,SAAmC,CAAC;AAAA,EACpC,iBAAiB,IAAI,gCAA2B;AAAA,EAChD,UAAqC,CAAC;AAAA,EAEtC,cAAc;AAAA,EACd;AAAA,EAEA;AAAA,EACA;AAAA,EAEA,YAAY,MAAuB;AACjC,UAAM;AAEN,SAAK,QAAQ;AACb,SAAK,SAAS,KAAK,QAAQ;AAAA,EAC7B;AAAA,EAEA,UAAU,OAAmB;AAC3B,SAAK,gBAAgB;AACrB,QAAI,CAAC,KAAK,YAAY;AACpB,WAAK,aAAa,EAAE,gBAAgB,GAAG,MAAM,MAAM;AACnD,WAAK,QAAQ,KAAK,KAAK,UAAU;AACjC,WAAK,eAAe,IAAI,CAAC;AAAA,IAC3B;AACA,SAAK,WAAW,kBAAkB,MAAM,oBAAoB,MAAM;AAAA,EACpE;AAAA,EAEA,SAAS,MAAc;AACrB,SAAK,gBAAgB;AACrB,QAAI,CAAC,KAAK,WAAW;AACnB,WAAK,YAAY;AAAA,QACf,gBAAgB,KAAK,MAAM,kBAAkB,OAAO;AAAA,QACpD,YAAY;AAAA,QACZ,MAAM;AAAA,QACN,kBAAkB;AAAA,QAClB,oBAAoB;AAAA,MACtB;AACA,WAAK,OAAO,KAAK,KAAK,SAAS;AAC/B,WAAK,cAAc,IAAI,CAAC;AAAA,IAC1B;AAEA,SAAK,UAAU,cAAc;AAC7B,SAAK,UAAU,eAAe,SAAS,IAAI;AAAA,EAC7C;AAAA,EAEA,sBAAsB;AACpB,SAAK,gBAAgB;AAErB,QAAI,CAAC,KAAK,YAAY;AAEpB,WAAK,UAAU,IAAI,2BAAW,IAAI,WAAW,GAAG,MAAO,GAAG,CAAC,CAAC;AAAA,IAC9D;AAEA,SAAK,WAAY,OAAO;AACxB,SAAK,aAAa;AAAA,EACpB;AAAA,EAEA,qBAAqB;AAvHvB;AAwHI,SAAK,gBAAgB;AAErB,QAAI,CAAC,KAAK,WAAW;AACnB,WAAK,SAAS,EAAE;AAAA,IAClB;AAEA,SAAK,UAAW,OAAO;AACvB,eAAK,cAAL,mBAAgB,eAAe;AAC/B,eAAK,cAAL,mBAAgB,eAAe;AAC/B,SAAK,YAAY;AAAA,EACnB;AAAA,EAEA,wBAAwB;AACtB,SAAK,gBAAgB;AACrB,SAAK;AAEL,QAAI,CAAC,KAAK,OAAO;AACf,WAAK,QAAQ,KAAK,UAAU;AAAA,IAC9B;AAAA,EACF;AAAA,EAEA,yBAAyB;AACvB,SAAK,gBAAgB;AACrB,SAAK;AAAA,EACP;AAAA,EAEA,IAAI,aAAqB;AACvB,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,MAAM,MAAM,WAAoB;AAC9B,QAAI,KAAK,SAAS;AAChB;AAAA,IACF;AACA,SAAK,UAAU;AACf,SAAK,eAAe;AACpB,SAAK,UAAU,QAAQ;AAEvB,eAAW,YAAY,KAAK,QAAQ;AAClC,2CAAU,eAAe;AAAA,IAC3B;AAEA,SAAK,OAAO,KAAK,MAAS;AAC1B,SAAK,QAAQ,KAAK,MAAS;AAC3B,SAAK,cAAc,IAAI,CAAC;AACxB,SAAK,eAAe,IAAI,CAAC;AAEzB,UAAM,KAAK;AAAA,EACb;AAAA,EAEA,MAAM,YAAY;AAChB,QAAI,WAAW;AACf,QAAI,QAAQ;AAEZ,WAAO,CAAC,OAAO;AACb,YAAM,KAAK,cAAc,KAAK;AAC9B,YAAM,KAAK,eAAe,KAAK;AAE/B,aAAO,KAAK,OAAO,UAAU,KAAK,QAAQ,QAAQ;AAChD,cAAM,WAAW,KAAK,OAAO,IAAI;AACjC,cAAM,YAAY,KAAK,QAAQ,IAAI;AAEnC,YAAI,EAAE,YAAY,YAAY;AAC5B,kBAAQ;AACR;AAAA,QACF;AAGA,eAAO,CAAC,KAAK,SAAS;AACpB,cAAI,KAAK,oBAAoB,SAAU;AACvC,gBAAM,KAAK,kBAAkB,GAAG;AAAA,QAClC;AAEA,cAAM,iBAAiB,SAAS;AAChC,cAAM,mBAAmB,KAAK,IAAI;AAElC,yBAAiB,MAAM,gBAAgB;AACrC,gBAAM,KAAK,cAAc,UAAU,kBAAkB,UAAU,WAAW,GAAG,KAAK;AAAA,QACpF;AAEA;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAM,cACJ,UACA,cACA,UACA,WACA,UACA;AACA,QAAI;AACJ,QAAI,UAAU,iBAAiB,KAAK,UAAU,MAAM;AAClD,kBAAY,KAAK,aAAa,SAAS,UAAU,EAAE,SAAS,UAAU;AAAA,IACxE;AAEA,UAAM,QAAQ,YAAQ,+BAAW;AACjC,UAAM,QAAQ,KAAK,MAAM,WAAW,QAAQ;AAC5C,UAAM,iBAA2B,CAAC;AAElC,UAAM,SAAS,KAAK;AAEpB,eAAW,CAAC,MAAM,GAAG,GAAG,KAAK,OAAO;AAClC,UAAI,YAAY,KAAK,kBAAmB;AACxC,UAAI,KAAK,aAAc;AAEvB,YAAM,cAAc,KAAK,MAAM,cAAc,IAAI,EAAE;AACnD,qBAAe,KAAK,IAAI;AAExB,YAAM,UAAU,KAAK,IAAI,IAAI;AAC7B,YAAM,OAAO,SAAS,MAAM,GAAG,GAAG;AAElC,UAAI,QAAQ,KAAK;AACjB,UAAI;AACJ,UAAI,WAAW;AACb,gBAAQ;AACR,cAAM,oBAAoB,SAAS,qBAAqB,KAAK,MAAM;AACnE,cAAM,aAAa,oBAAoB;AACvC,cAAM,gBAAgB,KAAK,MAAM,QAAQ,OAAO;AAChD,cAAM,KAAK,gBAAgB,SAAS,mBAAmB;AACvD,cAAM,gBAAgB,KAAK,IAAI,GAAG,cAAc,EAAE;AAClD,gBAAQ,gBAAgB;AAAA,MAC1B,OAAO;AACL,gBAAQ,cAAc;AAAA,MACxB;AAEA,YAAM,aAAa,KAAK,IAAI,QAAQ,GAAG,IAAI,KAAK;AAChD,YAAM,KAAK,kBAAkB,aAAa,GAAO;AAEjD,WAAK;AAAA,QACH;AAAA,QACA,IAAI,qCAAqB;AAAA,UACvB,IAAI;AAAA,UACJ;AAAA,UACA,WAAW,OAAO,CAAC;AAAA,UACnB,SAAS,OAAO,CAAC;AAAA,UACjB,OAAO;AAAA,UACP,UAAU,KAAK,MAAM;AAAA,QACvB,CAAC;AAAA,MACH;AAEA,WAAK,cAAc,GAAG,MAAM,IAAI,IAAI;AACpC,YAAM,KAAK,mBAAmB,QAAQ,cAAc,GAAO;AAC3D,eAAS,oBAAoB;AAAA,IAC/B;AAEA,SAAK;AAAA,MACH;AAAA,MACA,IAAI,qCAAqB;AAAA,QACvB,IAAI;AAAA,QACJ,MAAM;AAAA,QACN,WAAW,OAAO,CAAC;AAAA,QACnB,SAAS,OAAO,CAAC;AAAA,QACjB,OAAO;AAAA,QACP,UAAU,KAAK,MAAM;AAAA,MACvB,CAAC;AAAA,IACH;AAEA,SAAK,cAAc,GAAG,MAAM,IAAI,QAAQ;AAExC,UAAM,KAAK,kBAAkB,KAAK,MAAM,gBAAgB;AACxD,aAAS;AAAA,EACX;AAAA,EAEA,MAAM,kBAAkB,OAAe;AACrC,UAAM,QAAQ,KAAK;AAAA,MACjB,KAAK,UAAU;AAAA,MACf,IAAI,QAAQ,CAAC,YAAY,WAAW,SAAS,KAAK,CAAC;AAAA,IACrD,CAAC;AAAA,EACH;AAAA,EAEA,aAAa,MAAwB;AACnC,UAAM,UAAoB,CAAC;AAC3B,UAAM,QAAQ,KAAK,MAAM,WAAW,IAAI;AACxC,eAAW,QAAQ,OAAO;AACxB,YAAM,IAAI,KAAK,MAAM,cAAc,KAAK,CAAC,CAAC;AAC1C,cAAQ,KAAK,GAAG,CAAC;AAAA,IACnB;AACA,WAAO;AAAA,EACT;AAAA,EAEA,kBAAkB;AAChB,QAAI,KAAK,SAAS;AAChB,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAAA,EACF;AACF;","names":[]}