ssml-builder-js 2.9.0 → 2.11.0

package/dist/core.js

@@ -94,6 +94,7 @@ var SSML_ATTRS = {
   MSTTS_XMLNS: "xmlns:mstts",
   NAME: "name",
   VOICE: "voice",
+  SPEAKER: "speaker",
   EFFECT: "effect",
   RATE: "rate",
   PITCH: "pitch",
@@ -124,6 +125,11 @@ var SSML_ATTRS = {
   ALIAS: "alias",
   MARK: "mark",
   URI: "uri",
+  ID: "id",
+  MODEL: "model",
+  PROFILE: "profile",
+  URL: "url",
+  SPEAKER_PROFILE_ID: "speakerProfileId",
   TYPE: "type",
   VALUE: "value",
   FADE_IN: "fadein",
@@ -221,6 +227,7 @@ function getAttributes(element) {
       break;
     case SSML_TAGS.MSTTS_TURN:
       addAttribute(attributes, SSML_ATTRS.VOICE, element.voice);
+      addAttribute(attributes, SSML_ATTRS.SPEAKER, element.speaker);
       break;
     case SSML_TAGS.MSTTS_BACKGROUND_AUDIO:
       addAttribute(attributes, SSML_ATTRS.SRC, element.src);
@@ -229,9 +236,18 @@ function getAttributes(element) {
       addAttribute(attributes, SSML_ATTRS.FADE_OUT, element.fadeOut ?? element.fadeout);
       break;
     case SSML_TAGS.MSTTS_DIALOG:
+      break;
     case SSML_TAGS.MSTTS_TTS_EMBEDDING:
+      addAttribute(attributes, SSML_ATTRS.SPEAKER_PROFILE_ID, element.speakerProfileId);
+      break;
     case SSML_TAGS.MSTTS_EMBEDDING:
+      addAttribute(attributes, SSML_ATTRS.ID, element.id);
+      addAttribute(attributes, SSML_ATTRS.SPEAKER_PROFILE_ID, element.speakerProfileId);
+      break;
     case SSML_TAGS.MSTTS_VOICE_CONVERSION:
+      addAttribute(attributes, SSML_ATTRS.URL, element.url);
+      addAttribute(attributes, SSML_ATTRS.PROFILE, element.profile);
+      addAttribute(attributes, SSML_ATTRS.SPEAKER_PROFILE_ID, element.speakerProfileId);
       break;
     case SSML_TAGS.PARAGRAPH:
     case SSML_TAGS.SENTENCE:
@@ -806,7 +822,9 @@ function convertElement(node) {
     case SSML_TAGS.MSTTS_TURN: {
       const element = { type: SSML_TAGS.MSTTS_TURN };
       const voice = readAttribute(attributes, SSML_ATTRS.VOICE);
+      const speaker = readAttribute(attributes, SSML_ATTRS.SPEAKER);
       if (voice !== void 0) element.voice = voice;
+      if (speaker !== void 0) element.speaker = speaker;
       return finishElement(element, node, attributes);
     }
     case SSML_TAGS.MSTTS_BACKGROUND_AUDIO: {
@@ -823,14 +841,26 @@ function convertElement(node) {
     }
     case SSML_TAGS.MSTTS_TTS_EMBEDDING: {
       const element = { type: SSML_TAGS.MSTTS_TTS_EMBEDDING };
+      const speakerProfileId = readAttribute(attributes, SSML_ATTRS.SPEAKER_PROFILE_ID);
+      if (speakerProfileId !== void 0) element.speakerProfileId = speakerProfileId;
       return finishElement(element, node, attributes);
     }
     case SSML_TAGS.MSTTS_EMBEDDING: {
       const element = { type: SSML_TAGS.MSTTS_EMBEDDING };
+      const id = readAttribute(attributes, SSML_ATTRS.ID);
+      const speakerProfileId = readAttribute(attributes, SSML_ATTRS.SPEAKER_PROFILE_ID);
+      if (id !== void 0) element.id = id;
+      if (speakerProfileId !== void 0) element.speakerProfileId = speakerProfileId;
       return finishElement(element, node, attributes);
     }
     case SSML_TAGS.MSTTS_VOICE_CONVERSION: {
       const element = { type: SSML_TAGS.MSTTS_VOICE_CONVERSION };
+      const url = readAttribute(attributes, SSML_ATTRS.URL);
+      const profile = readAttribute(attributes, SSML_ATTRS.PROFILE);
+      const speakerProfileId = readAttribute(attributes, SSML_ATTRS.SPEAKER_PROFILE_ID);
+      if (url !== void 0) element.url = url;
+      if (profile !== void 0) element.profile = profile;
+      if (speakerProfileId !== void 0) element.speakerProfileId = speakerProfileId;
       return finishElement(element, node, attributes);
     }
     default: {
@@ -1206,6 +1236,7 @@ function elementAttributes(element) {
       break;
     case "mstts:turn":
       addAttribute2(attributes, "voice", element.voice);
+      addAttribute2(attributes, "speaker", element.speaker);
       break;
     case "mstts:backgroundaudio":
       addAttribute2(attributes, "src", element.src);
@@ -1213,6 +1244,18 @@ function elementAttributes(element) {
       addAttribute2(attributes, "fadein", element.fadeIn ?? element.fadein);
       addAttribute2(attributes, "fadeout", element.fadeOut ?? element.fadeout);
       break;
+    case "mstts:ttsembedding":
+      addAttribute2(attributes, "speakerProfileId", element.speakerProfileId);
+      break;
+    case "mstts:embedding":
+      addAttribute2(attributes, "id", element.id);
+      addAttribute2(attributes, "speakerProfileId", element.speakerProfileId);
+      break;
+    case "mstts:voiceconversion":
+      addAttribute2(attributes, "url", element.url);
+      addAttribute2(attributes, "profile", element.profile);
+      addAttribute2(attributes, "speakerProfileId", element.speakerProfileId);
+      break;
   }
   return Object.fromEntries(Object.entries(attributes).map(([name, value]) => [name, String(value)]));
 }
@@ -1444,6 +1487,9 @@ var AZURE_VOICE_DEFINITIONS = [
   },
   { name: "es-ES-ElviraNeural", locale: "es-ES" },
   { name: "fil-PH-AngeloNeural", locale: "fil-PH" },
+  { name: "fil-PH-Angelo:DragonHDLatestNeural", locale: "fil-PH" },
+  { name: "fil-PH-BlessicaNeural", locale: "fil-PH" },
+  { name: "fil-PH-Blessica:DragonHDLatestNeural", locale: "fil-PH" },
   { name: "fr-FR-DeniseNeural", locale: "fr-FR", styles: ["cheerful", "sad"] },
   { name: "fr-FR-HenriNeural", locale: "fr-FR", styles: ["cheerful", "sad"] },
   { name: "id-ID-GadisNeural", locale: "id-ID" },
@@ -1540,6 +1586,18 @@ var ALLOWED_SILENCE_TYPES = /* @__PURE__ */ new Set([
   "Enumerationcomma"
 ]);
 var ALLOWED_VISEME_TYPES = /* @__PURE__ */ new Set(["redlips_front", "FacialExpression"]);
+var DEFAULT_PREVIEW_TAGS = /* @__PURE__ */ new Set(["mstts:voiceconversion"]);
+function featureStatusForTag(name, options) {
+  const tagName = canonicalTagName(name);
+  const configured = Object.entries(options.tagStatuses ?? {}).find(
+    ([candidate]) => canonicalTagName(candidate) === tagName
+  )?.[1];
+  if (configured) return configured;
+  if ((options.previewTags ?? [...DEFAULT_PREVIEW_TAGS]).some((candidate) => canonicalTagName(candidate) === tagName))
+    return "preview";
+  if ((options.deprecatedTags ?? []).some((candidate) => canonicalTagName(candidate) === tagName)) return "deprecated";
+  return void 0;
+}
 function decodeAttribute(value) {
   return value.replace(
     /&(?:amp|apos|gt|lt|quot);/gi,
@@ -1599,11 +1657,14 @@ function tokenizeElements(source) {
     }
     const selfClosing = /\/\s*>$/.test(raw);
     const parent = openElements[openElements.length - 1];
+    const childElementIndex = parent?.childElementCount;
+    if (parent) parent.childElementCount += 1;
     const parentVoiceName = [...openElements].reverse().find((element) => element.voiceName)?.voiceName;
     const tokenName = nameMatch[1];
     const tokenVoiceName = tokenName.toLowerCase() === "voice" ? attributes.get("name") : tokenName.toLowerCase() === "mstts:turn" ? attributes.get("voice") ?? parentVoiceName : parentVoiceName;
     tokens.push({
       attributes,
+      childElementIndex,
       end,
       name: tokenName,
       parentName: parent?.name,
@@ -1613,6 +1674,7 @@ function tokenizeElements(source) {
     });
     if (!selfClosing) {
       openElements.push({
+        childElementCount: 0,
         name: tokenName,
         voiceName: tokenVoiceName
       });
@@ -1651,6 +1713,12 @@ function isValidAzureAudioDuration(value) {
   if (!clock) return false;
   return Number(clock[1]) > 0 || Number(clock[2]) > 0 || Number(clock[3]) > 0 || Number(clock[4] ?? 0) > 0;
 }
+function isValidAzureBackgroundAudioDuration(value) {
+  const match = /^(\d+)$/.exec(value.trim());
+  if (!match) return false;
+  const milliseconds = Number(match[1]);
+  return Number.isFinite(milliseconds) && milliseconds >= 0 && milliseconds <= 1e4;
+}
 function attr(token, name) {
   return token.attributes.get(name.toLowerCase());
 }
@@ -1790,9 +1858,16 @@ function validateAudioSource(token, source, diagnostics, options, elementName2)
   }
   if (parsed.protocol !== "https:" && !(options.allowHttpAudio && parsed.protocol === "http:"))
     addDiagnostic(diagnostics, source, token.start, `<${elementName2} src> must use HTTPS.`);
-  if (options.allowedAudioOrigins && !options.allowedAudioOrigins.includes(parsed.origin))
+  const isAllowedOrigin = options.allowedAudioOrigins?.some((allowedOrigin) => {
+    try {
+      return new URL(allowedOrigin).origin === parsed.origin;
+    } catch {
+      return allowedOrigin === parsed.origin;
+    }
+  }) ?? false;
+  if (options.allowedAudioOrigins && !isAllowedOrigin)
     addDiagnostic(diagnostics, source, token.start, `<${elementName2} src> origin "${parsed.origin}" is not allowed.`);
-  else if (!options.allowExternalAudio)
+  else if (!isAllowedOrigin && !options.allowExternalAudio)
     addDiagnostic(
       diagnostics,
       source,
@@ -1802,6 +1877,25 @@ function validateAudioSource(token, source, diagnostics, options, elementName2)
 }
 function validateElement(token, source, diagnostics, voiceName, options, voiceCatalog) {
   const name = token.name.toLowerCase();
+  const tagStatus = featureStatusForTag(token.name, options);
+  if (tagStatus === "preview")
+    addDiagnostic(
+      diagnostics,
+      source,
+      token.start,
+      `<${token.name}> is an Azure Speech preview feature and may change or require preview access.`,
+      "warning",
+      "azure-preview-tag"
+    );
+  if (tagStatus === "deprecated")
+    addDiagnostic(
+      diagnostics,
+      source,
+      token.start,
+      `<${token.name}> is deprecated by Azure Speech; migrate to a supported alternative.`,
+      "info",
+      "azure-deprecated-tag"
+    );
   if (name === "voice" && !attr(token, "name")?.trim())
     addDiagnostic(diagnostics, source, token.start, '<voice> requires a non-empty "name" attribute.');
   if (name === "break") {
@@ -1922,28 +2016,40 @@ function validateElement(token, source, diagnostics, voiceName, options, voiceCa
     validateAudioSource(token, source, diagnostics, options, "audio");
   }
   if (name === "mstts:turn") {
-    if (!attr(token, "voice")?.trim())
-      addDiagnostic(diagnostics, source, token.start, '<mstts:turn> requires a non-empty "voice" attribute.');
+    if (!attr(token, "voice")?.trim() && !attr(token, "speaker")?.trim())
+      addDiagnostic(
+        diagnostics,
+        source,
+        token.start,
+        '<mstts:turn> requires a non-empty "voice" or "speaker" attribute.'
+      );
     if (token.parentName?.toLowerCase() !== "mstts:dialog")
       addDiagnostic(diagnostics, source, token.start, "<mstts:turn> is only allowed directly inside <mstts:dialog>.");
   }
   if (name === "mstts:backgroundaudio") {
     validateAudioSource(token, source, diagnostics, options, "mstts:backgroundaudio");
     const volume = attr(token, "volume");
-    if (volume && !/^(silent|x-soft|soft|medium|loud|x-loud|[+-]?\d+(?:\.\d+)?(?:dB|%))$/i.test(volume.trim()))
+    if (volume !== void 0 && (!/^\d+(?:\.\d+)?$/.test(volume.trim()) || Number(volume) > 100))
       addDiagnostic(diagnostics, source, token.start, `Unsupported <mstts:backgroundaudio volume> value "${volume}".`);
     for (const [attribute, value] of [
       ["fadein", attr(token, "fadein")],
       ["fadeout", attr(token, "fadeout")]
     ]) {
-      if (value && !isValidAzureAudioDuration(value))
+      if (value !== void 0 && !isValidAzureBackgroundAudioDuration(value))
         addDiagnostic(
           diagnostics,
           source,
           token.start,
-          `<mstts:backgroundaudio ${attribute}> must be a positive duration such as "500ms" or "10s".`
+          `<mstts:backgroundaudio ${attribute}> must be between 0 and 10000 milliseconds, for example "500ms" or "10s".`
         );
     }
+    if (token.parentName?.toLowerCase() !== "speak" || token.childElementIndex !== 0)
+      addDiagnostic(
+        diagnostics,
+        source,
+        token.start,
+        "<mstts:backgroundaudio> must be the first element directly under <speak>."
+      );
     if (!token.selfClosing)
       addDiagnostic(diagnostics, source, token.start, "<mstts:backgroundaudio> must be self-closing.");
   }
@@ -1975,6 +2081,16 @@ function validateAzureSsml(ssml, options = {}) {
   const tokens = tokenizeElements(ssml);
   const speak = tokens.find((token) => token.name.toLowerCase() === "speak");
   const voices = tokens.filter((token) => token.name.toLowerCase() === "voice");
+  const backgroundAudioTokens = tokens.filter((token) => token.name.toLowerCase() === "mstts:backgroundaudio");
+  for (const [index, token] of backgroundAudioTokens.entries()) {
+    if (index > 0)
+      addDiagnostic(
+        diagnostics,
+        ssml,
+        token.start,
+        "An SSML document can contain at most one <mstts:backgroundaudio> element."
+      );
+  }
   if (!speak || voices.length === 0)
     addDiagnostic(
       diagnostics,
@@ -1991,6 +2107,24 @@ function validateAzureSsml(ssml, options = {}) {
     const name = attr(token, "name")?.trim();
     const language = attr(token, "xml:lang")?.trim() || (speak ? attr(speak, "xml:lang")?.trim() : void 0);
     const definition = name ? voiceCatalog.get(name.toLowerCase()) : void 0;
+    if (name && definition?.status === "preview")
+      addDiagnostic(
+        diagnostics,
+        ssml,
+        token.start,
+        `Voice "${name}" is an Azure Speech preview voice and may change or require preview access.`,
+        "warning",
+        "azure-preview-voice"
+      );
+    if (name && definition?.status === "deprecated")
+      addDiagnostic(
+        diagnostics,
+        ssml,
+        token.start,
+        `Voice "${name}" is deprecated by Azure Speech; migrate to a supported voice.`,
+        "info",
+        "azure-deprecated-voice"
+      );
     if (name && !definition && policySeverity)
       addDiagnostic(
         diagnostics,