ssml-builder-js 2.9.0 → 2.11.0

package/dist/elements.mjs

@@ -36,7 +36,7 @@ import {
   resolveExpressAsStyles,
   updateEditableText,
   validateAzureSsml
-} from "./chunk-JNJVTEL6.mjs";
+} from "./chunk-FWWMWI2C.mjs";
 import "./chunk-6S5ODO6A.mjs";
 
 // packages/ssml-editor-react/src/ssmlInsertions.ts

package/dist/index.d.mts

@@ -44,4 +44,29 @@ declare class AzureTtsClient {
 
 declare function synthesizeSpeech(ssml: string, config: TtsConfig): Promise<ArrayBuffer>;
 
-export { AzureTtsClient, type AzureTtsClientOptions, AzureTtsError, type AzureTtsLogger, AzureTtsSdkError, type TtsConfig, synthesizeSpeech };
+interface FetchAzureVoiceCatalogOptions {
+    apiKey: string;
+    region: string | string[];
+}
+interface AzureVoiceCatalogVoice {
+    name: string;
+    locale: string;
+    secondaryLocales?: readonly string[];
+    styles?: readonly string[];
+    regions: readonly string[];
+    status?: "ga" | "preview" | "deprecated";
+}
+interface FetchedAzureVoiceCatalogMetadata {
+    voiceCount: number;
+    generatedAt: string;
+    apiVersion: string;
+    regions: readonly string[];
+}
+interface AzureVoiceCatalog {
+    voices: readonly AzureVoiceCatalogVoice[];
+    metadata: FetchedAzureVoiceCatalogMetadata;
+}
+/** Fetches and deduplicates the current Azure Speech voice catalog for one or more regions. */
+declare function fetchAzureVoiceCatalog(options: FetchAzureVoiceCatalogOptions): Promise<AzureVoiceCatalog>;
+
+export { AzureTtsClient, type AzureTtsClientOptions, AzureTtsError, type AzureTtsLogger, AzureTtsSdkError, type AzureVoiceCatalog, type AzureVoiceCatalogVoice, type FetchAzureVoiceCatalogOptions, type FetchedAzureVoiceCatalogMetadata, type TtsConfig, fetchAzureVoiceCatalog, synthesizeSpeech };

package/dist/index.d.ts

@@ -44,4 +44,29 @@ declare class AzureTtsClient {
 
 declare function synthesizeSpeech(ssml: string, config: TtsConfig): Promise<ArrayBuffer>;
 
-export { AzureTtsClient, type AzureTtsClientOptions, AzureTtsError, type AzureTtsLogger, AzureTtsSdkError, type TtsConfig, synthesizeSpeech };
+interface FetchAzureVoiceCatalogOptions {
+    apiKey: string;
+    region: string | string[];
+}
+interface AzureVoiceCatalogVoice {
+    name: string;
+    locale: string;
+    secondaryLocales?: readonly string[];
+    styles?: readonly string[];
+    regions: readonly string[];
+    status?: "ga" | "preview" | "deprecated";
+}
+interface FetchedAzureVoiceCatalogMetadata {
+    voiceCount: number;
+    generatedAt: string;
+    apiVersion: string;
+    regions: readonly string[];
+}
+interface AzureVoiceCatalog {
+    voices: readonly AzureVoiceCatalogVoice[];
+    metadata: FetchedAzureVoiceCatalogMetadata;
+}
+/** Fetches and deduplicates the current Azure Speech voice catalog for one or more regions. */
+declare function fetchAzureVoiceCatalog(options: FetchAzureVoiceCatalogOptions): Promise<AzureVoiceCatalog>;
+
+export { AzureTtsClient, type AzureTtsClientOptions, AzureTtsError, type AzureTtsLogger, AzureTtsSdkError, type AzureVoiceCatalog, type AzureVoiceCatalogVoice, type FetchAzureVoiceCatalogOptions, type FetchedAzureVoiceCatalogMetadata, type TtsConfig, fetchAzureVoiceCatalog, synthesizeSpeech };

package/dist/index.js

@@ -45,6 +45,7 @@ __export(src_exports, {
   buildSsml: () => buildSsml,
   extractSsmlText: () => extractSsmlText,
   extractSsmlTranslatableText: () => extractSsmlTranslatableText,
+  fetchAzureVoiceCatalog: () => fetchAzureVoiceCatalog,
   fromPlainTextToSsml: () => fromPlainTextToSsml,
   getAzureVoiceCatalogMetadata: () => getAzureVoiceCatalogMetadata,
   getBuiltInVoiceCatalogMetadata: () => getBuiltInVoiceCatalogMetadata,
@@ -108,6 +109,7 @@ var SSML_ATTRS = {
   MSTTS_XMLNS: "xmlns:mstts",
   NAME: "name",
   VOICE: "voice",
+  SPEAKER: "speaker",
   EFFECT: "effect",
   RATE: "rate",
   PITCH: "pitch",
@@ -138,6 +140,11 @@ 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",
@@ -235,6 +242,7 @@ 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);
@@ -243,9 +251,18 @@ 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:
@@ -820,7 +837,9 @@ 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: {
@@ -837,14 +856,26 @@ 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: {
@@ -1220,6 +1251,7 @@ 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);
@@ -1227,6 +1259,18 @@ 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)]));
 }
@@ -1458,6 +1502,9 @@ 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" },
@@ -1554,6 +1601,18 @@ 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,
@@ -1613,11 +1672,14 @@ function tokenizeElements(source) {
     }
     const selfClosing = /\/\s*>$/.test(raw);
     const parent = openElements[openElements.length - 1];
+    const childElementIndex = parent?.childElementCount;
+    if (parent) parent.childElementCount += 1;
     const parentVoiceName = [...openElements].reverse().find((element) => element.voiceName)?.voiceName;
     const tokenName = nameMatch[1];
     const tokenVoiceName = tokenName.toLowerCase() === "voice" ? attributes.get("name") : tokenName.toLowerCase() === "mstts:turn" ? attributes.get("voice") ?? parentVoiceName : parentVoiceName;
     tokens.push({
       attributes,
+      childElementIndex,
       end,
       name: tokenName,
       parentName: parent?.name,
@@ -1627,6 +1689,7 @@ function tokenizeElements(source) {
     });
     if (!selfClosing) {
       openElements.push({
+        childElementCount: 0,
         name: tokenName,
         voiceName: tokenVoiceName
       });
@@ -1665,6 +1728,12 @@ function isValidAzureAudioDuration(value) {
   if (!clock) return false;
   return Number(clock[1]) > 0 || Number(clock[2]) > 0 || Number(clock[3]) > 0 || Number(clock[4] ?? 0) > 0;
 }
+function isValidAzureBackgroundAudioDuration(value) {
+  const match = /^(\d+)$/.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());
 }
@@ -1804,9 +1873,16 @@ 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 "${parsed.origin}" is not allowed.`);
-  else if (!options.allowExternalAudio)
+  else if (!isAllowedOrigin && !options.allowExternalAudio)
     addDiagnostic(
       diagnostics,
       source,
@@ -1816,6 +1892,25 @@ function validateAudioSource(token, source, diagnostics, options, elementName2)
 }
 function validateElement(token, source, diagnostics, voiceName, options, voiceCatalog) {
   const name = token.name.toLowerCase();
+  const tagStatus = featureStatusForTag(token.name, options);
+  if (tagStatus === "preview")
+    addDiagnostic(
+      diagnostics,
+      source,
+      token.start,
+      `<${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,
+      `<${token.name}> is deprecated by Azure Speech; migrate to a supported alternative.`,
+      "info",
+      "azure-deprecated-tag"
+    );
   if (name === "voice" && !attr(token, "name")?.trim())
     addDiagnostic(diagnostics, source, token.start, '<voice> requires a non-empty "name" attribute.');
   if (name === "break") {
@@ -1936,28 +2031,40 @@ function validateElement(token, source, diagnostics, voiceName, options, voiceCa
     validateAudioSource(token, source, diagnostics, options, "audio");
   }
   if (name === "mstts:turn") {
-    if (!attr(token, "voice")?.trim())
-      addDiagnostic(diagnostics, source, token.start, '<mstts:turn> requires a non-empty "voice" attribute.');
+    if (!attr(token, "voice")?.trim() && !attr(token, "speaker")?.trim())
+      addDiagnostic(
+        diagnostics,
+        source,
+        token.start,
+        '<mstts:turn> requires a non-empty "voice" or "speaker" attribute.'
+      );
     if (token.parentName?.toLowerCase() !== "mstts:dialog")
       addDiagnostic(diagnostics, source, token.start, "<mstts:turn> is only allowed directly inside <mstts:dialog>.");
   }
   if (name === "mstts:backgroundaudio") {
     validateAudioSource(token, source, diagnostics, options, "mstts:backgroundaudio");
     const volume = attr(token, "volume");
-    if (volume && !/^(silent|x-soft|soft|medium|loud|x-loud|[+-]?\d+(?:\.\d+)?(?:dB|%))$/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 ${attribute}> must be between 0 and 10000 milliseconds, for example "500ms" or "10s".`
         );
     }
+    if (token.parentName?.toLowerCase() !== "speak" || token.childElementIndex !== 0)
+      addDiagnostic(
+        diagnostics,
+        source,
+        token.start,
+        "<mstts:backgroundaudio> must be the first element directly under <speak>."
+      );
     if (!token.selfClosing)
       addDiagnostic(diagnostics, source, token.start, "<mstts:backgroundaudio> must be self-closing.");
   }
@@ -1989,6 +2096,16 @@ function validateAzureSsml(ssml, options = {}) {
   const tokens = tokenizeElements(ssml);
   const speak = tokens.find((token) => token.name.toLowerCase() === "speak");
   const voices = tokens.filter((token) => token.name.toLowerCase() === "voice");
+  const backgroundAudioTokens = tokens.filter((token) => token.name.toLowerCase() === "mstts:backgroundaudio");
+  for (const [index, token] of backgroundAudioTokens.entries()) {
+    if (index > 0)
+      addDiagnostic(
+        diagnostics,
+        ssml,
+        token.start,
+        "An SSML document can contain at most one <mstts:backgroundaudio> element."
+      );
+  }
   if (!speak || voices.length === 0)
     addDiagnostic(
       diagnostics,
@@ -2005,6 +2122,24 @@ function validateAzureSsml(ssml, options = {}) {
     const name = attr(token, "name")?.trim();
     const language = attr(token, "xml:lang")?.trim() || (speak ? attr(speak, "xml:lang")?.trim() : void 0);
     const definition = name ? voiceCatalog.get(name.toLowerCase()) : void 0;
+    if (name && definition?.status === "preview")
+      addDiagnostic(
+        diagnostics,
+        ssml,
+        token.start,
+        `Voice "${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 "${name}" is deprecated by Azure Speech; migrate to a supported voice.`,
+        "info",
+        "azure-deprecated-voice"
+      );
     if (name && !definition && policySeverity)
       addDiagnostic(
         diagnostics,
@@ -2242,6 +2377,84 @@ var AzureTtsClient = class {
   }
 };
 _options = new WeakMap();
+
+// packages/azure-tts-client/src/voiceCatalog.ts
+var AZURE_VOICE_API_VERSION = "2025-10-01";
+function stringValue(value) {
+  return typeof value === "string" && value.trim() ? value.trim() : void 0;
+}
+function stringList(value) {
+  if (!Array.isArray(value)) return [];
+  return [...new Set(value.map(stringValue).filter((item) => item !== void 0))];
+}
+function normalizeStatus(value) {
+  const status = stringValue(value)?.toLowerCase();
+  if (status === "preview" || status === "deprecated" || status === "ga") return status;
+  return void 0;
+}
+function normalizeRegions(region) {
+  const regions = Array.isArray(region) ? region : [region];
+  const result = [...new Set(regions.map((item) => item.trim()).filter(Boolean))];
+  if (result.length === 0) throw new TypeError("At least one Azure Speech region is required.");
+  return result;
+}
+async function fetchRegionVoices(region, apiKey) {
+  const endpoint = `https://${encodeURIComponent(region)}.tts.speech.microsoft.com/cognitiveservices/voices/list`;
+  const response = await fetch(endpoint, {
+    headers: {
+      Accept: "application/json",
+      "Ocp-Apim-Subscription-Key": apiKey
+    }
+  });
+  if (!response.ok) {
+    throw new Error(`Azure List Voices API request failed for region "${region}" with HTTP ${response.status}.`);
+  }
+  const payload = await response.json();
+  if (!Array.isArray(payload)) throw new Error(`Azure List Voices API returned an invalid response for "${region}".`);
+  return payload.filter((item) => Boolean(item && typeof item === "object"));
+}
+async function fetchAzureVoiceCatalog(options) {
+  if (!options || typeof options.apiKey !== "string" || !options.apiKey.trim())
+    throw new TypeError("An Azure Speech API key is required.");
+  const regions = normalizeRegions(options.region);
+  const payloads = await Promise.all(regions.map((region) => fetchRegionVoices(region, options.apiKey)));
+  const voices = /* @__PURE__ */ new Map();
+  for (let regionIndex = 0; regionIndex < payloads.length; regionIndex += 1) {
+    const region = regions[regionIndex];
+    for (const record of payloads[regionIndex]) {
+      const name = stringValue(record.ShortName) ?? stringValue(record.Name);
+      const locale = stringValue(record.Locale);
+      if (!name || !locale) continue;
+      const key = name.toLowerCase();
+      const existing = voices.get(key);
+      const secondaryLocales = stringList(record.SecondaryLocaleList);
+      const styles = stringList(record.StyleList);
+      const status = normalizeStatus(record.Status);
+      const merged = {
+        name: existing?.name ?? name,
+        locale: existing?.locale ?? locale,
+        regions: [.../* @__PURE__ */ new Set([...existing?.regions ?? [], region])]
+      };
+      const mergedSecondaryLocales = [.../* @__PURE__ */ new Set([...existing?.secondaryLocales ?? [], ...secondaryLocales])];
+      if (mergedSecondaryLocales.length > 0) merged.secondaryLocales = mergedSecondaryLocales;
+      const mergedStyles = [.../* @__PURE__ */ new Set([...existing?.styles ?? [], ...styles])];
+      if (mergedStyles.length > 0) merged.styles = mergedStyles;
+      if (status) merged.status = status;
+      else if (existing?.status) merged.status = existing.status;
+      voices.set(key, merged);
+    }
+  }
+  const sortedVoices = [...voices.values()].sort((first, second) => first.name.localeCompare(second.name));
+  return {
+    voices: sortedVoices,
+    metadata: {
+      voiceCount: sortedVoices.length,
+      generatedAt: (/* @__PURE__ */ new Date()).toISOString(),
+      apiVersion: AZURE_VOICE_API_VERSION,
+      regions
+    }
+  };
+}
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   AzureTtsClient,
@@ -2252,6 +2465,7 @@ _options = new WeakMap();
   buildSsml,
   extractSsmlText,
   extractSsmlTranslatableText,
+  fetchAzureVoiceCatalog,
   fromPlainTextToSsml,
   getAzureVoiceCatalogMetadata,
   getBuiltInVoiceCatalogMetadata,