ssml-builder-js 2.9.0 → 2.11.0

package/README.md

@@ -91,7 +91,7 @@ Pages Router と App Router のどちらでも、Monaco を `next/dynamic` の `
 "use client";
 import dynamic from "next/dynamic";
 const SsmlEditor = dynamic(
-  () => import("@ssml-builder-js/ssml-editor-react").then(({ SsmlEditor }) => SsmlEditor),
+  () => import("ssml-builder-js/react").then(({ SsmlEditor }) => SsmlEditor),
   { ssr: false },
 );
 
@@ -179,6 +179,8 @@ SSML の検証は、構文、Azure 固有の静的な意味、実サービスの
 
 `voice`、`prosody`、`break`、`express-as`、`say-as`、`phoneme`、`audio`、`lang`、`mark` に加えて、`mstts:dialog`、`mstts:turn`、`mstts:backgroundaudio`、`mstts:ttsembedding`、`mstts:embedding`、`mstts:voiceconversion` を型付きで表現できます。`type: "custom"` と `name` を指定すれば、未定義の XML 要素や追加属性も扱えます。`mstts:` 要素を含むドキュメントを生成すると、必要な Azure Speech 名前空間が自動的に追加されます。
 
+`mstts:turn` は `voice` またはマルチトーカー用の `speaker` を指定でき、`mstts:ttsembedding` は `speakerProfileId`、`mstts:embedding` は `id`、`mstts:voiceconversion` は `url` と `profile` を専用プロパティで指定できます。`mstts:backgroundaudio` は `<speak>` 直下の先頭要素として 1 文書に 1 つだけ配置し、`fadein`/`fadeout` は単位なしの生ミリ秒（0〜10000）で指定します。
+
 `validateAzureSsml` は Azure Speech へ送信する前に実行する事前静的チェックです。返却される診断の `source` はパッケージ側の静的解析結果であることを示し、Azure Speech サービス側でのランタイム生成結果や実際の合成可否を表すものではありません。
 
 翻訳などで本文だけを置き換える場合は、`mapSsmlTextNodes` に変換関数を渡します。変換関数には直近の親タグと祖先タグの `path` が渡され、戻り値は `string` または `Promise<string>` を指定できます。`validateAzureSsml` は音声、属性値、音声スタイル、文字数、`audio` URL/オリジンを検証します。
@@ -206,7 +208,7 @@ const diagnostics = validateAzureSsml(ssml, {
 });
 ```
 
-音声カタログは `AzureVoiceDefinition` の `name`、`locale`、`secondaryLocales`、`styles`、`supportedTags`、`unsupportedTags`、`models` で表現できます。`voiceDefinitions`（または `voiceCatalog`）を渡すと、組み込み台帳を外部定義で補完・上書きできます。`supportedTags`／`unsupportedTags` による音声別制約違反は `azure-unsupported-tag-for-voice` としてエラーになります。組み込み台帳にない音声は `azure-unknown-voice`（`unknownVoicePolicy` に従う）、登録済み音声への非対応スタイルは `azure-unsupported-style`、ロケール不一致は `azure-locale-mismatch` として区別されます。
+音声カタログは `AzureVoiceDefinition` の `name`、`locale`、`secondaryLocales`、`styles`、`supportedTags`、`unsupportedTags`、`models`、`regions`、`status` で表現できます。`voiceDefinitions`（または `voiceCatalog`）を渡すと、組み込み台帳を外部定義で補完・上書きできます。`supportedTags`／`unsupportedTags` による音声別制約違反は `azure-unsupported-tag-for-voice` としてエラーになります。組み込み台帳にない音声は `azure-unknown-voice`（`unknownVoicePolicy` に従う）、登録済み音声への非対応スタイルは `azure-unsupported-style`、ロケール不一致は `azure-locale-mismatch` として区別されます。
 
 `npm run sync:voices -- --regions eastus,japaneast` は Azure List Voices API をリージョンごとに取得し、重複を除いた TypeScript 音声定義と `azureVoiceCatalog.json`（生成日時、API バージョン、リージョン、収録数）を更新します。認証情報は `AZURE_SPEECH_KEY` と `AZURE_SPEECH_REGION(S)`、または CLI オプションで指定します。
 
@@ -245,6 +247,24 @@ Azure Speech は `<audio>` の URL を取得するため、任意の URL をそ
 
 `editMode="visual"` を指定するかツールバーの **Visual** を選ぶと、XML を直接見ずに構造ツリーとパンくずから親要素を選択し、フォームで本文の変更、rate、pitch、emotion、pause、pronunciation の適用ができます。Azure の配置違反や属性エラーはビジュアル領域にも表示されます。`editMode="code"` で Monaco に戻ります。
 
+Visual Editor と Code Editor の対応要素は次のとおりです。
+
+| 要素 | Visual Editor | Code Editor |
+| --- | --- | --- |
+| `voice`、`prosody`、`break`、`express-as`、`say-as`、`phoneme` | フォーム操作・構造ツリー | 完全対応 |
+| `mstts:dialog` / `mstts:turn` | 話者ターンの追加・音声/話者/本文編集 | 完全対応・補完あり |
+| `mstts:backgroundaudio` | URL、音量、フェードイン/アウト編集 | 完全対応・補完あり |
+| `mstts:ttsembedding`、`mstts:embedding`、`mstts:voiceconversion` | 構造ツリーで保持 | 完全対応 |
+| 未定義の XML 要素 | 構造ツリーで保持 | `custom` として編集 |
+
+Azure Speech のライフサイクル対応状況は次のとおりです。GA 要素は通常の静的検証対象、プレビュー要素・音声は Warning、非推奨要素・音声は Info の診断を返します。`AzureVoiceDefinition.status` と `AzureValidationOptions.tagStatuses` で外部カタログの状態も指定できます。
+
+| 状態 | 診断 | 対象 |
+| --- | --- | --- |
+| GA | なし | 標準 SSML、`mstts:dialog`、`mstts:backgroundaudio` |
+| Preview | Warning | `mstts:voiceconversion`、`previewTags` / `tagStatuses` に指定した要素、`status: "preview"` の音声 |
+| Deprecated | Info | `deprecatedTags` / `tagStatuses`、`status: "deprecated"` の音声 |
+
 `<mstts:express-as>` の `style` 属性補完と標準の感情メニューは、カーソルまたは選択範囲を囲む最内の `<voice name="...">` が対応するスタイルだけを表示します。音声名が未指定の場合は全候補を表示し、登録済みでスタイル非対応の音声や未登録の音声では候補がないことを表示します。`emotionStyles` を指定した場合、感情メニューではその値と登録済み音声の対応スタイルの共通部分を使用します。
 
 ```tsx
@@ -347,6 +367,19 @@ const audio = await client.synthesize(ssml);
 
 帯域幅を抑える場合は `audio-24khz-48kbitrate-mono-mp3` または `audio-16khz-32kbitrate-mono-mp3` を指定できます。
 
+最新の音声一覧は公開 API から取得できます。複数リージョンを指定した場合は音声名で重複排除され、各音声の `regions` と `metadata`（`voiceCount`、`generatedAt`、`apiVersion`、`regions`）が返ります。
+
+```ts
+import { fetchAzureVoiceCatalog } from "ssml-builder-js";
+
+const catalog = await fetchAzureVoiceCatalog({
+  apiKey: process.env.AZURE_SPEECH_KEY!,
+  region: ["eastus", "japaneast"],
+});
+```
+
+CLI では `npx ssml-builder sync-voices --region eastus --output ./azure-voices.json` を実行します。キーは `AZURE_SPEECH_KEY`（または `--key`）、リージョンは `AZURE_SPEECH_REGION(S)`（または `--region(s)`）から指定できます。
+
 ```ts
 const client = new AzureTtsClient({
   subscriptionKey: process.env.AZURE_SPEECH_KEY!,
@@ -378,8 +411,8 @@ Next.js Route Handler ではキーをブラウザへ渡さず、サーバー側
 
 ```ts
 // app/api/synthesize/route.ts
-import { AzureTtsClient, AzureTtsError } from "@ssml-builder-js/azure-tts-client";
-import { validateAzureSsml, validateSsml } from "@ssml-builder-js/ssml-core";
+import { AzureTtsClient, AzureTtsError } from "ssml-builder-js";
+import { validateAzureSsml, validateSsml } from "ssml-builder-js/core";
 
 export const runtime = "nodejs";
 
@@ -539,7 +572,7 @@ For both the Pages Router and App Router, keep Monaco in a client-only module wi
 "use client";
 import dynamic from "next/dynamic";
 const SsmlEditor = dynamic(
-  () => import("@ssml-builder-js/ssml-editor-react").then(({ SsmlEditor }) => SsmlEditor),
+  () => import("ssml-builder-js/react").then(({ SsmlEditor }) => SsmlEditor),
   { ssr: false },
 );
 
@@ -627,6 +660,8 @@ SSML validation separates XML syntax, Azure-specific static semantics, and runti
 
 Typed representations are available for elements such as `voice`, `prosody`, `break`, `express-as`, `say-as`, `phoneme`, `audio`, `lang`, and `mark`, plus `mstts:dialog`, `mstts:turn`, `mstts:backgroundaudio`, `mstts:ttsembedding`, `mstts:embedding`, and `mstts:voiceconversion`. Use `type: "custom"` and `name` to handle undefined XML elements or additional attributes. When a document contains `mstts:` elements, the required Azure Speech namespace is added automatically.
 
+`mstts:turn` accepts either `voice` or the multi-talker `speaker` property. The typed extension properties include `speakerProfileId` for `mstts:ttsembedding`, `id` for `mstts:embedding`, and `url` plus `profile` for `mstts:voiceconversion`. `mstts:backgroundaudio` must be the first direct child element of `<speak>` and may appear only once per document; `fadein` and `fadeout` are raw milliseconds from 0 through 10000.
+
 `validateAzureSsml` is a preflight static check performed by this package before sending SSML to Azure Speech. The `source` on each returned diagnostic identifies package-side static analysis; it is independent of Azure Speech's runtime generation result and does not guarantee that synthesis will succeed.
 
 Use `mapSsmlTextNodes` to replace translatable content without changing tags, attributes, or nesting. The transform receives the immediate parent tag and ancestor `path`, and may return a `string` or a `Promise<string>`. The third argument supports `skipTags` and a `filter` callback; `phoneme`, `say-as`, and `sub` are skipped by default. The callback receives `parentTag`, decoded `parentAttributes`, `ancestorTags`, and `path`.
@@ -650,7 +685,7 @@ const diagnostics = validateAzureSsml(ssml, {
 });
 ```
 
-The catalog is represented by `AzureVoiceDefinition` with `name`, `locale`, optional `secondaryLocales`, `styles`, `supportedTags`, `unsupportedTags`, and `models`. Pass `voiceDefinitions` (or `voiceCatalog`) to supplement or override the built-in catalog with an external definition. A tag that violates `supportedTags` or `unsupportedTags` produces `azure-unsupported-tag-for-voice` with error severity. `customVoiceStyleMap` remains supported for backward compatibility and overrides styles for the named voice. Diagnostics distinguish an unregistered voice (`azure-unknown-voice`, controlled by `unknownVoicePolicy`), an unsupported style on a registered voice (`azure-unsupported-style`), and a locale mismatch (`azure-locale-mismatch`). The `<mstts:audioduration value="10s"/>` element accepts positive `ms` or `s` values and `hh:mm:ss[.fff]` clock values.
+The catalog is represented by `AzureVoiceDefinition` with `name`, `locale`, optional `secondaryLocales`, `styles`, `supportedTags`, `unsupportedTags`, `models`, `regions`, and `status`. Pass `voiceDefinitions` (or `voiceCatalog`) to supplement or override the built-in catalog with an external definition. A tag that violates `supportedTags` or `unsupportedTags` produces `azure-unsupported-tag-for-voice` with error severity. `customVoiceStyleMap` remains supported for backward compatibility and overrides styles for the named voice. Diagnostics distinguish an unregistered voice (`azure-unknown-voice`, controlled by `unknownVoicePolicy`), an unsupported style on a registered voice (`azure-unsupported-style`), and a locale mismatch (`azure-locale-mismatch`). The `<mstts:audioduration value="10s"/>` element accepts positive `ms` or `s` values and `hh:mm:ss[.fff]` clock values.
 
 `npm run sync:voices -- --regions eastus,japaneast` fetches Azure's List Voices API for each region, deduplicates the results, and updates the generated TypeScript definitions plus `azureVoiceCatalog.json` with generation time, API version, regions, and voice count. Provide credentials through `AZURE_SPEECH_KEY` and `AZURE_SPEECH_REGION(S)`, or CLI options.
 
@@ -689,6 +724,24 @@ Azure Speech fetches `<audio>` URLs, so a server that accepts arbitrary user-pro
 
 Set `editMode="visual"`, or choose **Visual** in the toolbar, to edit without viewing XML source. The structured editor provides a structure tree, parent breadcrumb selection, text editing, and form actions for rate, pitch, emotion, pause, and pronunciation. Azure placement and attribute diagnostics are shown in the visual area. Choose `editMode="code"` to return to Monaco.
 
+The Visual Editor and Code Editor support the following elements:
+
+| Element | Visual Editor | Code Editor |
+| --- | --- | --- |
+| `voice`, `prosody`, `break`, `express-as`, `say-as`, `phoneme` | Form actions and structure tree | Full support |
+| `mstts:dialog` / `mstts:turn` | Add turns and edit voice, speaker, and text | Full support with completion |
+| `mstts:backgroundaudio` | Edit URL, volume, fade-in, and fade-out | Full support with completion |
+| `mstts:ttsembedding`, `mstts:embedding`, `mstts:voiceconversion` | Preserved in the structure tree | Full support |
+| Unknown XML elements | Preserved in the structure tree | Editable as `custom` |
+
+Azure lifecycle support is reported by static validation: GA elements produce no lifecycle diagnostic, preview elements and voices produce a Warning, and deprecated elements and voices produce an Info diagnostic. Use `AzureVoiceDefinition.status` and `AzureValidationOptions.tagStatuses` for external catalog metadata.
+
+| Status | Diagnostic | Examples |
+| --- | --- | --- |
+| GA | None | Standard SSML, `mstts:dialog`, `mstts:backgroundaudio` |
+| Preview | Warning | `mstts:voiceconversion`, tags listed in `previewTags` / `tagStatuses`, voices with `status: "preview"` |
+| Deprecated | Info | Tags listed in `deprecatedTags` / `tagStatuses`, voices with `status: "deprecated"` |
+
 Completion for the `<mstts:express-as>` `style` attribute and the built-in Emotion menu only show styles supported by the innermost `<voice name="...">` around the cursor or selection. When no voice name is available, all candidates are shown; a registered voice without supported styles or an explicitly unregistered voice displays an empty-state message. When `emotionStyles` is supplied, the Emotion menu uses its intersection with the registered voice's supported styles.
 
 ```tsx
@@ -791,6 +844,19 @@ const audio = await client.synthesize(ssml);
 
 For lower bandwidth, choose `audio-24khz-48kbitrate-mono-mp3` or `audio-16khz-32kbitrate-mono-mp3`.
 
+Fetch the current voice catalog from Azure with `fetchAzureVoiceCatalog`. Multiple regions are deduplicated by voice name, and each voice includes its available `regions`; the result also contains `metadata` with `voiceCount`, `generatedAt`, `apiVersion`, and `regions`.
+
+```ts
+import { fetchAzureVoiceCatalog } from "ssml-builder-js";
+
+const catalog = await fetchAzureVoiceCatalog({
+  apiKey: process.env.AZURE_SPEECH_KEY!,
+  region: ["eastus", "japaneast"],
+});
+```
+
+The public updater CLI is `npx ssml-builder sync-voices --region eastus --output ./azure-voices.json`. It reads the key from `AZURE_SPEECH_KEY` (or `--key`) and regions from `AZURE_SPEECH_REGION(S)` (or `--region(s)`).
+
 ```ts
 const client = new AzureTtsClient({
   subscriptionKey: process.env.AZURE_SPEECH_KEY!,
@@ -822,8 +888,8 @@ Keep the subscription key on the server by using a Next.js Route Handler (or an
 
 ```ts
 // app/api/synthesize/route.ts
-import { AzureTtsClient, AzureTtsError } from "@ssml-builder-js/azure-tts-client";
-import { validateAzureSsml, validateSsml } from "@ssml-builder-js/ssml-core";
+import { AzureTtsClient, AzureTtsError } from "ssml-builder-js";
+import { validateAzureSsml, validateSsml } from "ssml-builder-js/core";
 
 export const runtime = "nodejs";
 

package/bin/sync-voices.js

@@ -0,0 +1,62 @@
+#!/usr/bin/env node
+
+const { writeFile } = require("node:fs/promises");
+const { resolve } = require("node:path");
+const { fetchAzureVoiceCatalog } = require("../dist/index.js");
+
+function usage() {
+  return [
+    "Usage: ssml-builder sync-voices --region <region[,region...]> --output <file>",
+    "",
+    "Environment variables: AZURE_SPEECH_KEY, AZURE_SPEECH_REGION(S)",
+  ].join("\n");
+}
+
+function parseArgs(args) {
+  const options = {};
+  for (let index = 0; index < args.length; index += 1) {
+    const argument = args[index];
+    if (argument === "--region" || argument === "--regions") {
+      options.regions = (args[++index] || "")
+        .split(",")
+        .map((value) => value.trim())
+        .filter(Boolean);
+    } else if (argument === "--output") {
+      options.output = args[++index];
+    } else if (argument === "--key") {
+      options.apiKey = args[++index];
+    } else if (argument === "--help" || argument === "-h") {
+      console.log(usage());
+      process.exit(0);
+    } else {
+      throw new Error(`Unknown option: ${argument}\n\n${usage()}`);
+    }
+  }
+  return options;
+}
+
+async function main() {
+  const [command, ...args] = process.argv.slice(2);
+  if (command !== "sync-voices") throw new Error(usage());
+  const options = parseArgs(args);
+  const regions =
+    options.regions ||
+    (process.env.AZURE_SPEECH_REGIONS || process.env.AZURE_SPEECH_REGION || "")
+      .split(",")
+      .map((value) => value.trim())
+      .filter(Boolean);
+  const apiKey = options.apiKey || process.env.AZURE_SPEECH_KEY;
+  const output = options.output;
+  if (!apiKey) throw new Error("Set AZURE_SPEECH_KEY or pass --key.");
+  if (regions.length === 0) throw new Error("Set AZURE_SPEECH_REGION(S) or pass --region(s).");
+  if (!output) throw new Error(`Pass --output <file>.\n\n${usage()}`);
+
+  const catalog = await fetchAzureVoiceCatalog({ apiKey, region: regions });
+  await writeFile(resolve(output), `${JSON.stringify(catalog, null, 2)}\n`, "utf8");
+  console.log(`Wrote ${catalog.metadata.voiceCount} Azure voice definitions to ${resolve(output)}.`);
+}
+
+main().catch((error) => {
+  console.error(error instanceof Error ? error.message : String(error));
+  process.exitCode = 1;
+});

package/dist/{chunk-JNJVTEL6.mjs → chunk-FWWMWI2C.mjs}

@@ -542,6 +542,7 @@ var ssmlPresets_exports = {};
 __export(ssmlPresets_exports, {
   AUDIO_DURATION_DESCRIPTIONS: () => AUDIO_DURATION_DESCRIPTIONS,
   AUDIO_DURATION_PRESETS: () => AUDIO_DURATION_PRESETS,
+  BACKGROUND_AUDIO_DURATION_PRESETS: () => BACKGROUND_AUDIO_DURATION_PRESETS,
   BREAK_STRENGTH_PRESETS: () => BREAK_STRENGTH_PRESETS,
   BREAK_TIME_DESCRIPTIONS: () => BREAK_TIME_DESCRIPTIONS,
   BREAK_TIME_PRESETS: () => BREAK_TIME_PRESETS,
@@ -623,6 +624,9 @@ var AZURE_VOICE_STYLE_MAP = {
   ],
   "es-ES-ElviraNeural": [],
   "fil-PH-AngeloNeural": [],
+  "fil-PH-Angelo:DragonHDLatestNeural": [],
+  "fil-PH-BlessicaNeural": [],
+  "fil-PH-Blessica:DragonHDLatestNeural": [],
   "fr-FR-DeniseNeural": ["cheerful", "sad"],
   "fr-FR-HenriNeural": ["cheerful", "sad"],
   "id-ID-GadisNeural": [],
@@ -900,6 +904,7 @@ var SAY_AS_PRESETS = [
 var LANGUAGE_PRESETS = ["ja-JP", "en-US", "de-DE", "fr-FR"];
 var SILENCE_VALUE_PRESETS = ["300ms", "500ms", "1s"];
 var AUDIO_DURATION_PRESETS = ["5s", "10s", "30s"];
+var BACKGROUND_AUDIO_DURATION_PRESETS = ["0", "500", "1000", "3000", "10000"];
 var SILENCE_TYPE_PRESETS = [
   "Leading",
   "Tailing",
@@ -961,8 +966,8 @@ var SSML_ATTRIBUTE_PRESETS = {
   },
   "mstts:backgroundaudio": {
     volume: PROSODY_VOLUME_PRESETS,
-    fadein: AUDIO_DURATION_PRESETS,
-    fadeout: AUDIO_DURATION_PRESETS
+    fadein: BACKGROUND_AUDIO_DURATION_PRESETS,
+    fadeout: BACKGROUND_AUDIO_DURATION_PRESETS
   },
   silence: {
     type: SILENCE_TYPE_PRESETS,
@@ -1364,6 +1369,7 @@ var SSML_ATTRS = {
   MSTTS_XMLNS: "xmlns:mstts",
   NAME: "name",
   VOICE: "voice",
+  SPEAKER: "speaker",
   EFFECT: "effect",
   RATE: "rate",
   PITCH: "pitch",
@@ -1394,6 +1400,11 @@ 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",
@@ -1489,6 +1500,7 @@ 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);
@@ -1497,9 +1509,18 @@ 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:
@@ -2072,7 +2093,9 @@ 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: {
@@ -2089,14 +2112,26 @@ 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: {
@@ -2278,6 +2313,9 @@ 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" },
@@ -2372,6 +2410,18 @@ 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,
@@ -2431,11 +2481,14 @@ function tokenizeElements(source) {
     }
     const selfClosing = /\/\s*>$/.test(raw);
     const parent = openElements[openElements.length - 1];
+    const childElementIndex = parent?.childElementCount;
+    if (parent) parent.childElementCount += 1;
     const parentVoiceName = [...openElements].reverse().find((element) => element.voiceName)?.voiceName;
     const tokenName = nameMatch[1];
     const tokenVoiceName = tokenName.toLowerCase() === "voice" ? attributes.get("name") : tokenName.toLowerCase() === "mstts:turn" ? attributes.get("voice") ?? parentVoiceName : parentVoiceName;
     tokens.push({
       attributes,
+      childElementIndex,
       end,
       name: tokenName,
       parentName: parent?.name,
@@ -2445,6 +2498,7 @@ function tokenizeElements(source) {
     });
     if (!selfClosing) {
       openElements.push({
+        childElementCount: 0,
         name: tokenName,
         voiceName: tokenVoiceName
       });
@@ -2483,6 +2537,12 @@ function isValidAzureAudioDuration(value) {
   if (!clock) return false;
   return Number(clock[1]) > 0 || Number(clock[2]) > 0 || Number(clock[3]) > 0 || Number(clock[4] ?? 0) > 0;
 }
+function isValidAzureBackgroundAudioDuration(value) {
+  const match = /^(\d+)$/.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());
 }
@@ -2612,9 +2672,16 @@ 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 "${parsed.origin}" is not allowed.`);
-  else if (!options.allowExternalAudio)
+  else if (!isAllowedOrigin && !options.allowExternalAudio)
     addDiagnostic(
       diagnostics,
       source,
@@ -2624,6 +2691,25 @@ function validateAudioSource(token, source, diagnostics, options, elementName2)
 }
 function validateElement(token, source, diagnostics, voiceName, options, voiceCatalog) {
   const name = token.name.toLowerCase();
+  const tagStatus = featureStatusForTag(token.name, options);
+  if (tagStatus === "preview")
+    addDiagnostic(
+      diagnostics,
+      source,
+      token.start,
+      `<${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,
+      `<${token.name}> is deprecated by Azure Speech; migrate to a supported alternative.`,
+      "info",
+      "azure-deprecated-tag"
+    );
   if (name === "voice" && !attr(token, "name")?.trim())
     addDiagnostic(diagnostics, source, token.start, '<voice> requires a non-empty "name" attribute.');
   if (name === "break") {
@@ -2744,28 +2830,40 @@ function validateElement(token, source, diagnostics, voiceName, options, voiceCa
     validateAudioSource(token, source, diagnostics, options, "audio");
   }
   if (name === "mstts:turn") {
-    if (!attr(token, "voice")?.trim())
-      addDiagnostic(diagnostics, source, token.start, '<mstts:turn> requires a non-empty "voice" attribute.');
+    if (!attr(token, "voice")?.trim() && !attr(token, "speaker")?.trim())
+      addDiagnostic(
+        diagnostics,
+        source,
+        token.start,
+        '<mstts:turn> requires a non-empty "voice" or "speaker" attribute.'
+      );
     if (token.parentName?.toLowerCase() !== "mstts:dialog")
       addDiagnostic(diagnostics, source, token.start, "<mstts:turn> is only allowed directly inside <mstts:dialog>.");
   }
   if (name === "mstts:backgroundaudio") {
     validateAudioSource(token, source, diagnostics, options, "mstts:backgroundaudio");
     const volume = attr(token, "volume");
-    if (volume && !/^(silent|x-soft|soft|medium|loud|x-loud|[+-]?\d+(?:\.\d+)?(?:dB|%))$/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 ${attribute}> must be between 0 and 10000 milliseconds, for example "500ms" or "10s".`
         );
     }
+    if (token.parentName?.toLowerCase() !== "speak" || token.childElementIndex !== 0)
+      addDiagnostic(
+        diagnostics,
+        source,
+        token.start,
+        "<mstts:backgroundaudio> must be the first element directly under <speak>."
+      );
     if (!token.selfClosing)
       addDiagnostic(diagnostics, source, token.start, "<mstts:backgroundaudio> must be self-closing.");
   }
@@ -2797,6 +2895,16 @@ function validateAzureSsml(ssml, options = {}) {
   const tokens = tokenizeElements(ssml);
   const speak = tokens.find((token) => token.name.toLowerCase() === "speak");
   const voices = tokens.filter((token) => token.name.toLowerCase() === "voice");
+  const backgroundAudioTokens = tokens.filter((token) => token.name.toLowerCase() === "mstts:backgroundaudio");
+  for (const [index, token] of backgroundAudioTokens.entries()) {
+    if (index > 0)
+      addDiagnostic(
+        diagnostics,
+        ssml,
+        token.start,
+        "An SSML document can contain at most one <mstts:backgroundaudio> element."
+      );
+  }
   if (!speak || voices.length === 0)
     addDiagnostic(
       diagnostics,
@@ -2813,6 +2921,24 @@ function validateAzureSsml(ssml, options = {}) {
     const name = attr(token, "name")?.trim();
     const language = attr(token, "xml:lang")?.trim() || (speak ? attr(speak, "xml:lang")?.trim() : void 0);
     const definition = name ? voiceCatalog.get(name.toLowerCase()) : void 0;
+    if (name && definition?.status === "preview")
+      addDiagnostic(
+        diagnostics,
+        ssml,
+        token.start,
+        `Voice "${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,
@@ -3654,7 +3780,7 @@ var SSML_COMPLETION_SNIPPETS = [
   },
   {
     label: "mstts:backgroundaudio",
-    insertText: `<mstts:backgroundaudio src="\${1:https://example.com/audio.mp3}" volume="\${2:-3dB}" />`
+    insertText: `<mstts:backgroundaudio src="\${1:https://example.com/audio.mp3}" volume="\${2:70}" fadein="\${3:1000}" fadeout="\${4:1000}" />`
   },
   {
     label: "mstts:ttsembedding",
@@ -4058,18 +4184,18 @@ var SSML_TAG_DEFINITIONS = [
       },
       {
         name: "volume",
-        description: "The background audio volume, for example `-3dB` or `medium`.",
-        example: "-3dB"
+        description: "The background audio volume from 0 to 100.",
+        example: "70"
       },
       {
         name: "fadein",
-        description: "The fade-in duration, for example `1s`.",
-        example: "1s"
+        description: "The fade-in duration in milliseconds from 0 to 10000.",
+        example: "1000"
       },
       {
         name: "fadeout",
-        description: "The fade-out duration, for example `500ms`.",
-        example: "500ms"
+        description: "The fade-out duration in milliseconds from 0 to 10000.",
+        example: "1000"
       }
     ]
   },
@@ -4439,4 +4565,4 @@ export {
   findSsmlHoverTarget,
   formatSsmlHover
 };
-//# sourceMappingURL=chunk-JNJVTEL6.mjs.map
+//# sourceMappingURL=chunk-FWWMWI2C.mjs.map