ssml-builder-js 2.9.0 → 2.10.0

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。
@@ -378,8 +380,8 @@ 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 +541,7 @@ 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 +629,8 @@ 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.
+
 `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`.
@@ -822,8 +826,8 @@ 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/dist/{chunk-JNJVTEL6.mjs → chunk-BYIZQL2W.mjs}

@@ -1364,6 +1364,7 @@ var SSML_ATTRS = {
   MSTTS_XMLNS: "xmlns:mstts",
   NAME: "name",
   VOICE: "voice",
+  SPEAKER: "speaker",
   EFFECT: "effect",
   RATE: "rate",
   PITCH: "pitch",
@@ -1394,6 +1395,11 @@ 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 +1495,7 @@ 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 +1504,18 @@ 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 +2088,9 @@ 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 +2107,26 @@ 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: {
@@ -2431,11 +2461,14 @@ function tokenizeElements(source) {
     }
     const selfClosing = /\/\s*>$/.test(raw);
     const parent = openElements[openElements.length - 1];
+    const childElementIndex = parent?.childElementCount;
+    if (parent) parent.childElementCount += 1;
     const parentVoiceName = [...openElements].reverse().find((element) => element.voiceName)?.voiceName;
     const tokenName = nameMatch[1];
     const tokenVoiceName = tokenName.toLowerCase() === "voice" ? attributes.get("name") : tokenName.toLowerCase() === "mstts:turn" ? attributes.get("voice") ?? parentVoiceName : parentVoiceName;
     tokens.push({
       attributes,
+      childElementIndex,
       end,
       name: tokenName,
       parentName: parent?.name,
@@ -2445,6 +2478,7 @@ function tokenizeElements(source) {
     });
     if (!selfClosing) {
       openElements.push({
+        childElementCount: 0,
         name: tokenName,
         voiceName: tokenVoiceName
       });
@@ -2483,6 +2517,12 @@ function isValidAzureAudioDuration(value) {
   if (!clock) return false;
   return Number(clock[1]) > 0 || Number(clock[2]) > 0 || Number(clock[3]) > 0 || Number(clock[4] ?? 0) > 0;
 }
+function isValidAzureBackgroundAudioDuration(value) {
+  const match = /^(\d+(?:\.\d+)?)(ms|s)?$/i.exec(value.trim());
+  if (!match) return false;
+  const milliseconds = Number(match[1]) * (match[2]?.toLowerCase() === "s" ? 1e3 : 1);
+  return Number.isFinite(milliseconds) && milliseconds >= 0 && milliseconds <= 1e4;
+}
 function attr(token, name) {
   return token.attributes.get(name.toLowerCase());
 }
@@ -2612,9 +2652,16 @@ 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 "${parsed.origin}" is not allowed.`);
-  else if (!options.allowExternalAudio)
+  else if (!isAllowedOrigin && !options.allowExternalAudio)
     addDiagnostic(
       diagnostics,
       source,
@@ -2744,28 +2791,40 @@ function validateElement(token, source, diagnostics, voiceName, options, voiceCa
     validateAudioSource(token, source, diagnostics, options, "audio");
   }
   if (name === "mstts:turn") {
-    if (!attr(token, "voice")?.trim())
-      addDiagnostic(diagnostics, source, token.start, '<mstts:turn> requires a non-empty "voice" attribute.');
+    if (!attr(token, "voice")?.trim() && !attr(token, "speaker")?.trim())
+      addDiagnostic(
+        diagnostics,
+        source,
+        token.start,
+        '<mstts:turn> requires a non-empty "voice" or "speaker" attribute.'
+      );
     if (token.parentName?.toLowerCase() !== "mstts:dialog")
       addDiagnostic(diagnostics, source, token.start, "<mstts:turn> is only allowed directly inside <mstts:dialog>.");
   }
   if (name === "mstts:backgroundaudio") {
     validateAudioSource(token, source, diagnostics, options, "mstts:backgroundaudio");
     const volume = attr(token, "volume");
-    if (volume && !/^(silent|x-soft|soft|medium|loud|x-loud|[+-]?\d+(?:\.\d+)?(?:dB|%))$/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 +2856,16 @@ 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,
@@ -4439,4 +4508,4 @@ export {
   findSsmlHoverTarget,
   formatSsmlHover
 };
-//# sourceMappingURL=chunk-JNJVTEL6.mjs.map
+//# sourceMappingURL=chunk-BYIZQL2W.mjs.map