@unboundcx/sdk 4.13.97 → 4.13.98

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@unboundcx/sdk",
-  "version": "4.13.97",
+  "version": "4.13.98",
   "description": "Official JavaScript SDK for the Unbound API - A comprehensive toolkit for integrating with Unbound's communication, AI, and data management services",
   "main": "index.js",
   "type": "module",

package/services/ai/tts.js

@@ -0,0 +1,155 @@
+import { internalRequest } from '../../base.js';
+
+/**
+ * Text-to-speech. `sdk.ai.tts.*`.
+ *
+ * @see app1-api src/services/ai/routes/tts.js
+ */
+export class TextToSpeechService {
+  constructor(sdk) {
+    this.sdk = sdk;
+  }
+
+  /**
+   * Render text to speech and get back a storage id/url (buffered — waits
+   * for the whole file). For low-latency playback (e.g. a voice bot) use
+   * `stream()` instead.
+   * @param {Object} params
+   * @param {string} params.text
+   * @param {string} [params.voice]
+   * @param {string} [params.languageCode]
+   * @param {string} [params.ssmlGender]
+   * @param {string} [params.audioEncoding]
+   * @param {number} [params.speakingRate]
+   * @param {number} [params.pitch]
+   * @param {number} [params.volumeGainDb]
+   * @param {string[]} [params.effectsProfileIds]
+   * @param {boolean} [params.createAccessKey]
+   * @returns {Promise<Object>} `{ id, storageId, url? }`
+   */
+  async create({
+    text,
+    voice,
+    languageCode,
+    ssmlGender,
+    audioEncoding,
+    speakingRate,
+    pitch,
+    volumeGainDb,
+    effectsProfileIds,
+    createAccessKey,
+  }) {
+    this.sdk.validateParams(
+      {
+        text,
+        voice,
+        languageCode,
+        ssmlGender,
+        audioEncoding,
+        speakingRate,
+        pitch,
+        volumeGainDb,
+        effectsProfileIds,
+        createAccessKey,
+      },
+      {
+        text: { type: 'string', required: true },
+        voice: { type: 'string', required: false },
+        languageCode: { type: 'string', required: false },
+        ssmlGender: { type: 'string', required: false },
+        audioEncoding: { type: 'string', required: false },
+        speakingRate: { type: 'number', required: false },
+        pitch: { type: 'number', required: false },
+        volumeGainDb: { type: 'number', required: false },
+        effectsProfileIds: { type: 'array', required: false },
+        createAccessKey: { type: 'boolean', required: false },
+      },
+    );
+
+    const ttsData = { text };
+    if (voice) ttsData.voice = voice;
+    if (languageCode) ttsData.languageCode = languageCode;
+    if (ssmlGender) ttsData.ssmlGender = ssmlGender;
+    if (audioEncoding) ttsData.audioEncoding = audioEncoding;
+    if (speakingRate) ttsData.speakingRate = speakingRate;
+    if (pitch) ttsData.pitch = pitch;
+    if (volumeGainDb) ttsData.volumeGainDb = volumeGainDb;
+    if (effectsProfileIds) ttsData.effectsProfileIds = effectsProfileIds;
+    if (createAccessKey) ttsData.createAccessKey = createAccessKey;
+
+    const params = {
+      body: ttsData,
+    };
+
+    const result = await internalRequest(this.sdk, '/ai/tts', 'POST', params);
+    return result;
+  }
+
+  /**
+   * List available TTS voices
+   * @returns {Promise<Object>} { voices: Array, count: number, supportedEncodings: Array, supportedLanguages: Array }
+   */
+  async list() {
+    const result = await internalRequest(this.sdk, '/ai/tts', 'GET');
+    return result;
+  }
+
+  /**
+   * Stream TTS audio as it's generated — first bytes can arrive well before
+   * the whole utterance finishes rendering (Groq voices: TTFB ~0.2s).
+   * Always renders wav; shares its cache with `create()` when
+   * audioEncoding:'wav' is used there.
+   *
+   * Node: wrap the result in `Readable.fromWeb(result.body)` to get a
+   * normal Node stream (e.g. to pipe into ffmpeg). Browser: `result.body`
+   * is already a web ReadableStream you can read directly or hand to
+   * `new Response(result.body)` / a `<audio>` element via a Blob.
+   *
+   * Response headers of note: `x-tts-cache` (`hit`|`miss`) and `x-tts-id`
+   * (set on a cache hit, when the id is already known).
+   *
+   * @param {Object} params
+   * @param {string} params.text
+   * @param {string} [params.voice]
+   * @param {string} [params.languageCode]
+   * @returns {Promise<{body: ReadableStream, headers: Headers, status: number}>}
+   * @example
+   * const result = await sdk.ai.tts.stream({ text: 'Hi there', voice: 'hannah' });
+   * const nodeStream = Readable.fromWeb(result.body);
+   * nodeStream.pipe(ffmpegProcess.stdin);
+   */
+  async stream({ text, voice, languageCode }) {
+    this.sdk.validateParams(
+      { text, voice, languageCode },
+      {
+        text: { type: 'string', required: true },
+        voice: { type: 'string', required: false },
+        languageCode: { type: 'string', required: false },
+      },
+    );
+
+    const ttsData = { text };
+    if (voice) ttsData.voice = voice;
+    if (languageCode) ttsData.languageCode = languageCode;
+
+    const params = {
+      body: ttsData,
+      returnRawResponse: true,
+    };
+
+    // forceFetch: true — NATS transport can't carry a streamed body.
+    const response = await internalRequest(
+      this.sdk,
+      '/ai/tts/stream',
+      'POST',
+      params,
+      true,
+    );
+
+    return {
+      body: response.body,
+      headers: response.headers,
+      status: response.status,
+    };
+  }
+}

package/services/ai.js

@@ -5,6 +5,7 @@ import { AssistService } from './ai/assist.js';
 import { VocabularyService } from './ai/vocabulary.js';
 import { EmailService } from './ai/email.js';
 import { ModelsService } from './ai/models.js';
+import { TextToSpeechService } from './ai/tts.js';
 import { translate as translateItems } from './ai/translate.js';
 import {
   getSettings as getAiSettings,
@@ -487,79 +488,6 @@ export class GenerativeService {
   // }
 }
 
-export class TextToSpeechService {
-  constructor(sdk) {
-    this.sdk = sdk;
-  }
-
-  async create({
-    text,
-    voice,
-    languageCode,
-    ssmlGender,
-    audioEncoding,
-    speakingRate,
-    pitch,
-    volumeGainDb,
-    effectsProfileIds,
-    createAccessKey,
-  }) {
-    this.sdk.validateParams(
-      {
-        text,
-        voice,
-        languageCode,
-        ssmlGender,
-        audioEncoding,
-        speakingRate,
-        pitch,
-        volumeGainDb,
-        effectsProfileIds,
-        createAccessKey,
-      },
-      {
-        text: { type: 'string', required: true },
-        voice: { type: 'string', required: false },
-        languageCode: { type: 'string', required: false },
-        ssmlGender: { type: 'string', required: false },
-        audioEncoding: { type: 'string', required: false },
-        speakingRate: { type: 'number', required: false },
-        pitch: { type: 'number', required: false },
-        volumeGainDb: { type: 'number', required: false },
-        effectsProfileIds: { type: 'array', required: false },
-        createAccessKey: { type: 'boolean', required: false },
-      },
-    );
-
-    const ttsData = { text };
-    if (voice) ttsData.voice = voice;
-    if (languageCode) ttsData.languageCode = languageCode;
-    if (ssmlGender) ttsData.ssmlGender = ssmlGender;
-    if (audioEncoding) ttsData.audioEncoding = audioEncoding;
-    if (speakingRate) ttsData.speakingRate = speakingRate;
-    if (pitch) ttsData.pitch = pitch;
-    if (volumeGainDb) ttsData.volumeGainDb = volumeGainDb;
-    if (effectsProfileIds) ttsData.effectsProfileIds = effectsProfileIds;
-    if (createAccessKey) ttsData.createAccessKey = createAccessKey;
-
-    const params = {
-      body: ttsData,
-    };
-
-    const result = await internalRequest(this.sdk, '/ai/tts', 'POST', params);
-    return result;
-  }
-
-  /**
-   * List available TTS voices
-   * @returns {Promise<Object>} { voices: Array, count: number, supportedEncodings: Array, supportedLanguages: Array }
-   */
-  async list() {
-    const result = await internalRequest(this.sdk, '/ai/tts', 'GET');
-    return result;
-  }
-}
-
 export class SpeechToTextService {
   constructor(sdk) {
     this.sdk = sdk;