@shogo-ai/voice 1.8.20 → 1.9.0

package/dist/route/agent.js

@@ -1,8 +1,8 @@
 import {
   defaultHandlers
-} from "../chunk-D2EVSUZX.js";
-import "../chunk-QHDIEE7O.js";
-import "../chunk-ZGYHDC5G.js";
+} from "../chunk-T3T4EBWY.js";
+import "../chunk-O362EJQR.js";
+import "../chunk-B4JC3DK2.js";
 import "../chunk-Q3TJAJ54.js";
 
 // src/route/agent.ts

package/dist/route/audio-tags.cjs

@@ -203,6 +203,7 @@ function resolveConvaiTtsModel(requested) {
   if (requested && CONVAI_SUPPORTED_TTS_MODELS.has(requested)) return requested;
   return CONVAI_TTS_MODEL_FALLBACK;
 }
+var DEFAULT_MUSIC_MODEL = "music_v1";
 var ElevenLabsApiError = class extends Error {
   constructor(message, status, body) {
     super(message);
@@ -403,6 +404,110 @@ var ElevenLabsClient = class {
     const contentType = res.headers.get("content-type") ?? accept;
     return { audio, modelId: used, contentType };
   }
+  /**
+   * Low-level escape hatch for any ElevenLabs endpoint not covered by a
+   * dedicated method (music, sound-generation, speech-to-text, dubbing,
+   * voice-design, …). Handles auth, query encoding, JSON body encoding,
+   * and JSON-vs-binary response parsing; non-2xx responses throw
+   * {@link ElevenLabsApiError} like the typed methods do.
+   *
+   * @example
+   * ```ts
+   * const { json } = await el.request({ method: 'GET', path: '/v1/models' })
+   * const { audio } = await el.request({
+   *   method: 'POST',
+   *   path: '/v1/sound-generation',
+   *   body: { text: 'door creak' },
+   *   accept: 'audio/mpeg',
+   * })
+   * ```
+   */
+  async request(opts) {
+    const accept = opts.accept ?? "application/json";
+    let url = `${this.baseUrl}${opts.path.startsWith("/") ? "" : "/"}${opts.path}`;
+    if (opts.query) {
+      const params = new URLSearchParams();
+      for (const [k, v] of Object.entries(opts.query)) {
+        if (v !== null && v !== void 0) params.set(k, String(v));
+      }
+      const qs = params.toString();
+      if (qs) url += `${url.includes("?") ? "&" : "?"}${qs}`;
+    }
+    const headers = this.headers({ accept, ...opts.headers ?? {} });
+    const init = { method: opts.method, headers };
+    if (opts.body !== void 0 && opts.body !== null) {
+      const isPlainPayload = typeof opts.body === "object" && !(opts.body instanceof ArrayBuffer) && !(opts.body instanceof Uint8Array) && !(typeof Blob !== "undefined" && opts.body instanceof Blob) && !(typeof FormData !== "undefined" && opts.body instanceof FormData);
+      if (isPlainPayload) {
+        headers["content-type"] = "application/json";
+        init.body = JSON.stringify(opts.body);
+      } else {
+        init.body = opts.body;
+      }
+    }
+    const res = await this.fetchImpl(url, init);
+    if (!res.ok) {
+      const text = await res.text();
+      throw new ElevenLabsApiError(
+        `request ${opts.method} ${opts.path} failed: ${res.status}`,
+        res.status,
+        text
+      );
+    }
+    const contentType = res.headers.get("content-type") ?? accept;
+    if (contentType.includes("application/json")) {
+      return { status: res.status, contentType, json: await res.json() };
+    }
+    return { status: res.status, contentType, audio: await res.arrayBuffer() };
+  }
+  /**
+   * Generate music via `POST /v1/music`. Provide exactly one of `prompt`
+   * or `compositionPlan` (ElevenLabs returns 422 otherwise). Returns the
+   * raw audio bytes plus the response content type.
+   *
+   * @example
+   * ```ts
+   * const { audio } = await el.composeMusic({
+   *   prompt: 'Upbeat synthwave with driving bass',
+   *   musicLengthMs: 30_000,
+   * })
+   * ```
+   */
+  async composeMusic(params) {
+    const hasPrompt = typeof params.prompt === "string" && params.prompt.length > 0;
+    const hasPlan = params.compositionPlan != null;
+    if (hasPrompt === hasPlan) {
+      throw new Error(
+        "composeMusic: provide exactly one of `prompt` or `compositionPlan`"
+      );
+    }
+    const body = {
+      model_id: params.modelId ?? DEFAULT_MUSIC_MODEL
+    };
+    if (hasPrompt) {
+      body.prompt = params.prompt;
+      if (params.musicLengthMs !== void 0) body.music_length_ms = params.musicLengthMs;
+      if (params.forceInstrumental !== void 0) {
+        body.force_instrumental = params.forceInstrumental;
+      }
+    } else {
+      body.composition_plan = params.compositionPlan;
+    }
+    const { audio, contentType } = await this.request({
+      method: "POST",
+      path: "/v1/music",
+      body,
+      accept: "audio/mpeg",
+      ...params.outputFormat ? { query: { output_format: params.outputFormat } } : {}
+    });
+    if (!audio) {
+      throw new ElevenLabsApiError(
+        "composeMusic: expected audio bytes but got a JSON response",
+        500,
+        ""
+      );
+    }
+    return { audio, contentType };
+  }
   /** Probe whether a given voice id still exists on the caller's account. */
   async voiceExists(voiceId) {
     const res = await this.fetchImpl(
@@ -584,6 +689,28 @@ async function readJson(req) {
 }
 var NOOP_LOGGER = () => {
 };
+function parseMusicBody(raw) {
+  if (raw === null || typeof raw !== "object") return { error: "Invalid JSON body" };
+  const b = raw;
+  const hasPrompt = typeof b.prompt === "string" && b.prompt.trim().length > 0;
+  const hasPlan = b.compositionPlan != null && typeof b.compositionPlan === "object";
+  if (hasPrompt === hasPlan) {
+    return { error: "Provide exactly one of `prompt` or `compositionPlan`" };
+  }
+  const params = {};
+  if (hasPrompt) {
+    params.prompt = b.prompt.trim();
+    if (typeof b.musicLengthMs === "number") params.musicLengthMs = b.musicLengthMs;
+    if (typeof b.forceInstrumental === "boolean") params.forceInstrumental = b.forceInstrumental;
+  } else {
+    params.compositionPlan = b.compositionPlan;
+  }
+  if (typeof b.modelId === "string" && b.modelId.length > 0) params.modelId = b.modelId;
+  if (typeof b.outputFormat === "string" && b.outputFormat.length > 0) {
+    params.outputFormat = b.outputFormat;
+  }
+  return { params };
+}
 function resolveProxyOptions(config) {
   if (config.proxy) {
     if (!config.proxy.runtimeToken || !config.proxy.projectId || !config.proxy.apiUrl) {
@@ -669,6 +796,34 @@ function createProxyHandlers(proxy, log) {
       }
     });
   }
+  async function doMusic(req) {
+    if (req.method !== "POST") return json({ error: "Method Not Allowed" }, 405);
+    const raw = await readJson(req);
+    const parsed = parseMusicBody(raw);
+    if ("error" in parsed) return json({ error: parsed.error }, 400);
+    try {
+      const upstream = await fetchImpl(proxyUrl("/api/voice/music", {}), {
+        method: "POST",
+        headers: {
+          "x-runtime-token": proxy.runtimeToken,
+          "content-type": "application/json",
+          accept: "audio/mpeg"
+        },
+        body: JSON.stringify(parsed.params),
+        credentials: "omit"
+      });
+      const buf = await upstream.arrayBuffer();
+      const headers = {};
+      const ct = upstream.headers.get("content-type");
+      if (ct) headers["content-type"] = ct;
+      return new Response(buf, { status: upstream.status, headers });
+    } catch (err) {
+      log("error", "proxy music failed", {
+        error: err instanceof Error ? err.message : String(err)
+      });
+      return json({ error: "music proxy failed" }, 502);
+    }
+  }
   function notImplemented(label) {
     return async () => json(
       {
@@ -681,6 +836,7 @@ function createProxyHandlers(proxy, log) {
   return {
     signedUrl: doSignedUrl,
     tts: notImplemented("tts"),
+    music: doMusic,
     agent: {
       create: notImplemented("agent.create"),
       patch: notImplemented("agent.patch"),
@@ -821,6 +977,25 @@ function createVoiceHandlers(config) {
       return handleElError(e, "TTS failed");
     }
   }
+  async function doMusic(req) {
+    if (req.method !== "POST") return json({ error: "Method Not Allowed" }, 405);
+    const u = await resolveUser(req);
+    if (u instanceof Response) return u;
+    const raw = await readJson(req);
+    const parsed = parseMusicBody(raw);
+    if ("error" in parsed) return json({ error: parsed.error }, 400);
+    try {
+      const { audio, contentType } = await el.composeMusic(parsed.params);
+      return new Response(audio, {
+        headers: {
+          "content-type": contentType,
+          "cache-control": "private, max-age=3600"
+        }
+      });
+    } catch (e) {
+      return handleElError(e, "Music generation failed");
+    }
+  }
   async function doAgentCreate(req) {
     if (req.method !== "POST") return json({ error: "Method Not Allowed" }, 405);
     const u = await resolveUser(req);
@@ -951,6 +1126,7 @@ function createVoiceHandlers(config) {
   return {
     signedUrl: doSignedUrl,
     tts: doTts,
+    music: doMusic,
     agent: {
       create: doAgentCreate,
       patch: doAgentPatch,

package/dist/route/audio-tags.js

@@ -1,8 +1,8 @@
 import {
   defaultHandlers
-} from "../chunk-D2EVSUZX.js";
-import "../chunk-QHDIEE7O.js";
-import "../chunk-ZGYHDC5G.js";
+} from "../chunk-T3T4EBWY.js";
+import "../chunk-O362EJQR.js";
+import "../chunk-B4JC3DK2.js";
 import "../chunk-Q3TJAJ54.js";
 
 // src/route/audio-tags.ts

package/dist/route/index.cjs

@@ -204,6 +204,7 @@ function resolveConvaiTtsModel(requested) {
   if (requested && CONVAI_SUPPORTED_TTS_MODELS.has(requested)) return requested;
   return CONVAI_TTS_MODEL_FALLBACK;
 }
+var DEFAULT_MUSIC_MODEL = "music_v1";
 var ElevenLabsApiError = class extends Error {
   constructor(message, status, body) {
     super(message);
@@ -404,6 +405,110 @@ var ElevenLabsClient = class {
     const contentType = res.headers.get("content-type") ?? accept;
     return { audio, modelId: used, contentType };
   }
+  /**
+   * Low-level escape hatch for any ElevenLabs endpoint not covered by a
+   * dedicated method (music, sound-generation, speech-to-text, dubbing,
+   * voice-design, …). Handles auth, query encoding, JSON body encoding,
+   * and JSON-vs-binary response parsing; non-2xx responses throw
+   * {@link ElevenLabsApiError} like the typed methods do.
+   *
+   * @example
+   * ```ts
+   * const { json } = await el.request({ method: 'GET', path: '/v1/models' })
+   * const { audio } = await el.request({
+   *   method: 'POST',
+   *   path: '/v1/sound-generation',
+   *   body: { text: 'door creak' },
+   *   accept: 'audio/mpeg',
+   * })
+   * ```
+   */
+  async request(opts) {
+    const accept = opts.accept ?? "application/json";
+    let url = `${this.baseUrl}${opts.path.startsWith("/") ? "" : "/"}${opts.path}`;
+    if (opts.query) {
+      const params = new URLSearchParams();
+      for (const [k, v] of Object.entries(opts.query)) {
+        if (v !== null && v !== void 0) params.set(k, String(v));
+      }
+      const qs = params.toString();
+      if (qs) url += `${url.includes("?") ? "&" : "?"}${qs}`;
+    }
+    const headers = this.headers({ accept, ...opts.headers ?? {} });
+    const init = { method: opts.method, headers };
+    if (opts.body !== void 0 && opts.body !== null) {
+      const isPlainPayload = typeof opts.body === "object" && !(opts.body instanceof ArrayBuffer) && !(opts.body instanceof Uint8Array) && !(typeof Blob !== "undefined" && opts.body instanceof Blob) && !(typeof FormData !== "undefined" && opts.body instanceof FormData);
+      if (isPlainPayload) {
+        headers["content-type"] = "application/json";
+        init.body = JSON.stringify(opts.body);
+      } else {
+        init.body = opts.body;
+      }
+    }
+    const res = await this.fetchImpl(url, init);
+    if (!res.ok) {
+      const text = await res.text();
+      throw new ElevenLabsApiError(
+        `request ${opts.method} ${opts.path} failed: ${res.status}`,
+        res.status,
+        text
+      );
+    }
+    const contentType = res.headers.get("content-type") ?? accept;
+    if (contentType.includes("application/json")) {
+      return { status: res.status, contentType, json: await res.json() };
+    }
+    return { status: res.status, contentType, audio: await res.arrayBuffer() };
+  }
+  /**
+   * Generate music via `POST /v1/music`. Provide exactly one of `prompt`
+   * or `compositionPlan` (ElevenLabs returns 422 otherwise). Returns the
+   * raw audio bytes plus the response content type.
+   *
+   * @example
+   * ```ts
+   * const { audio } = await el.composeMusic({
+   *   prompt: 'Upbeat synthwave with driving bass',
+   *   musicLengthMs: 30_000,
+   * })
+   * ```
+   */
+  async composeMusic(params) {
+    const hasPrompt = typeof params.prompt === "string" && params.prompt.length > 0;
+    const hasPlan = params.compositionPlan != null;
+    if (hasPrompt === hasPlan) {
+      throw new Error(
+        "composeMusic: provide exactly one of `prompt` or `compositionPlan`"
+      );
+    }
+    const body = {
+      model_id: params.modelId ?? DEFAULT_MUSIC_MODEL
+    };
+    if (hasPrompt) {
+      body.prompt = params.prompt;
+      if (params.musicLengthMs !== void 0) body.music_length_ms = params.musicLengthMs;
+      if (params.forceInstrumental !== void 0) {
+        body.force_instrumental = params.forceInstrumental;
+      }
+    } else {
+      body.composition_plan = params.compositionPlan;
+    }
+    const { audio, contentType } = await this.request({
+      method: "POST",
+      path: "/v1/music",
+      body,
+      accept: "audio/mpeg",
+      ...params.outputFormat ? { query: { output_format: params.outputFormat } } : {}
+    });
+    if (!audio) {
+      throw new ElevenLabsApiError(
+        "composeMusic: expected audio bytes but got a JSON response",
+        500,
+        ""
+      );
+    }
+    return { audio, contentType };
+  }
   /** Probe whether a given voice id still exists on the caller's account. */
   async voiceExists(voiceId) {
     const res = await this.fetchImpl(
@@ -585,6 +690,28 @@ async function readJson(req) {
 }
 var NOOP_LOGGER = () => {
 };
+function parseMusicBody(raw) {
+  if (raw === null || typeof raw !== "object") return { error: "Invalid JSON body" };
+  const b = raw;
+  const hasPrompt = typeof b.prompt === "string" && b.prompt.trim().length > 0;
+  const hasPlan = b.compositionPlan != null && typeof b.compositionPlan === "object";
+  if (hasPrompt === hasPlan) {
+    return { error: "Provide exactly one of `prompt` or `compositionPlan`" };
+  }
+  const params = {};
+  if (hasPrompt) {
+    params.prompt = b.prompt.trim();
+    if (typeof b.musicLengthMs === "number") params.musicLengthMs = b.musicLengthMs;
+    if (typeof b.forceInstrumental === "boolean") params.forceInstrumental = b.forceInstrumental;
+  } else {
+    params.compositionPlan = b.compositionPlan;
+  }
+  if (typeof b.modelId === "string" && b.modelId.length > 0) params.modelId = b.modelId;
+  if (typeof b.outputFormat === "string" && b.outputFormat.length > 0) {
+    params.outputFormat = b.outputFormat;
+  }
+  return { params };
+}
 function resolveProxyOptions(config) {
   if (config.proxy) {
     if (!config.proxy.runtimeToken || !config.proxy.projectId || !config.proxy.apiUrl) {
@@ -670,6 +797,34 @@ function createProxyHandlers(proxy, log) {
       }
     });
   }
+  async function doMusic(req) {
+    if (req.method !== "POST") return json({ error: "Method Not Allowed" }, 405);
+    const raw = await readJson(req);
+    const parsed = parseMusicBody(raw);
+    if ("error" in parsed) return json({ error: parsed.error }, 400);
+    try {
+      const upstream = await fetchImpl(proxyUrl("/api/voice/music", {}), {
+        method: "POST",
+        headers: {
+          "x-runtime-token": proxy.runtimeToken,
+          "content-type": "application/json",
+          accept: "audio/mpeg"
+        },
+        body: JSON.stringify(parsed.params),
+        credentials: "omit"
+      });
+      const buf = await upstream.arrayBuffer();
+      const headers = {};
+      const ct = upstream.headers.get("content-type");
+      if (ct) headers["content-type"] = ct;
+      return new Response(buf, { status: upstream.status, headers });
+    } catch (err) {
+      log("error", "proxy music failed", {
+        error: err instanceof Error ? err.message : String(err)
+      });
+      return json({ error: "music proxy failed" }, 502);
+    }
+  }
   function notImplemented(label) {
     return async () => json(
       {
@@ -682,6 +837,7 @@ function createProxyHandlers(proxy, log) {
   return {
     signedUrl: doSignedUrl,
     tts: notImplemented("tts"),
+    music: doMusic,
     agent: {
       create: notImplemented("agent.create"),
       patch: notImplemented("agent.patch"),
@@ -822,6 +978,25 @@ function createVoiceHandlers(config) {
       return handleElError(e, "TTS failed");
     }
   }
+  async function doMusic(req) {
+    if (req.method !== "POST") return json({ error: "Method Not Allowed" }, 405);
+    const u = await resolveUser(req);
+    if (u instanceof Response) return u;
+    const raw = await readJson(req);
+    const parsed = parseMusicBody(raw);
+    if ("error" in parsed) return json({ error: parsed.error }, 400);
+    try {
+      const { audio, contentType } = await el.composeMusic(parsed.params);
+      return new Response(audio, {
+        headers: {
+          "content-type": contentType,
+          "cache-control": "private, max-age=3600"
+        }
+      });
+    } catch (e) {
+      return handleElError(e, "Music generation failed");
+    }
+  }
   async function doAgentCreate(req) {
     if (req.method !== "POST") return json({ error: "Method Not Allowed" }, 405);
     const u = await resolveUser(req);
@@ -952,6 +1127,7 @@ function createVoiceHandlers(config) {
   return {
     signedUrl: doSignedUrl,
     tts: doTts,
+    music: doMusic,
     agent: {
       create: doAgentCreate,
       patch: doAgentPatch,
@@ -971,6 +1147,7 @@ function createVoiceRoute(config) {
   return {
     signedUrl: { GET: (req) => v.signedUrl(req) },
     ttsPreview: { POST: (req) => v.tts(req) },
+    music: { POST: (req) => v.music(req) },
     agent: {
       POST: (req) => v.agent.create(req),
       PATCH: (req) => v.agent.patch(req),

package/dist/route/index.d.cts

@@ -1,6 +1,6 @@
 import { VoiceHandlersConfig, VoiceHandlers } from '../server.cjs';
 import 'node:http';
-import '../types-BKJtOUbE.cjs';
+import '../types-W9y6eMj3.cjs';
 
 /**
  * @shogo-ai/sdk/voice/route
@@ -79,6 +79,10 @@ interface VoiceRoute {
     ttsPreview: {
         POST: (req: Request) => Promise<Response>;
     };
+    /** `POST /voice/music` */
+    music: {
+        POST: (req: Request) => Promise<Response>;
+    };
     /** `POST | PATCH | DELETE /voice/agent` */
     agent: {
         POST: (req: Request) => Promise<Response>;

package/dist/route/index.d.ts

@@ -1,6 +1,6 @@
 import { VoiceHandlersConfig, VoiceHandlers } from '../server.js';
 import 'node:http';
-import '../types-BKJtOUbE.js';
+import '../types-W9y6eMj3.js';
 
 /**
  * @shogo-ai/sdk/voice/route
@@ -79,6 +79,10 @@ interface VoiceRoute {
     ttsPreview: {
         POST: (req: Request) => Promise<Response>;
     };
+    /** `POST /voice/music` */
+    music: {
+        POST: (req: Request) => Promise<Response>;
+    };
     /** `POST | PATCH | DELETE /voice/agent` */
     agent: {
         POST: (req: Request) => Promise<Response>;

package/dist/route/index.js

@@ -1,9 +1,9 @@
 import {
   createVoiceRoute,
   defaultHandlers
-} from "../chunk-D2EVSUZX.js";
-import "../chunk-QHDIEE7O.js";
-import "../chunk-ZGYHDC5G.js";
+} from "../chunk-T3T4EBWY.js";
+import "../chunk-O362EJQR.js";
+import "../chunk-B4JC3DK2.js";
 import "../chunk-Q3TJAJ54.js";
 export {
   createVoiceRoute,