@cartesia/cartesia-js 1.0.3 → 1.2.1-alpha.2

package/dist/index.cjs

@@ -9,6 +9,9 @@ var __getOwnPropSymbols = Object.getOwnPropertySymbols;
 var __getProtoOf = Object.getPrototypeOf;
 var __hasOwnProp = Object.prototype.hasOwnProperty;
 var __propIsEnum = Object.prototype.propertyIsEnumerable;
+var __typeError = (msg) => {
+  throw TypeError(msg);
+};
 var __defNormalProp = (obj, key, value) => key in obj ? __defProp(obj, key, { enumerable: true, configurable: true, writable: true, value }) : obj[key] = value;
 var __spreadValues = (a, b) => {
   for (var prop in b || (b = {}))
@@ -22,18 +25,6 @@ var __spreadValues = (a, b) => {
   return a;
 };
 var __spreadProps = (a, b) => __defProps(a, __getOwnPropDescs(b));
-var __objRest = (source, exclude) => {
-  var target = {};
-  for (var prop in source)
-    if (__hasOwnProp.call(source, prop) && exclude.indexOf(prop) < 0)
-      target[prop] = source[prop];
-  if (source != null && __getOwnPropSymbols)
-    for (var prop of __getOwnPropSymbols(source)) {
-      if (exclude.indexOf(prop) < 0 && __propIsEnum.call(source, prop))
-        target[prop] = source[prop];
-    }
-  return target;
-};
 var __export = (target, all) => {
   for (var name in all)
     __defProp(target, name, { get: all[name], enumerable: true });
@@ -55,28 +46,11 @@ var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__ge
   mod
 ));
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
-var __accessCheck = (obj, member, msg) => {
-  if (!member.has(obj))
-    throw TypeError("Cannot " + msg);
-};
-var __privateGet = (obj, member, getter) => {
-  __accessCheck(obj, member, "read from private field");
-  return getter ? getter.call(obj) : member.get(obj);
-};
-var __privateAdd = (obj, member, value) => {
-  if (member.has(obj))
-    throw TypeError("Cannot add the same private member more than once");
-  member instanceof WeakSet ? member.add(obj) : member.set(obj, value);
-};
-var __privateSet = (obj, member, value, setter) => {
-  __accessCheck(obj, member, "write to private field");
-  setter ? setter.call(obj, value) : member.set(obj, value);
-  return value;
-};
-var __privateMethod = (obj, member, method) => {
-  __accessCheck(obj, member, "access private method");
-  return method;
-};
+var __accessCheck = (obj, member, msg) => member.has(obj) || __typeError("Cannot " + msg);
+var __privateGet = (obj, member, getter) => (__accessCheck(obj, member, "read from private field"), getter ? getter.call(obj) : member.get(obj));
+var __privateAdd = (obj, member, value) => member.has(obj) ? __typeError("Cannot add the same private member more than once") : member instanceof WeakSet ? member.add(obj) : member.set(obj, value);
+var __privateSet = (obj, member, value, setter) => (__accessCheck(obj, member, "write to private field"), setter ? setter.call(obj, value) : member.set(obj, value), value);
+var __privateMethod = (obj, member, method) => (__accessCheck(obj, member, "access private method"), method);
 var __async = (__this, __arguments, generator) => {
   return new Promise((resolve, reject) => {
     var fulfilled = (value) => {
@@ -161,7 +135,7 @@ var ENCODING_MAP = {
   pcm_alaw: { arrayType: Uint8Array, bytesPerElement: 1 },
   pcm_mulaw: { arrayType: Uint8Array, bytesPerElement: 1 }
 };
-var _emitter, _buffer, _readIndex, _writeIndex, _closed, _sampleRate, _encoding, _container, _createBuffer, createBuffer_fn;
+var _emitter, _buffer, _readIndex, _writeIndex, _closed, _sampleRate, _encoding, _container, _Source_instances, createBuffer_fn;
 var Source = class {
   /**
    * Create a new Source.
@@ -174,21 +148,15 @@ var Source = class {
     encoding,
     container
   }) {
-    /**
-     * Create a new buffer for the source.
-     *
-     * @param size - The size of the buffer to create.
-     * @returns The new buffer as a TypedArray based on the encoding.
-     */
-    __privateAdd(this, _createBuffer);
+    __privateAdd(this, _Source_instances);
     __privateAdd(this, _emitter, new import_emittery.default());
-    __privateAdd(this, _buffer, void 0);
+    __privateAdd(this, _buffer);
     __privateAdd(this, _readIndex, 0);
     __privateAdd(this, _writeIndex, 0);
     __privateAdd(this, _closed, false);
-    __privateAdd(this, _sampleRate, void 0);
-    __privateAdd(this, _encoding, void 0);
-    __privateAdd(this, _container, void 0);
+    __privateAdd(this, _sampleRate);
+    __privateAdd(this, _encoding);
+    __privateAdd(this, _container);
     this.on = __privateGet(this, _emitter).on.bind(__privateGet(this, _emitter));
     this.once = __privateGet(this, _emitter).once.bind(__privateGet(this, _emitter));
     this.events = __privateGet(this, _emitter).events.bind(__privateGet(this, _emitter));
@@ -196,7 +164,7 @@ var Source = class {
     __privateSet(this, _sampleRate, sampleRate);
     __privateSet(this, _encoding, encoding);
     __privateSet(this, _container, container);
-    __privateSet(this, _buffer, __privateMethod(this, _createBuffer, createBuffer_fn).call(this, 1024));
+    __privateSet(this, _buffer, __privateMethod(this, _Source_instances, createBuffer_fn).call(this, 1024));
   }
   get sampleRate() {
     return __privateGet(this, _sampleRate);
@@ -220,7 +188,7 @@ var Source = class {
         while (newCapacity < requiredCapacity) {
           newCapacity *= 2;
         }
-        const newBuffer = __privateMethod(this, _createBuffer, createBuffer_fn).call(this, newCapacity);
+        const newBuffer = __privateMethod(this, _Source_instances, createBuffer_fn).call(this, newCapacity);
         newBuffer.set(__privateGet(this, _buffer));
         __privateSet(this, _buffer, newBuffer);
       }
@@ -253,6 +221,37 @@ var Source = class {
       return read;
     });
   }
+  /**
+   * Seek in the buffer.
+   *
+   * @param offset The offset to seek to.
+   * @param whence The position to seek from.
+   * @returns The new position in the buffer.
+   * @throws {Error} If the seek is invalid.
+   */
+  seek(offset, whence) {
+    return __async(this, null, function* () {
+      let position = __privateGet(this, _readIndex);
+      switch (whence) {
+        case "start":
+          position = offset;
+          break;
+        case "current":
+          position += offset;
+          break;
+        case "end":
+          position = __privateGet(this, _writeIndex) + offset;
+          break;
+        default:
+          throw new Error(`Invalid seek mode: ${whence}`);
+      }
+      if (position < 0 || position > __privateGet(this, _writeIndex)) {
+        throw new Error("Seek out of bounds");
+      }
+      __privateSet(this, _readIndex, position);
+      return position;
+    });
+  }
   /**
    * Get the number of samples in a given duration.
    *
@@ -294,7 +293,13 @@ _closed = new WeakMap();
 _sampleRate = new WeakMap();
 _encoding = new WeakMap();
 _container = new WeakMap();
-_createBuffer = new WeakSet();
+_Source_instances = new WeakSet();
+/**
+ * Create a new buffer for the source.
+ *
+ * @param size - The size of the buffer to create.
+ * @returns The new buffer as a TypedArray based on the encoding.
+ */
 createBuffer_fn = function(size) {
   const { arrayType: ArrayType } = ENCODING_MAP[__privateGet(this, _encoding)];
   return new ArrayType(size);
@@ -370,7 +375,7 @@ function getEmitteryCallbacks(emitter) {
 }
 
 // src/tts/websocket.ts
-var _isConnected, _sampleRate2, _container2, _encoding2, _generateId, generateId_fn;
+var _isConnected, _sampleRate2, _container2, _encoding2, _WebSocket_instances, generateId_fn;
 var WebSocket = class extends Client {
   /**
    * Create a new WebSocket client.
@@ -379,19 +384,11 @@ var WebSocket = class extends Client {
    */
   constructor({ sampleRate, container, encoding }, ...args) {
     super(...args);
-    /**
-     * Generate a unique ID suitable for a streaming context.
-     *
-     * Not suitable for security purposes or as a primary key, since
-     * it lacks the amount of entropy required for those use cases.
-     *
-     * @returns A unique ID.
-     */
-    __privateAdd(this, _generateId);
+    __privateAdd(this, _WebSocket_instances);
     __privateAdd(this, _isConnected, false);
-    __privateAdd(this, _sampleRate2, void 0);
-    __privateAdd(this, _container2, void 0);
-    __privateAdd(this, _encoding2, void 0);
+    __privateAdd(this, _sampleRate2);
+    __privateAdd(this, _container2);
+    __privateAdd(this, _encoding2);
     __privateSet(this, _sampleRate2, sampleRate);
     __privateSet(this, _container2, container != null ? container : "raw");
     __privateSet(this, _encoding2, encoding != null ? encoding : "pcm_f32le");
@@ -399,7 +396,7 @@ var WebSocket = class extends Client {
   /**
    * Send a message over the WebSocket to start a stream.
    *
-   * @param inputs - Stream options. Defined in the StreamRequest type.
+   * @param inputs - Generation parameters. Defined in the StreamRequest type.
    * @param options - Options for the stream.
    * @param options.timeout - The maximum time to wait for a chunk before cancelling the stream.
    *                          If set to `0`, the stream will not time out.
@@ -407,14 +404,13 @@ var WebSocket = class extends Client {
    * @returns An Emittery instance that emits messages from the WebSocket.
    * @returns An abort function that can be called to cancel the stream.
    */
-  send(_a, { timeout = 0 } = {}) {
-    var inputs = __objRest(_a, []);
-    var _a2, _b, _c, _d;
+  send(inputs, { timeout = 0 } = {}) {
+    var _a, _b, _c, _d;
     if (!__privateGet(this, _isConnected)) {
       throw new Error("Not connected to WebSocket. Call .connect() first.");
     }
     if (!inputs.context_id) {
-      inputs.context_id = __privateMethod(this, _generateId, generateId_fn).call(this);
+      inputs.context_id = __privateMethod(this, _WebSocket_instances, generateId_fn).call(this);
     }
     if (!inputs.output_format) {
       inputs.output_format = {
@@ -423,7 +419,7 @@ var WebSocket = class extends Client {
         sample_rate: __privateGet(this, _sampleRate2)
       };
     }
-    (_a2 = this.socket) == null ? void 0 : _a2.send(
+    (_a = this.socket) == null ? void 0 : _a.send(
       JSON.stringify(__spreadValues({}, inputs))
     );
     const emitter = new import_emittery2.default();
@@ -493,6 +489,32 @@ var WebSocket = class extends Client {
       stop: streamCompleteController.abort.bind(streamCompleteController)
     });
   }
+  /**
+   * Continue a stream.
+   *
+   * @param inputs - Generation parameters. Defined in the StreamRequest type, but must include a `context_id` field. `continue` is set to true by default.
+   */
+  continue(inputs) {
+    var _a;
+    if (!__privateGet(this, _isConnected)) {
+      throw new Error("Not connected to WebSocket. Call .connect() first.");
+    }
+    if (!inputs.context_id) {
+      throw new Error("context_id is required to continue a context.");
+    }
+    if (!inputs.output_format) {
+      inputs.output_format = {
+        container: __privateGet(this, _container2),
+        encoding: __privateGet(this, _encoding2),
+        sample_rate: __privateGet(this, _sampleRate2)
+      };
+    }
+    (_a = this.socket) == null ? void 0 : _a.send(
+      JSON.stringify(__spreadValues({
+        continue: true
+      }, inputs))
+    );
+  }
   /**
    * Authenticate and connect to a Cartesia streaming WebSocket.
    *
@@ -568,7 +590,15 @@ _isConnected = new WeakMap();
 _sampleRate2 = new WeakMap();
 _container2 = new WeakMap();
 _encoding2 = new WeakMap();
-_generateId = new WeakSet();
+_WebSocket_instances = new WeakSet();
+/**
+ * Generate a unique ID suitable for a streaming context.
+ *
+ * Not suitable for security purposes or as a primary key, since
+ * it lacks the amount of entropy required for those use cases.
+ *
+ * @returns A unique ID.
+ */
 generateId_fn = function() {
   return (0, import_human_id.humanId)({
     separator: "-",
@@ -589,6 +619,53 @@ var TTS = class extends Client {
       baseUrl: this.baseUrl
     });
   }
+  /**
+   * Generate audio bytes from text.
+   *
+   * @param options - The options for the request.
+   * @returns {Promise<ArrayBuffer>} A promise that resolves to an ArrayBuffer containing the audio bytes.
+   */
+  bytes(options) {
+    return __async(this, null, function* () {
+      const response = yield this._fetch("/tts/bytes", {
+        method: "POST",
+        body: JSON.stringify(options)
+      });
+      return response.arrayBuffer();
+    });
+  }
+};
+
+// src/voice-changer/index.ts
+var VoiceChanger = class extends Client {
+  bytes(options) {
+    return __async(this, null, function* () {
+      const formData = new FormData();
+      formData.append("clip", options.clip);
+      formData.append("voice[id]", options.voice.id);
+      const fmt = options.output_format;
+      formData.append("output_format[container]", fmt.container);
+      if ("encoding" in fmt) {
+        formData.append("output_format[encoding]", fmt.encoding);
+      }
+      if ("bit_rate" in fmt) {
+        formData.append("output_format[bit_rate]", fmt.bit_rate.toString());
+      }
+      if ("sample_rate" in fmt) {
+        formData.append("output_format[sample_rate]", fmt.sample_rate.toString());
+      }
+      const response = yield this._fetch("/voice-changer/bytes", {
+        method: "POST",
+        body: formData
+      });
+      if (!response.ok) {
+        throw new Error(
+          `Voice changer error! status: ${response.status}, message: ${yield response.text()}`
+        );
+      }
+      return { buffer: yield response.arrayBuffer() };
+    });
+  }
 };
 
 // src/voices/index.ts
@@ -642,10 +719,18 @@ var Voices = class extends Client {
   }
   mix(options) {
     return __async(this, null, function* () {
-      const request = options;
       const response = yield this._fetch("/voices/mix", {
         method: "POST",
-        body: JSON.stringify(request)
+        body: JSON.stringify(options)
+      });
+      return response.json();
+    });
+  }
+  localize(options) {
+    return __async(this, null, function* () {
+      const response = yield this._fetch("/voices/localize", {
+        method: "POST",
+        body: JSON.stringify(options)
       });
       return response.json();
     });
@@ -658,11 +743,12 @@ var Cartesia = class extends Client {
     super(options);
     this.tts = new TTS(options);
     this.voices = new Voices(options);
+    this.voiceChanger = new VoiceChanger(options);
   }
 };
 
 // src/tts/player.ts
-var _context, _startNextPlaybackAt, _bufferDuration, _playBuffer, playBuffer_fn;
+var _context, _startNextPlaybackAt, _bufferDuration, _Player_instances, playBuffer_fn;
 var Player = class {
   /**
    * Create a new Player.
@@ -671,10 +757,10 @@ var Player = class {
    * @param options.bufferDuration - The duration of the audio buffer to play.
    */
   constructor({ bufferDuration }) {
-    __privateAdd(this, _playBuffer);
+    __privateAdd(this, _Player_instances);
     __privateAdd(this, _context, null);
     __privateAdd(this, _startNextPlaybackAt, 0);
-    __privateAdd(this, _bufferDuration, void 0);
+    __privateAdd(this, _bufferDuration);
     __privateSet(this, _bufferDuration, bufferDuration);
   }
   /**
@@ -694,7 +780,7 @@ var Player = class {
       while (true) {
         const read = yield source.read(buffer);
         const playableAudio = buffer.subarray(0, read);
-        plays.push(__privateMethod(this, _playBuffer, playBuffer_fn).call(this, playableAudio, source.sampleRate));
+        plays.push(__privateMethod(this, _Player_instances, playBuffer_fn).call(this, playableAudio, source.sampleRate));
         if (read < buffer.length) {
           break;
         }
@@ -763,7 +849,7 @@ var Player = class {
 _context = new WeakMap();
 _startNextPlaybackAt = new WeakMap();
 _bufferDuration = new WeakMap();
-_playBuffer = new WeakSet();
+_Player_instances = new WeakSet();
 playBuffer_fn = function(buf, sampleRate) {
   return __async(this, null, function* () {
     if (!__privateGet(this, _context)) {

package/dist/index.d.cts

@@ -1,10 +1,11 @@
 export { Cartesia, Cartesia as default } from './lib/index.cjs';
-export { Chunk, ClientOptions, CloneOptions, CloneResponse, ConnectionEventData, CreateVoice, EmitteryCallbacks, Emotion, EmotionControl, Encoding, Intensity, MixVoicesOptions, MixVoicesResponse, Sentinel, SourceEventData, StreamOptions, StreamRequest, TypedArray, UpdateVoice, Voice, VoiceOptions, VoiceSpecifier, VoiceToMix, WebSocketBaseResponse, WebSocketChunkResponse, WebSocketErrorResponse, WebSocketOptions, WebSocketResponse, WebSocketTimestampsResponse, WordTimestamps } from './types/index.cjs';
+export { BytesRequest, Chunk, ClientOptions, CloneOptions, CloneResponse, ConnectionEventData, ContinueRequest, CreateVoice, EmitteryCallbacks, Emotion, EmotionControl, Encoding, Intensity, Language, LocalizeOptions, LocalizeResponse, MixVoicesOptions, MixVoicesResponse, Sentinel, SourceEventData, StreamOptions, StreamRequest, TypedArray, UpdateVoice, Voice, VoiceChangerBytesResponse, VoiceChangerOptions, VoiceOptions, VoiceSpecifier, VoiceToMix, WebSocketBaseResponse, WebSocketChunkResponse, WebSocketErrorResponse, WebSocketOptions, WebSocketResponse, WebSocketTimestampsResponse, WordTimestamps } from './types/index.cjs';
 export { default as WebPlayer } from './tts/player.cjs';
 export { default as Source } from './tts/source.cjs';
 export { default as WebSocket } from './tts/websocket.cjs';
 import './tts/index.cjs';
 import './lib/client.cjs';
+import './voice-changer/index.cjs';
 import './voices/index.cjs';
 import 'emittery';
 import 'partysocket';

package/dist/index.d.ts

@@ -1,10 +1,11 @@
 export { Cartesia, Cartesia as default } from './lib/index.js';
-export { Chunk, ClientOptions, CloneOptions, CloneResponse, ConnectionEventData, CreateVoice, EmitteryCallbacks, Emotion, EmotionControl, Encoding, Intensity, MixVoicesOptions, MixVoicesResponse, Sentinel, SourceEventData, StreamOptions, StreamRequest, TypedArray, UpdateVoice, Voice, VoiceOptions, VoiceSpecifier, VoiceToMix, WebSocketBaseResponse, WebSocketChunkResponse, WebSocketErrorResponse, WebSocketOptions, WebSocketResponse, WebSocketTimestampsResponse, WordTimestamps } from './types/index.js';
+export { BytesRequest, Chunk, ClientOptions, CloneOptions, CloneResponse, ConnectionEventData, ContinueRequest, CreateVoice, EmitteryCallbacks, Emotion, EmotionControl, Encoding, Intensity, Language, LocalizeOptions, LocalizeResponse, MixVoicesOptions, MixVoicesResponse, Sentinel, SourceEventData, StreamOptions, StreamRequest, TypedArray, UpdateVoice, Voice, VoiceChangerBytesResponse, VoiceChangerOptions, VoiceOptions, VoiceSpecifier, VoiceToMix, WebSocketBaseResponse, WebSocketChunkResponse, WebSocketErrorResponse, WebSocketOptions, WebSocketResponse, WebSocketTimestampsResponse, WordTimestamps } from './types/index.js';
 export { default as WebPlayer } from './tts/player.js';
 export { default as Source } from './tts/source.js';
 export { default as WebSocket } from './tts/websocket.js';
 import './tts/index.js';
 import './lib/client.js';
+import './voice-changer/index.js';
 import './voices/index.js';
 import 'emittery';
 import 'partysocket';

package/dist/index.js

@@ -1,22 +1,23 @@
 import "./chunk-FXPGR372.js";
 import {
   Cartesia
-} from "./chunk-ZF6HASZT.js";
-import "./chunk-YFN6TFR4.js";
-import "./chunk-NWCW6C7H.js";
+} from "./chunk-QWNB544W.js";
+import "./chunk-WLEVU3HN.js";
+import "./chunk-RJICGVPL.js";
+import "./chunk-GLZYI5DM.js";
 import {
   WebSocket
-} from "./chunk-EUW2435M.js";
-import "./chunk-PISCPZK4.js";
+} from "./chunk-WSIVWXEI.js";
+import "./chunk-FLWYXP5Z.js";
 import "./chunk-2BFEKY3F.js";
 import {
   Player
-} from "./chunk-LZO6K34D.js";
-import "./chunk-BHY7MNGT.js";
+} from "./chunk-CSOXALSC.js";
+import "./chunk-5SBAQNWQ.js";
 import {
   Source
-} from "./chunk-6YQ6KDIQ.js";
-import "./chunk-GHY2WEOK.js";
+} from "./chunk-I5YVYTNK.js";
+import "./chunk-NJDRWDQ3.js";
 export {
   Cartesia,
   Source,

package/dist/lib/client.js

@@ -1,8 +1,8 @@
 import {
   Client
-} from "../chunk-PISCPZK4.js";
+} from "../chunk-FLWYXP5Z.js";
 import "../chunk-2BFEKY3F.js";
-import "../chunk-GHY2WEOK.js";
+import "../chunk-NJDRWDQ3.js";
 export {
   Client
 };

package/dist/lib/constants.js

@@ -3,7 +3,7 @@ import {
   CARTESIA_VERSION,
   constructApiUrl
 } from "../chunk-2BFEKY3F.js";
-import "../chunk-GHY2WEOK.js";
+import "../chunk-NJDRWDQ3.js";
 export {
   BASE_URL,
   CARTESIA_VERSION,