@cartesia/cartesia-js 2.2.5 → 2.2.7

package/dist/api/resources/tts/types/GenerationRequest.d.ts

@@ -10,6 +10,7 @@ export interface GenerationRequest {
     voice: Cartesia.TtsRequestVoiceSpecifier;
     language?: Cartesia.SupportedLanguage;
     outputFormat: Cartesia.WebSocketRawOutputFormat;
+    generationConfig?: Cartesia.GenerationConfig;
     /**
      * The maximum duration of the audio in seconds. You do not usually need to specify this.
      * If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
@@ -32,10 +33,10 @@ export interface GenerationRequest {
     maxBufferDelayMs?: number;
     /** Whether to flush the context. */
     flush?: boolean;
-    /** Whether to return word-level timestamps. */
+    /** Whether to return word-level timestamps. If `false` (default), no word timestamps will be produced at all. If `true`, the server will return timestamp events containing word-level timing information. */
     addTimestamps?: boolean;
-    /** Whether to return phoneme-level timestamps. */
+    /** Whether to return phoneme-level timestamps. If `false` (default), no phoneme timestamps will be produced. If `true`, the server will return timestamp events containing phoneme-level timing information. */
     addPhonemeTimestamps?: boolean;
-    /** Whether to use the original transcript for timestamps. */
-    useOriginalTimestamps?: boolean;
+    /** Whether to use normalized timestamps (True) or original timestamps (False). */
+    useNormalizedTimestamps?: boolean;
 }

package/dist/api/resources/tts/types/Mp3OutputFormat.d.ts

@@ -2,6 +2,7 @@
  * This file was auto-generated by Fern from our API Definition.
  */
 export interface Mp3OutputFormat {
+    /** The sample rate of the audio in Hz. Supported sample rates are 8000, 16000, 22050, 24000, 44100, 48000. */
     sampleRate: number;
     /** The bit rate of the audio in bits per second. Supported bit rates are 32000, 64000, 96000, 128000, 192000. */
     bitRate: number;

package/dist/api/resources/tts/types/RawOutputFormat.d.ts

@@ -4,6 +4,7 @@
 import * as Cartesia from "../../../index";
 export interface RawOutputFormat {
     encoding: Cartesia.RawEncoding;
+    /** The sample rate of the audio in Hz. Supported sample rates are 8000, 16000, 22050, 24000, 44100, 48000. */
     sampleRate: number;
     bitRate?: number;
 }

package/dist/api/resources/tts/types/SseOutputFormat.d.ts

@@ -0,0 +1,10 @@
+/**
+ * This file was auto-generated by Fern from our API Definition.
+ */
+import * as Cartesia from "../../../index";
+export interface SseOutputFormat {
+    container: "raw";
+    encoding: Cartesia.RawEncoding;
+    /** The sample rate of the audio in Hz. Supported sample rates are 8000, 16000, 22050, 24000, 44100, 48000. */
+    sampleRate: number;
+}

package/dist/api/resources/tts/types/SseOutputFormat.js

@@ -0,0 +1,5 @@
+"use strict";
+/**
+ * This file was auto-generated by Fern from our API Definition.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });

package/dist/api/resources/tts/types/TtsRequest.d.ts

@@ -9,6 +9,7 @@ export interface TtsRequest {
     voice: Cartesia.TtsRequestVoiceSpecifier;
     language?: Cartesia.SupportedLanguage;
     outputFormat: Cartesia.OutputFormat;
+    generationConfig?: Cartesia.GenerationConfig;
     /**
      * The maximum duration of the audio in seconds. You do not usually need to specify this.
      * If the duration is not appropriate for the length of the transcript, the output audio may be truncated.

package/dist/api/resources/tts/types/TtssseRequest.d.ts

@@ -0,0 +1,27 @@
+/**
+ * This file was auto-generated by Fern from our API Definition.
+ */
+import * as Cartesia from "../../../index";
+export interface TtssseRequest {
+    /** The ID of the model to use for the generation. See [Models](/build-with-cartesia/models) for available models. */
+    modelId: string;
+    transcript: string;
+    voice: Cartesia.TtsRequestVoiceSpecifier;
+    language?: Cartesia.SupportedLanguage;
+    outputFormat: Cartesia.SseOutputFormat;
+    generationConfig?: Cartesia.GenerationConfig;
+    /**
+     * The maximum duration of the audio in seconds. You do not usually need to specify this.
+     * If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
+     */
+    duration?: number;
+    speed?: Cartesia.ModelSpeed;
+    /** Whether to return word-level timestamps. If `false` (default), no word timestamps will be produced at all. If `true`, the server will return timestamp events containing word-level timing information. */
+    addTimestamps?: boolean;
+    /** Whether to return phoneme-level timestamps. If `false` (default), no phoneme timestamps will be produced - if `add_timestamps` is `true`, the produced timestamps will be word timestamps instead. If `true`, the server will return timestamp events containing phoneme-level timing information. */
+    addPhonemeTimestamps?: boolean;
+    /** Whether to use normalized timestamps (True) or original timestamps (False). */
+    useNormalizedTimestamps?: boolean;
+    /** Optional context ID for this request. */
+    contextId?: Cartesia.ContextId;
+}

package/dist/api/resources/tts/types/TtssseRequest.js

@@ -0,0 +1,5 @@
+"use strict";
+/**
+ * This file was auto-generated by Fern from our API Definition.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });

package/dist/api/resources/tts/types/WebSocketRawOutputFormat.d.ts

@@ -5,5 +5,6 @@ import * as Cartesia from "../../../index";
 export interface WebSocketRawOutputFormat {
     container: "raw";
     encoding: Cartesia.RawEncoding;
+    /** The sample rate of the audio in Hz. Supported sample rates are 8000, 16000, 22050, 24000, 44100, 48000. */
     sampleRate: number;
 }

package/dist/api/resources/tts/types/WebSocketRequest.d.ts

@@ -4,10 +4,8 @@
 import * as Cartesia from "../../../index";
 export type WebSocketRequest = 
 /**
- * Use this to generate speech for a transcript.
- *  */
+ * Use this to generate speech for a transcript. */
 Cartesia.GenerationRequest
 /**
- * Use this to cancel a context, so that no more messages are generated for that context.
- *  */
+ * Use this to cancel a context, so that no more messages are generated for that context. */
  | Cartesia.CancelContextRequest;

package/dist/api/resources/tts/types/WebSocketTtsRequest.d.ts

@@ -6,13 +6,16 @@ export interface WebSocketTtsRequest {
     /** The ID of the model to use for the generation. See [Models](/build-with-cartesia/models) for available models. */
     modelId: string;
     outputFormat?: Cartesia.OutputFormat;
+    generationConfig?: Cartesia.GenerationConfig;
     transcript?: string;
     voice: Cartesia.TtsRequestVoiceSpecifier;
     duration?: number;
     language?: string;
+    /** Whether to return word-level timestamps. If `false` (default), no word timestamps will be produced at all. If `true`, the server will return timestamp events containing word-level timing information. */
     addTimestamps?: boolean;
+    /** Whether to return phoneme-level timestamps. If `false` (default), no phoneme timestamps will be produced - if `add_timestamps` is `true`, the produced timestamps will be word timestamps instead. If `true`, the server will return timestamp events containing phoneme-level timing information. */
     addPhonemeTimestamps?: boolean;
-    useOriginalTimestamps?: boolean;
+    useNormalizedTimestamps?: boolean;
     continue?: boolean;
     contextId?: string;
     maxBufferDelayMs?: number;

package/dist/api/resources/tts/types/index.d.ts

@@ -19,9 +19,13 @@ export * from "./WebSocketRawOutputFormat";
 export * from "./WebSocketRequest";
 export * from "./WebSocketTtsRequest";
 export * from "./TtsRequest";
+export * from "./TtssseRequest";
 export * from "./SupportedLanguage";
+export * from "./GenerationConfig";
+export * from "./Emotion";
 export * from "./OutputFormat";
 export * from "./RawOutputFormat";
+export * from "./SseOutputFormat";
 export * from "./RawEncoding";
 export * from "./WavOutputFormat";
 export * from "./Mp3OutputFormat";
@@ -32,4 +36,4 @@ export * from "./Controls";
 export * from "./Speed";
 export * from "./NumericalSpecifier";
 export * from "./NaturalSpecifier";
-export * from "./Emotion";
+export * from "./EmotionDeprecated";

package/dist/api/resources/tts/types/index.js

@@ -35,9 +35,13 @@ __exportStar(require("./WebSocketRawOutputFormat"), exports);
 __exportStar(require("./WebSocketRequest"), exports);
 __exportStar(require("./WebSocketTtsRequest"), exports);
 __exportStar(require("./TtsRequest"), exports);
+__exportStar(require("./TtssseRequest"), exports);
 __exportStar(require("./SupportedLanguage"), exports);
+__exportStar(require("./GenerationConfig"), exports);
+__exportStar(require("./Emotion"), exports);
 __exportStar(require("./OutputFormat"), exports);
 __exportStar(require("./RawOutputFormat"), exports);
+__exportStar(require("./SseOutputFormat"), exports);
 __exportStar(require("./RawEncoding"), exports);
 __exportStar(require("./WavOutputFormat"), exports);
 __exportStar(require("./Mp3OutputFormat"), exports);
@@ -48,4 +52,4 @@ __exportStar(require("./Controls"), exports);
 __exportStar(require("./Speed"), exports);
 __exportStar(require("./NumericalSpecifier"), exports);
 __exportStar(require("./NaturalSpecifier"), exports);
-__exportStar(require("./Emotion"), exports);
+__exportStar(require("./EmotionDeprecated"), exports);

package/dist/api/resources/voiceChanger/client/Client.d.ts

@@ -1,14 +1,17 @@
 /**
  * This file was auto-generated by Fern from our API Definition.
  */
-import { Blob } from "buffer";
-import * as fs from "fs";
-import * as core from "../../../../core";
 import * as environments from "../../../../environments";
+import * as core from "../../../../core";
+import * as fs from "fs";
+import { Blob } from "buffer";
 import * as Cartesia from "../../../index";
+import * as stream from "stream";
 export declare namespace VoiceChanger {
     interface Options {
         environment?: core.Supplier<environments.CartesiaEnvironment | string>;
+        /** Specify a custom URL to connect the client to. */
+        baseUrl?: core.Supplier<string>;
         apiKey?: core.Supplier<string | undefined>;
         /** Override the Cartesia-Version header */
         cartesiaVersion?: "2024-06-10";
@@ -23,6 +26,8 @@ export declare namespace VoiceChanger {
         abortSignal?: AbortSignal;
         /** Override the Cartesia-Version header */
         cartesiaVersion?: "2024-06-10";
+        /** Additional headers to include in the request. */
+        headers?: Record<string, string>;
     }
 }
 export declare class VoiceChanger {
@@ -33,7 +38,7 @@ export declare class VoiceChanger {
      *
      * This endpoint is priced at 15 characters per second of input audio.
      */
-    bytes(clip: File | fs.ReadStream | Blob, request: Cartesia.VoiceChangerBytesRequest, requestOptions?: VoiceChanger.RequestOptions): Promise<ArrayBuffer>;
+    bytes(clip: File | fs.ReadStream | Blob, request: Cartesia.VoiceChangerBytesRequest, requestOptions?: VoiceChanger.RequestOptions): Promise<stream.Readable>;
     sse(clip: File | fs.ReadStream | Blob, request: Cartesia.VoiceChangerSseRequest, requestOptions?: VoiceChanger.RequestOptions): Promise<core.Stream<Cartesia.StreamingResponse>>;
     protected _getCustomAuthorizationHeaders(): Promise<{
         "X-API-Key": string | undefined;

package/dist/api/resources/voiceChanger/client/Client.js

@@ -49,11 +49,11 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
 };
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.VoiceChanger = void 0;
-const url_join_1 = __importDefault(require("url-join"));
-const core = __importStar(require("../../../../core"));
 const environments = __importStar(require("../../../../environments"));
-const errors = __importStar(require("../../../../errors/index"));
+const core = __importStar(require("../../../../core"));
 const serializers = __importStar(require("../../../../serialization/index"));
+const url_join_1 = __importDefault(require("url-join"));
+const errors = __importStar(require("../../../../errors/index"));
 class VoiceChanger {
     constructor(_options = {}) {
         this._options = _options;
@@ -65,27 +65,29 @@ class VoiceChanger {
      */
     bytes(clip, request, requestOptions) {
         return __awaiter(this, void 0, void 0, function* () {
-            var _a, _b, _c, _d, _e;
+            var _a, _b, _c, _d, _e, _f;
             const _request = yield core.newFormData();
             yield _request.appendFile("clip", clip);
-            yield _request.append("voice[id]", request.voiceId);
-            yield _request.append("output_format[container]", request.outputFormatContainer);
-            yield _request.append("output_format[sample_rate]", request.outputFormatSampleRate.toString());
+            _request.append("voice[id]", request.voiceId);
+            _request.append("output_format[container]", serializers.OutputFormatContainer.jsonOrThrow(request.outputFormatContainer, {
+                unrecognizedObjectKeys: "strip",
+            }));
+            _request.append("output_format[sample_rate]", request.outputFormatSampleRate.toString());
             if (request.outputFormatEncoding != null) {
-                yield _request.append("output_format[encoding]", request.outputFormatEncoding);
+                _request.append("output_format[encoding]", serializers.RawEncoding.jsonOrThrow(request.outputFormatEncoding, { unrecognizedObjectKeys: "strip" }));
             }
             if (request.outputFormatBitRate != null) {
-                yield _request.append("output_format[bit_rate]", request.outputFormatBitRate.toString());
+                _request.append("output_format[bit_rate]", request.outputFormatBitRate.toString());
             }
             const _maybeEncodedRequest = yield _request.getRequest();
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
-                url: (0, url_join_1.default)((_b = (yield core.Supplier.get(this._options.environment))) !== null && _b !== void 0 ? _b : environments.CartesiaEnvironment.Production, "/voice-changer/bytes"),
+                url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, "/voice-changer/bytes"),
                 method: "POST",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_e = (_c = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _c !== void 0 ? _c : (_d = this._options) === null || _d === void 0 ? void 0 : _d.cartesiaVersion) !== null && _e !== void 0 ? _e : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.0.4-alpha0", "User-Agent": "@cartesia/cartesia-js/2.0.4-alpha0", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), _maybeEncodedRequest.headers),
+                headers: Object.assign(Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), _maybeEncodedRequest.headers), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 requestType: "file",
                 duplex: _maybeEncodedRequest.duplex,
                 body: _maybeEncodedRequest.body,
-                responseType: "arraybuffer",
+                responseType: "streaming",
                 timeoutMs: (requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.timeoutInSeconds) != null ? requestOptions.timeoutInSeconds * 1000 : 60000,
                 maxRetries: requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.maxRetries,
                 abortSignal: requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.abortSignal,
@@ -116,23 +118,25 @@ class VoiceChanger {
     }
     sse(clip, request, requestOptions) {
         return __awaiter(this, void 0, void 0, function* () {
-            var _a, _b, _c, _d, _e;
+            var _a, _b, _c, _d, _e, _f;
             const _request = yield core.newFormData();
             yield _request.appendFile("clip", clip);
-            yield _request.append("voice[id]", request.voiceId);
-            yield _request.append("output_format[container]", request.outputFormatContainer);
-            yield _request.append("output_format[sample_rate]", request.outputFormatSampleRate.toString());
+            _request.append("voice[id]", request.voiceId);
+            _request.append("output_format[container]", serializers.OutputFormatContainer.jsonOrThrow(request.outputFormatContainer, {
+                unrecognizedObjectKeys: "strip",
+            }));
+            _request.append("output_format[sample_rate]", request.outputFormatSampleRate.toString());
             if (request.outputFormatEncoding != null) {
-                yield _request.append("output_format[encoding]", request.outputFormatEncoding);
+                _request.append("output_format[encoding]", serializers.RawEncoding.jsonOrThrow(request.outputFormatEncoding, { unrecognizedObjectKeys: "strip" }));
             }
             if (request.outputFormatBitRate != null) {
-                yield _request.append("output_format[bit_rate]", request.outputFormatBitRate.toString());
+                _request.append("output_format[bit_rate]", request.outputFormatBitRate.toString());
             }
             const _maybeEncodedRequest = yield _request.getRequest();
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
-                url: (0, url_join_1.default)((_b = (yield core.Supplier.get(this._options.environment))) !== null && _b !== void 0 ? _b : environments.CartesiaEnvironment.Production, "/voice-changer/sse"),
+                url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, "/voice-changer/sse"),
                 method: "POST",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_e = (_c = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _c !== void 0 ? _c : (_d = this._options) === null || _d === void 0 ? void 0 : _d.cartesiaVersion) !== null && _e !== void 0 ? _e : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.0.4-alpha0", "User-Agent": "@cartesia/cartesia-js/2.0.4-alpha0", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), _maybeEncodedRequest.headers),
+                headers: Object.assign(Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), _maybeEncodedRequest.headers), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 requestType: "file",
                 duplex: _maybeEncodedRequest.duplex,
                 body: _maybeEncodedRequest.body,

package/dist/api/resources/voiceChanger/client/requests/VoiceChangerBytesRequest.d.ts

@@ -30,15 +30,10 @@ import * as Cartesia from "../../../../index";
 export interface VoiceChangerBytesRequest {
     voiceId: string;
     outputFormatContainer: Cartesia.OutputFormatContainer;
+    /** The sample rate of the output audio in Hz. Supported sample rates are 8000, 16000, 22050, 24000, 44100, 48000. */
     outputFormatSampleRate: number;
-    /**
-     * Required for `raw` and `wav` containers.
-     *
-     */
+    /** Required for `raw` and `wav` containers. */
     outputFormatEncoding?: Cartesia.RawEncoding;
-    /**
-     * Required for `mp3` containers.
-     *
-     */
+    /** Required for `mp3` containers. */
     outputFormatBitRate?: number;
 }

package/dist/api/resources/voiceChanger/client/requests/VoiceChangerSseRequest.d.ts

@@ -30,15 +30,10 @@ import * as Cartesia from "../../../../index";
 export interface VoiceChangerSseRequest {
     voiceId: string;
     outputFormatContainer: Cartesia.OutputFormatContainer;
+    /** The sample rate of the audio in Hz. Supported sample rates are 8000, 16000, 22050, 24000, 44100, 48000. */
     outputFormatSampleRate: number;
-    /**
-     * Required for `raw` and `wav` containers.
-     *
-     */
+    /** Required for `raw` and `wav` containers. */
     outputFormatEncoding?: Cartesia.RawEncoding;
-    /**
-     * Required for `mp3` containers.
-     *
-     */
+    /** Required for `mp3` containers. */
     outputFormatBitRate?: number;
 }

package/dist/api/resources/voices/client/Client.js

@@ -70,7 +70,7 @@ class Voices {
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
                 url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, "/voices/"),
                 method: "GET",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.5", "User-Agent": "@cartesia/cartesia-js/2.2.5", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
+                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 contentType: "application/json",
                 requestType: "json",
                 timeoutMs: (requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.timeoutInSeconds) != null ? requestOptions.timeoutInSeconds * 1000 : 60000,
@@ -155,7 +155,7 @@ class Voices {
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
                 url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, "/voices/clone"),
                 method: "POST",
-                headers: Object.assign(Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.5", "User-Agent": "@cartesia/cartesia-js/2.2.5", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), _maybeEncodedRequest.headers), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
+                headers: Object.assign(Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), _maybeEncodedRequest.headers), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 requestType: "file",
                 duplex: _maybeEncodedRequest.duplex,
                 body: _maybeEncodedRequest.body,
@@ -206,7 +206,7 @@ class Voices {
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
                 url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, `/voices/${encodeURIComponent(serializers.VoiceId.jsonOrThrow(id))}`),
                 method: "DELETE",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.5", "User-Agent": "@cartesia/cartesia-js/2.2.5", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
+                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 contentType: "application/json",
                 requestType: "json",
                 timeoutMs: (requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.timeoutInSeconds) != null ? requestOptions.timeoutInSeconds * 1000 : 60000,
@@ -254,7 +254,7 @@ class Voices {
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
                 url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, `/voices/${encodeURIComponent(serializers.VoiceId.jsonOrThrow(id))}`),
                 method: "PATCH",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.5", "User-Agent": "@cartesia/cartesia-js/2.2.5", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
+                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 contentType: "application/json",
                 requestType: "json",
                 body: serializers.UpdateVoiceRequest.jsonOrThrow(request, { unrecognizedObjectKeys: "strip" }),
@@ -305,7 +305,7 @@ class Voices {
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
                 url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, `/voices/${encodeURIComponent(serializers.VoiceId.jsonOrThrow(id))}`),
                 method: "GET",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.5", "User-Agent": "@cartesia/cartesia-js/2.2.5", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
+                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 contentType: "application/json",
                 requestType: "json",
                 timeoutMs: (requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.timeoutInSeconds) != null ? requestOptions.timeoutInSeconds * 1000 : 60000,
@@ -364,7 +364,7 @@ class Voices {
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
                 url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, "/voices/localize"),
                 method: "POST",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.5", "User-Agent": "@cartesia/cartesia-js/2.2.5", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
+                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 contentType: "application/json",
                 requestType: "json",
                 body: serializers.LocalizeVoiceRequest.jsonOrThrow(request, { unrecognizedObjectKeys: "strip" }),
@@ -423,7 +423,7 @@ class Voices {
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
                 url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, "/voices/mix"),
                 method: "POST",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.5", "User-Agent": "@cartesia/cartesia-js/2.2.5", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
+                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 contentType: "application/json",
                 requestType: "json",
                 body: serializers.MixVoicesRequest.jsonOrThrow(request, { unrecognizedObjectKeys: "strip" }),
@@ -482,7 +482,7 @@ class Voices {
             const _response = yield ((_a = this._options.fetcher) !== null && _a !== void 0 ? _a : core.fetcher)({
                 url: (0, url_join_1.default)((_c = (_b = (yield core.Supplier.get(this._options.baseUrl))) !== null && _b !== void 0 ? _b : (yield core.Supplier.get(this._options.environment))) !== null && _c !== void 0 ? _c : environments.CartesiaEnvironment.Production, "/voices/"),
                 method: "POST",
-                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.5", "User-Agent": "@cartesia/cartesia-js/2.2.5", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
+                headers: Object.assign(Object.assign({ "Cartesia-Version": (_f = (_d = requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.cartesiaVersion) !== null && _d !== void 0 ? _d : (_e = this._options) === null || _e === void 0 ? void 0 : _e.cartesiaVersion) !== null && _f !== void 0 ? _f : "2024-06-10", "X-Fern-Language": "JavaScript", "X-Fern-SDK-Name": "@cartesia/cartesia-js", "X-Fern-SDK-Version": "2.2.7", "User-Agent": "@cartesia/cartesia-js/2.2.7", "X-Fern-Runtime": core.RUNTIME.type, "X-Fern-Runtime-Version": core.RUNTIME.version }, (yield this._getCustomAuthorizationHeaders())), requestOptions === null || requestOptions === void 0 ? void 0 : requestOptions.headers),
                 contentType: "application/json",
                 requestType: "json",
                 body: serializers.CreateVoiceRequest.jsonOrThrow(request, { unrecognizedObjectKeys: "strip" }),

package/dist/api/resources/voices/client/requests/CloneVoiceRequest.d.ts

@@ -20,34 +20,16 @@ import * as Cartesia from "../../../../index";
  *     }
  */
 export interface CloneVoiceRequest {
-    /**
-     * The name of the voice.
-     *
-     */
+    /** The name of the voice. */
     name: string;
-    /**
-     * A description for the voice.
-     *
-     */
+    /** A description for the voice. */
     description?: string;
-    /**
-     * The language of the voice.
-     *
-     */
+    /** The language of the voice. */
     language: Cartesia.SupportedLanguage;
-    /**
-     * Tradeoff between similarity and stability. Similarity clones sound more like the source clip, but may reproduce background noise. Stability clones always sound like a studio recording, but may not sound as similar to the source clip.
-     *
-     */
+    /** Tradeoff between similarity and stability. Similarity clones sound more like the source clip, but may reproduce background noise. Stability clones always sound like a studio recording, but may not sound as similar to the source clip. */
     mode: Cartesia.CloneMode;
-    /**
-     * Whether to apply AI enhancements to the clip to reduce background noise. This leads to cleaner generated speech at the cost of reduced similarity to the source clip.
-     *
-     */
+    /** Whether to apply AI enhancements to the clip to reduce background noise. This leads to cleaner generated speech at the cost of reduced similarity to the source clip. */
     enhance?: boolean;
-    /**
-     * Optional base voice ID that the cloned voice is derived from.
-     *
-     */
+    /** Optional base voice ID that the cloned voice is derived from. */
     baseVoiceId?: Cartesia.VoiceId;
 }

package/dist/api/resources/voices/types/LocalizeDialect.d.ts

@@ -7,18 +7,14 @@ import * as Cartesia from "../../../index";
  */
 export type LocalizeDialect = 
 /**
- * Only available when language is set to English (`en`). Options: Australian (`au`), Indian (`in`), Southern (`so`), British (`uk`), or American (`us`).
- *  */
+ * Only available when language is set to English (`en`). Options: Australian (`au`), Indian (`in`), Southern (`so`), British (`uk`), or American (`us`). */
 Cartesia.LocalizeEnglishDialect
 /**
- * Only available when language is set to Spanish (`es`). Options: Latin American (`mx`) and Peninsular (`pe`).
- *  */
+ * Only available when language is set to Spanish (`es`). Options: Latin American (`mx`) and Peninsular (`pe`). */
  | Cartesia.LocalizeSpanishDialect
 /**
- * Only available when language is set to Portuguese (`pt`). Options: Brazilian (`br`) and European Portuguese (`eu`).
- *  */
+ * Only available when language is set to Portuguese (`pt`). Options: Brazilian (`br`) and European Portuguese (`eu`). */
  | Cartesia.LocalizePortugueseDialect
 /**
- * Only available when language is set to French (`fr`). Options: Standard Parisian/Metropolitan (`eu`) and Canadian (`ca`).
- *  */
+ * Only available when language is set to French (`fr`). Options: Standard Parisian/Metropolitan (`eu`) and Canadian (`ca`). */
  | Cartesia.LocalizeFrenchDialect;

package/dist/core/fetcher/Fetcher.d.ts

@@ -6,14 +6,14 @@ export declare namespace Fetcher {
         method: string;
         contentType?: string;
         headers?: Record<string, string | undefined>;
-        queryParameters?: Record<string, string | string[] | object | object[]>;
+        queryParameters?: Record<string, string | string[] | object | object[] | null>;
         body?: unknown;
         timeoutMs?: number;
         maxRetries?: number;
         withCredentials?: boolean;
         abortSignal?: AbortSignal;
         requestType?: "json" | "file" | "bytes";
-        responseType?: "json" | "blob" | "sse" | "streaming" | "text" | "arraybuffer";
+        responseType?: "json" | "blob" | "sse" | "streaming" | "text" | "arrayBuffer";
         duplex?: "half";
     }
     type Error = FailedStatusCodeError | NonJsonError | TimeoutError | UnknownError;

package/dist/core/fetcher/Fetcher.js

@@ -11,6 +11,7 @@ var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, ge
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.fetcher = void 0;
 exports.fetcherImpl = fetcherImpl;
+const json_1 = require("../json");
 const createRequestUrl_1 = require("./createRequestUrl");
 const getFetchFn_1 = require("./getFetchFn");
 const getRequestBody_1 = require("./getRequestBody");
@@ -31,7 +32,7 @@ function fetcherImpl(args) {
             }
         }
         const url = (0, createRequestUrl_1.createRequestUrl)(args.url, args.queryParameters);
-        let requestBody = yield (0, getRequestBody_1.getRequestBody)({
+        const requestBody = yield (0, getRequestBody_1.getRequestBody)({
             body: args.body,
             type: args.requestType === "json" ? "json" : "other",
         });
@@ -40,7 +41,7 @@ function fetcherImpl(args) {
             const response = yield (0, requestWithRetries_1.requestWithRetries)(() => __awaiter(this, void 0, void 0, function* () {
                 return (0, makeRequest_1.makeRequest)(fetchFn, url, args.method, headers, requestBody, args.timeoutMs, args.abortSignal, args.withCredentials, args.duplex);
             }), args.maxRetries);
-            let responseBody = yield (0, getResponseBody_1.getResponseBody)(response, args.responseType);
+            const responseBody = yield (0, getResponseBody_1.getResponseBody)(response, args.responseType);
             if (response.status >= 200 && response.status < 400) {
                 return {
                     ok: true,
@@ -90,7 +91,7 @@ function fetcherImpl(args) {
                 ok: false,
                 error: {
                     reason: "unknown",
-                    errorMessage: JSON.stringify(error),
+                    errorMessage: (0, json_1.toJson)(error),
                 },
             };
         }

package/dist/core/fetcher/getResponseBody.js

@@ -16,15 +16,15 @@ function getResponseBody(response, responseType) {
         if (response.body != null && responseType === "blob") {
             return yield response.blob();
         }
+        else if (response.body != null && responseType === "arrayBuffer") {
+            return yield response.arrayBuffer();
+        }
         else if (response.body != null && responseType === "sse") {
             return response.body;
         }
         else if (response.body != null && responseType === "streaming") {
             return (0, chooseStreamWrapper_1.chooseStreamWrapper)(response.body);
         }
-        else if (response.body != null && responseType === "arraybuffer") {
-            return yield response.arrayBuffer();
-        }
         else if (response.body != null && responseType === "text") {
             return yield response.text();
         }

package/dist/index.d.ts

@@ -1,5 +1,7 @@
 export * as Cartesia from "./api";
 export { CartesiaClient } from "./wrapper/Client";
 export { WebPlayer } from "./wrapper/WebPlayer";
+export { default as SttWebsocket } from "./wrapper/SttWebsocket";
+export { StreamingSTTClient } from "./wrapper/StreamingSTTClient";
 export { CartesiaEnvironment } from "./environments";
 export { CartesiaError, CartesiaTimeoutError } from "./errors";