@cartesia/cartesia-js 2.2.4 → 2.2.7

package/api/resources/stt/types/TranscriptionWord.d.ts

@@ -0,0 +1,11 @@
+/**
+ * This file was auto-generated by Fern from our API Definition.
+ */
+export interface TranscriptionWord {
+    /** The transcribed word. */
+    word: string;
+    /** Start time of the word in seconds. */
+    start: number;
+    /** End time of the word in seconds. */
+    end: number;
+}

package/api/resources/stt/types/TranscriptionWord.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/api/resources/stt/types/index.d.ts

@@ -1,3 +1,5 @@
+export * from "./TimestampGranularity";
+export * from "./TranscriptionWord";
 export * from "./TranscriptionResponse";
 export * from "./StreamingTranscriptionResponse";
 export * from "./TranscriptMessage";

package/api/resources/stt/types/index.js

@@ -14,6 +14,8 @@ var __exportStar = (this && this.__exportStar) || function(m, exports) {
     for (var p in m) if (p !== "default" && !Object.prototype.hasOwnProperty.call(exports, p)) __createBinding(exports, m, p);
 };
 Object.defineProperty(exports, "__esModule", { value: true });
+__exportStar(require("./TimestampGranularity"), exports);
+__exportStar(require("./TranscriptionWord"), exports);
 __exportStar(require("./TranscriptionResponse"), exports);
 __exportStar(require("./StreamingTranscriptionResponse"), exports);
 __exportStar(require("./TranscriptMessage"), exports);

package/api/resources/tts/client/Client.d.ts

@@ -4,9 +4,12 @@
 import * as environments from "../../../../environments";
 import * as core from "../../../../core";
 import * as Cartesia from "../../../index";
+import * as stream from "stream";
 export declare namespace Tts {
     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";
@@ -21,13 +24,15 @@ export declare namespace Tts {
         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 Tts {
     protected readonly _options: Tts.Options;
     constructor(_options?: Tts.Options);
-    bytes(request: Cartesia.TtsRequest, requestOptions?: Tts.RequestOptions): Promise<ArrayBuffer>;
-    sse(request: Cartesia.TtsRequest, requestOptions?: Tts.RequestOptions): Promise<core.Stream<Cartesia.WebSocketResponse>>;
+    bytes(request: Cartesia.TtsRequest, requestOptions?: Tts.RequestOptions): Promise<stream.Readable>;
+    sse(request: Cartesia.TtssseRequest, requestOptions?: Tts.RequestOptions): Promise<core.Stream<Cartesia.WebSocketResponse>>;
     protected _getCustomAuthorizationHeaders(): Promise<{
         "X-API-Key": string | undefined;
     }>;

package/api/resources/tts/client/Client.js

@@ -60,15 +60,15 @@ class Tts {
     }
     bytes(request, requestOptions) {
         return __awaiter(this, void 0, void 0, function* () {
-            var _a, _b, _c, _d, _e;
+            var _a, _b, _c, _d, _e, _f;
             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, "/tts/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, "/tts/bytes"),
                 method: "POST",
-                headers: 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())),
+                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.TtsRequest.jsonOrThrow(request, { unrecognizedObjectKeys: "strip" }),
-                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,
@@ -99,14 +99,14 @@ class Tts {
     }
     sse(request, requestOptions) {
         return __awaiter(this, void 0, void 0, function* () {
-            var _a, _b, _c, _d, _e;
+            var _a, _b, _c, _d, _e, _f;
             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, "/tts/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, "/tts/sse"),
                 method: "POST",
-                headers: 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())),
+                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.TtsRequest.jsonOrThrow(request, { unrecognizedObjectKeys: "strip" }),
+                body: serializers.TtssseRequest.jsonOrThrow(request, { unrecognizedObjectKeys: "strip" }),
                 responseType: "sse",
                 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,

package/api/resources/tts/types/Controls.d.ts

@@ -4,5 +4,5 @@
 import * as Cartesia from "../../../index";
 export interface Controls {
     speed: Cartesia.Speed;
-    emotion: Cartesia.Emotion[];
+    emotion: Cartesia.EmotionDeprecated[];
 }

package/api/resources/tts/types/Emotion.d.ts

@@ -2,37 +2,6 @@
  * This file was auto-generated by Fern from our API Definition.
  */
 /**
- * An array of emotion:level tags.
- *
- * Supported emotions are: anger, positivity, surprise, sadness, and curiosity.
- *
- * Supported levels are: lowest, low, (omit), high, highest.
+ * Guide the emotion of the generated speech.
  */
-export type Emotion = "anger:lowest" | "anger:low" | "anger" | "anger:high" | "anger:highest" | "positivity:lowest" | "positivity:low" | "positivity" | "positivity:high" | "positivity:highest" | "surprise:lowest" | "surprise:low" | "surprise" | "surprise:high" | "surprise:highest" | "sadness:lowest" | "sadness:low" | "sadness" | "sadness:high" | "sadness:highest" | "curiosity:lowest" | "curiosity:low" | "curiosity" | "curiosity:high" | "curiosity:highest";
-export declare const Emotion: {
-    readonly AngerLowest: "anger:lowest";
-    readonly AngerLow: "anger:low";
-    readonly Anger: "anger";
-    readonly AngerHigh: "anger:high";
-    readonly AngerHighest: "anger:highest";
-    readonly PositivityLowest: "positivity:lowest";
-    readonly PositivityLow: "positivity:low";
-    readonly Positivity: "positivity";
-    readonly PositivityHigh: "positivity:high";
-    readonly PositivityHighest: "positivity:highest";
-    readonly SurpriseLowest: "surprise:lowest";
-    readonly SurpriseLow: "surprise:low";
-    readonly Surprise: "surprise";
-    readonly SurpriseHigh: "surprise:high";
-    readonly SurpriseHighest: "surprise:highest";
-    readonly SadnessLowest: "sadness:lowest";
-    readonly SadnessLow: "sadness:low";
-    readonly Sadness: "sadness";
-    readonly SadnessHigh: "sadness:high";
-    readonly SadnessHighest: "sadness:highest";
-    readonly CuriosityLowest: "curiosity:lowest";
-    readonly CuriosityLow: "curiosity:low";
-    readonly Curiosity: "curiosity";
-    readonly CuriosityHigh: "curiosity:high";
-    readonly CuriosityHighest: "curiosity:highest";
-};
+export type Emotion = string;

package/api/resources/tts/types/Emotion.js

@@ -3,31 +3,3 @@
  * This file was auto-generated by Fern from our API Definition.
  */
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.Emotion = void 0;
-exports.Emotion = {
-    AngerLowest: "anger:lowest",
-    AngerLow: "anger:low",
-    Anger: "anger",
-    AngerHigh: "anger:high",
-    AngerHighest: "anger:highest",
-    PositivityLowest: "positivity:lowest",
-    PositivityLow: "positivity:low",
-    Positivity: "positivity",
-    PositivityHigh: "positivity:high",
-    PositivityHighest: "positivity:highest",
-    SurpriseLowest: "surprise:lowest",
-    SurpriseLow: "surprise:low",
-    Surprise: "surprise",
-    SurpriseHigh: "surprise:high",
-    SurpriseHighest: "surprise:highest",
-    SadnessLowest: "sadness:lowest",
-    SadnessLow: "sadness:low",
-    Sadness: "sadness",
-    SadnessHigh: "sadness:high",
-    SadnessHighest: "sadness:highest",
-    CuriosityLowest: "curiosity:lowest",
-    CuriosityLow: "curiosity:low",
-    Curiosity: "curiosity",
-    CuriosityHigh: "curiosity:high",
-    CuriosityHighest: "curiosity:highest",
-};

package/api/resources/tts/types/EmotionDeprecated.d.ts

@@ -0,0 +1,38 @@
+/**
+ * This file was auto-generated by Fern from our API Definition.
+ */
+/**
+ * An array of emotion:level tags.
+ *
+ * Supported emotions are: anger, positivity, surprise, sadness, and curiosity.
+ *
+ * Supported levels are: lowest, low, (omit), high, highest.
+ */
+export type EmotionDeprecated = "anger:lowest" | "anger:low" | "anger" | "anger:high" | "anger:highest" | "positivity:lowest" | "positivity:low" | "positivity" | "positivity:high" | "positivity:highest" | "surprise:lowest" | "surprise:low" | "surprise" | "surprise:high" | "surprise:highest" | "sadness:lowest" | "sadness:low" | "sadness" | "sadness:high" | "sadness:highest" | "curiosity:lowest" | "curiosity:low" | "curiosity" | "curiosity:high" | "curiosity:highest";
+export declare const EmotionDeprecated: {
+    readonly AngerLowest: "anger:lowest";
+    readonly AngerLow: "anger:low";
+    readonly Anger: "anger";
+    readonly AngerHigh: "anger:high";
+    readonly AngerHighest: "anger:highest";
+    readonly PositivityLowest: "positivity:lowest";
+    readonly PositivityLow: "positivity:low";
+    readonly Positivity: "positivity";
+    readonly PositivityHigh: "positivity:high";
+    readonly PositivityHighest: "positivity:highest";
+    readonly SurpriseLowest: "surprise:lowest";
+    readonly SurpriseLow: "surprise:low";
+    readonly Surprise: "surprise";
+    readonly SurpriseHigh: "surprise:high";
+    readonly SurpriseHighest: "surprise:highest";
+    readonly SadnessLowest: "sadness:lowest";
+    readonly SadnessLow: "sadness:low";
+    readonly Sadness: "sadness";
+    readonly SadnessHigh: "sadness:high";
+    readonly SadnessHighest: "sadness:highest";
+    readonly CuriosityLowest: "curiosity:lowest";
+    readonly CuriosityLow: "curiosity:low";
+    readonly Curiosity: "curiosity";
+    readonly CuriosityHigh: "curiosity:high";
+    readonly CuriosityHighest: "curiosity:highest";
+};

package/api/resources/tts/types/EmotionDeprecated.js

@@ -0,0 +1,33 @@
+"use strict";
+/**
+ * This file was auto-generated by Fern from our API Definition.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.EmotionDeprecated = void 0;
+exports.EmotionDeprecated = {
+    AngerLowest: "anger:lowest",
+    AngerLow: "anger:low",
+    Anger: "anger",
+    AngerHigh: "anger:high",
+    AngerHighest: "anger:highest",
+    PositivityLowest: "positivity:lowest",
+    PositivityLow: "positivity:low",
+    Positivity: "positivity",
+    PositivityHigh: "positivity:high",
+    PositivityHighest: "positivity:highest",
+    SurpriseLowest: "surprise:lowest",
+    SurpriseLow: "surprise:low",
+    Surprise: "surprise",
+    SurpriseHigh: "surprise:high",
+    SurpriseHighest: "surprise:highest",
+    SadnessLowest: "sadness:lowest",
+    SadnessLow: "sadness:low",
+    Sadness: "sadness",
+    SadnessHigh: "sadness:high",
+    SadnessHighest: "sadness:highest",
+    CuriosityLowest: "curiosity:lowest",
+    CuriosityLow: "curiosity:low",
+    Curiosity: "curiosity",
+    CuriosityHigh: "curiosity:high",
+    CuriosityHighest: "curiosity:highest",
+};

package/api/resources/tts/types/GenerationConfig.d.ts

@@ -0,0 +1,15 @@
+/**
+ * This file was auto-generated by Fern from our API Definition.
+ */
+import * as Cartesia from "../../../index";
+/**
+ * Configure the various attributes of the generated speech. These controls only for `sonic-3` and have no effect on earlier models.
+ */
+export interface GenerationConfig {
+    /** Adjust the volume of the generated speech between 0.5x and 2.0x the original volume (default is 1.0x). Valid values are between 0.5 and 2.0 inclusive. */
+    volume?: number;
+    /** Adjust the speed of the generated speech between 0.6x and 1.5x the original speed (default is 1.0x). Valid values are between 0.6 and 1.5 inclusive. */
+    speed?: number;
+    /** Guide the emotion of the generated speech. */
+    emotion?: Cartesia.Emotion;
+}

package/api/resources/tts/types/GenerationConfig.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/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/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/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/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/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/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/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/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/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/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/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/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/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/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/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,