@livekit/agents-plugin-cartesia 1.4.4 → 1.4.6

package/README.md

@@ -9,7 +9,7 @@ The Agents Framework is designed for building realtime, programmable
 participants that run on servers. Use it to create conversational, multi-modal
 voice agents that can see, hear, and understand.
 
-This package contains the Cartesia plugin, which allows for voice synthesis.
+This package contains the Cartesia plugin, which allows for voice synthesis and speech recognition.
 Refer to the [documentation](https://docs.livekit.io/agents/overview/) for
 information on how to use it, or browse the [API
 reference](https://docs.livekit.io/agents-js/modules/plugins_agents_plugin_cartesia.html).

package/dist/index.cjs

@@ -16,12 +16,13 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 var index_exports = {};
 module.exports = __toCommonJS(index_exports);
 var import_agents = require("@livekit/agents");
+__reExport(index_exports, require("./stt.cjs"), module.exports);
 __reExport(index_exports, require("./tts.cjs"), module.exports);
 class CartesiaPlugin extends import_agents.Plugin {
   constructor() {
     super({
       title: "cartesia",
-      version: "1.4.4",
+      version: "1.4.6",
       package: "@livekit/agents-plugin-cartesia"
     });
   }
@@ -29,6 +30,7 @@ class CartesiaPlugin extends import_agents.Plugin {
 import_agents.Plugin.registerPlugin(new CartesiaPlugin());
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
+  ...require("./stt.cjs"),
   ...require("./tts.cjs")
 });
 //# sourceMappingURL=index.cjs.map

package/dist/index.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2025 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { Plugin } from '@livekit/agents';\n\nexport * from './tts.js';\n\nclass CartesiaPlugin extends Plugin {\n  constructor() {\n    super({\n      title: 'cartesia',\n      version: __PACKAGE_VERSION__,\n      package: __PACKAGE_NAME__,\n    });\n  }\n}\n\nPlugin.registerPlugin(new CartesiaPlugin());\n"],"mappings":";;;;;;;;;;;;;;;AAAA;AAAA;AAGA,oBAAuB;AAEvB,0BAAc,qBALd;AAOA,MAAM,uBAAuB,qBAAO;AAAA,EAClC,cAAc;AACZ,UAAM;AAAA,MACJ,OAAO;AAAA,MACP,SAAS;AAAA,MACT,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACF;AAEA,qBAAO,eAAe,IAAI,eAAe,CAAC;","names":[]}
+{"version":3,"sources":["../src/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2025 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { Plugin } from '@livekit/agents';\n\nexport * from './stt.js';\nexport * from './tts.js';\n\nclass CartesiaPlugin extends Plugin {\n  constructor() {\n    super({\n      title: 'cartesia',\n      version: __PACKAGE_VERSION__,\n      package: __PACKAGE_NAME__,\n    });\n  }\n}\n\nPlugin.registerPlugin(new CartesiaPlugin());\n"],"mappings":";;;;;;;;;;;;;;;AAAA;AAAA;AAGA,oBAAuB;AAEvB,0BAAc,qBALd;AAMA,0BAAc,qBANd;AAQA,MAAM,uBAAuB,qBAAO;AAAA,EAClC,cAAc;AACZ,UAAM;AAAA,MACJ,OAAO;AAAA,MACP,SAAS;AAAA,MACT,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACF;AAEA,qBAAO,eAAe,IAAI,eAAe,CAAC;","names":[]}

package/dist/index.d.cts

@@ -1,2 +1,3 @@
+export * from './stt.js';
 export * from './tts.js';
 //# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts

@@ -1,2 +1,3 @@
+export * from './stt.js';
 export * from './tts.js';
 //# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAKA,cAAc,UAAU,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAKA,cAAc,UAAU,CAAC;AACzB,cAAc,UAAU,CAAC"}

package/dist/index.js

@@ -1,10 +1,11 @@
 import { Plugin } from "@livekit/agents";
+export * from "./stt.js";
 export * from "./tts.js";
 class CartesiaPlugin extends Plugin {
   constructor() {
     super({
       title: "cartesia",
-      version: "1.4.4",
+      version: "1.4.6",
       package: "@livekit/agents-plugin-cartesia"
     });
   }

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2025 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { Plugin } from '@livekit/agents';\n\nexport * from './tts.js';\n\nclass CartesiaPlugin extends Plugin {\n  constructor() {\n    super({\n      title: 'cartesia',\n      version: __PACKAGE_VERSION__,\n      package: __PACKAGE_NAME__,\n    });\n  }\n}\n\nPlugin.registerPlugin(new CartesiaPlugin());\n"],"mappings":"AAGA,SAAS,cAAc;AAEvB,cAAc;AAEd,MAAM,uBAAuB,OAAO;AAAA,EAClC,cAAc;AACZ,UAAM;AAAA,MACJ,OAAO;AAAA,MACP,SAAS;AAAA,MACT,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACF;AAEA,OAAO,eAAe,IAAI,eAAe,CAAC;","names":[]}
+{"version":3,"sources":["../src/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2025 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { Plugin } from '@livekit/agents';\n\nexport * from './stt.js';\nexport * from './tts.js';\n\nclass CartesiaPlugin extends Plugin {\n  constructor() {\n    super({\n      title: 'cartesia',\n      version: __PACKAGE_VERSION__,\n      package: __PACKAGE_NAME__,\n    });\n  }\n}\n\nPlugin.registerPlugin(new CartesiaPlugin());\n"],"mappings":"AAGA,SAAS,cAAc;AAEvB,cAAc;AACd,cAAc;AAEd,MAAM,uBAAuB,OAAO;AAAA,EAClC,cAAc;AACZ,UAAM;AAAA,MACJ,OAAO;AAAA,MACP,SAAS;AAAA,MACT,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACF;AAEA,OAAO,eAAe,IAAI,eAAe,CAAC;","names":[]}

package/dist/models.cjs

@@ -18,14 +18,17 @@ var __copyProps = (to, from, except, desc) => {
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 var models_exports = {};
 __export(models_exports, {
+  AUDIO_ENCODING: () => AUDIO_ENCODING,
   TTSDefaultVoiceId: () => TTSDefaultVoiceId,
   isSonic3: () => isSonic3
 });
 module.exports = __toCommonJS(models_exports);
+const AUDIO_ENCODING = "pcm_s16le";
 const TTSDefaultVoiceId = "f786b574-daa5-4673-aa0c-cbe3e8534c02";
 const isSonic3 = (model) => model.startsWith("sonic-3");
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
+  AUDIO_ENCODING,
   TTSDefaultVoiceId,
   isSonic3
 });

package/dist/models.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/models.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport type TTSModels =\n  | 'sonic'\n  | 'sonic-2'\n  | 'sonic-3'\n  | 'sonic-lite'\n  | 'sonic-preview'\n  | 'sonic-turbo';\n\nexport type TTSLanguages = 'en' | 'es' | 'fr' | 'de' | 'pt' | 'zh' | 'ja';\n\nexport const TTSDefaultVoiceId = 'f786b574-daa5-4673-aa0c-cbe3e8534c02';\n\nexport const isSonic3 = (model: string): boolean => model.startsWith('sonic-3');\n\nexport type TTSVoiceSpeed = 'fastest' | 'fast' | 'normal' | 'slow' | 'slowest';\n\nexport type TTSVoiceEmotion =\n  | 'anger:lowest'\n  | 'anger:low'\n  | 'anger'\n  | 'anger:high'\n  | 'anger:highest'\n  | 'positivity:lowest'\n  | 'positivity:low'\n  | 'positivity'\n  | 'positivity:high'\n  | 'positivity:highest'\n  | 'surprise:lowest'\n  | 'surprise:low'\n  | 'surprise'\n  | 'surprise:high'\n  | 'surprise:highest'\n  | 'sadness:lowest'\n  | 'sadness:low'\n  | 'sadness'\n  | 'sadness:high'\n  | 'sadness:highest'\n  | 'curiosity:lowest'\n  | 'curiosity:low'\n  | 'curiosity'\n  | 'curiosity:high'\n  | 'curiosity:highest';\n\nexport type TTSEncoding =\n  // XXX(nbsp): not yet supported\n  // | 'pcm_f32le'\n  // | 'pcm_mulaw'\n  // | 'pcm_alaw'\n  'pcm_s16le';\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAcO,MAAM,oBAAoB;AAE1B,MAAM,WAAW,CAAC,UAA2B,MAAM,WAAW,SAAS;","names":[]}
+{"version":3,"sources":["../src/models.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\n/**\n * LiveKit uses this encoding for all audio\n */\nexport const AUDIO_ENCODING = 'pcm_s16le';\n\n// ============================================================================\n//                                   TTS\n// ============================================================================\n\n/**\n * See [the docs](https://docs.cartesia.ai/build-with-cartesia/tts-models/latest) for all options.\n */\nexport type TTSModels =\n  | 'sonic'\n  | 'sonic-2'\n  | 'sonic-3'\n  | 'sonic-lite'\n  | 'sonic-preview'\n  | 'sonic-turbo';\n\n/**\n * See [the docs](https://docs.cartesia.ai/build-with-cartesia/tts-models/latest) for all options.\n */\nexport type TTSLanguages = 'en' | 'es' | 'fr' | 'de' | 'pt' | 'zh' | 'ja';\n\nexport const TTSDefaultVoiceId = 'f786b574-daa5-4673-aa0c-cbe3e8534c02';\n\nexport const isSonic3 = (model: string): boolean => model.startsWith('sonic-3');\n\nexport type TTSVoiceSpeed = 'fastest' | 'fast' | 'normal' | 'slow' | 'slowest';\n\nexport type TTSVoiceEmotion =\n  | 'anger:lowest'\n  | 'anger:low'\n  | 'anger'\n  | 'anger:high'\n  | 'anger:highest'\n  | 'positivity:lowest'\n  | 'positivity:low'\n  | 'positivity'\n  | 'positivity:high'\n  | 'positivity:highest'\n  | 'surprise:lowest'\n  | 'surprise:low'\n  | 'surprise'\n  | 'surprise:high'\n  | 'surprise:highest'\n  | 'sadness:lowest'\n  | 'sadness:low'\n  | 'sadness'\n  | 'sadness:high'\n  | 'sadness:highest'\n  | 'curiosity:lowest'\n  | 'curiosity:low'\n  | 'curiosity'\n  | 'curiosity:high'\n  | 'curiosity:highest';\n\n/**\n * @deprecated Encoding should not be parameterized. Only `pcm_s16le`is allowed. Prefer using {@link AUDIO_ENCODING}.\n */\nexport type TTSEncoding = 'pcm_s16le';\n\n// ============================================================================\n//                                   STT\n// ============================================================================\n\n/**\n * See [the docs](https://docs.cartesia.ai/build-with-cartesia/stt-models/latest) for all options.\n */\nexport type STTModel = 'ink-2';\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAOO,MAAM,iBAAiB;AAsBvB,MAAM,oBAAoB;AAE1B,MAAM,WAAW,CAAC,UAA2B,MAAM,WAAW,SAAS;","names":[]}

package/dist/models.d.cts

@@ -1,8 +1,25 @@
+/**
+ * LiveKit uses this encoding for all audio
+ */
+export declare const AUDIO_ENCODING = "pcm_s16le";
+/**
+ * See [the docs](https://docs.cartesia.ai/build-with-cartesia/tts-models/latest) for all options.
+ */
 export type TTSModels = 'sonic' | 'sonic-2' | 'sonic-3' | 'sonic-lite' | 'sonic-preview' | 'sonic-turbo';
+/**
+ * See [the docs](https://docs.cartesia.ai/build-with-cartesia/tts-models/latest) for all options.
+ */
 export type TTSLanguages = 'en' | 'es' | 'fr' | 'de' | 'pt' | 'zh' | 'ja';
 export declare const TTSDefaultVoiceId = "f786b574-daa5-4673-aa0c-cbe3e8534c02";
 export declare const isSonic3: (model: string) => boolean;
 export type TTSVoiceSpeed = 'fastest' | 'fast' | 'normal' | 'slow' | 'slowest';
 export type TTSVoiceEmotion = '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';
+/**
+ * @deprecated Encoding should not be parameterized. Only `pcm_s16le`is allowed. Prefer using {@link AUDIO_ENCODING}.
+ */
 export type TTSEncoding = 'pcm_s16le';
+/**
+ * See [the docs](https://docs.cartesia.ai/build-with-cartesia/stt-models/latest) for all options.
+ */
+export type STTModel = 'ink-2';
 //# sourceMappingURL=models.d.ts.map

package/dist/models.d.ts

@@ -1,8 +1,25 @@
+/**
+ * LiveKit uses this encoding for all audio
+ */
+export declare const AUDIO_ENCODING = "pcm_s16le";
+/**
+ * See [the docs](https://docs.cartesia.ai/build-with-cartesia/tts-models/latest) for all options.
+ */
 export type TTSModels = 'sonic' | 'sonic-2' | 'sonic-3' | 'sonic-lite' | 'sonic-preview' | 'sonic-turbo';
+/**
+ * See [the docs](https://docs.cartesia.ai/build-with-cartesia/tts-models/latest) for all options.
+ */
 export type TTSLanguages = 'en' | 'es' | 'fr' | 'de' | 'pt' | 'zh' | 'ja';
 export declare const TTSDefaultVoiceId = "f786b574-daa5-4673-aa0c-cbe3e8534c02";
 export declare const isSonic3: (model: string) => boolean;
 export type TTSVoiceSpeed = 'fastest' | 'fast' | 'normal' | 'slow' | 'slowest';
 export type TTSVoiceEmotion = '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';
+/**
+ * @deprecated Encoding should not be parameterized. Only `pcm_s16le`is allowed. Prefer using {@link AUDIO_ENCODING}.
+ */
 export type TTSEncoding = 'pcm_s16le';
+/**
+ * See [the docs](https://docs.cartesia.ai/build-with-cartesia/stt-models/latest) for all options.
+ */
+export type STTModel = 'ink-2';
 //# sourceMappingURL=models.d.ts.map

package/dist/models.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"models.d.ts","sourceRoot":"","sources":["../src/models.ts"],"names":[],"mappings":"AAIA,MAAM,MAAM,SAAS,GACjB,OAAO,GACP,SAAS,GACT,SAAS,GACT,YAAY,GACZ,eAAe,GACf,aAAa,CAAC;AAElB,MAAM,MAAM,YAAY,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,CAAC;AAE1E,eAAO,MAAM,iBAAiB,yCAAyC,CAAC;AAExE,eAAO,MAAM,QAAQ,GAAI,OAAO,MAAM,KAAG,OAAsC,CAAC;AAEhF,MAAM,MAAM,aAAa,GAAG,SAAS,GAAG,MAAM,GAAG,QAAQ,GAAG,MAAM,GAAG,SAAS,CAAC;AAE/E,MAAM,MAAM,eAAe,GACvB,cAAc,GACd,WAAW,GACX,OAAO,GACP,YAAY,GACZ,eAAe,GACf,mBAAmB,GACnB,gBAAgB,GAChB,YAAY,GACZ,iBAAiB,GACjB,oBAAoB,GACpB,iBAAiB,GACjB,cAAc,GACd,UAAU,GACV,eAAe,GACf,kBAAkB,GAClB,gBAAgB,GAChB,aAAa,GACb,SAAS,GACT,cAAc,GACd,iBAAiB,GACjB,kBAAkB,GAClB,eAAe,GACf,WAAW,GACX,gBAAgB,GAChB,mBAAmB,CAAC;AAExB,MAAM,MAAM,WAAW,GAKrB,WAAW,CAAC"}
+{"version":3,"file":"models.d.ts","sourceRoot":"","sources":["../src/models.ts"],"names":[],"mappings":"AAIA;;GAEG;AACH,eAAO,MAAM,cAAc,cAAc,CAAC;AAM1C;;GAEG;AACH,MAAM,MAAM,SAAS,GACjB,OAAO,GACP,SAAS,GACT,SAAS,GACT,YAAY,GACZ,eAAe,GACf,aAAa,CAAC;AAElB;;GAEG;AACH,MAAM,MAAM,YAAY,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,CAAC;AAE1E,eAAO,MAAM,iBAAiB,yCAAyC,CAAC;AAExE,eAAO,MAAM,QAAQ,GAAI,OAAO,MAAM,KAAG,OAAsC,CAAC;AAEhF,MAAM,MAAM,aAAa,GAAG,SAAS,GAAG,MAAM,GAAG,QAAQ,GAAG,MAAM,GAAG,SAAS,CAAC;AAE/E,MAAM,MAAM,eAAe,GACvB,cAAc,GACd,WAAW,GACX,OAAO,GACP,YAAY,GACZ,eAAe,GACf,mBAAmB,GACnB,gBAAgB,GAChB,YAAY,GACZ,iBAAiB,GACjB,oBAAoB,GACpB,iBAAiB,GACjB,cAAc,GACd,UAAU,GACV,eAAe,GACf,kBAAkB,GAClB,gBAAgB,GAChB,aAAa,GACb,SAAS,GACT,cAAc,GACd,iBAAiB,GACjB,kBAAkB,GAClB,eAAe,GACf,WAAW,GACX,gBAAgB,GAChB,mBAAmB,CAAC;AAExB;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,WAAW,CAAC;AAMtC;;GAEG;AACH,MAAM,MAAM,QAAQ,GAAG,OAAO,CAAC"}

package/dist/models.js

@@ -1,6 +1,8 @@
+const AUDIO_ENCODING = "pcm_s16le";
 const TTSDefaultVoiceId = "f786b574-daa5-4673-aa0c-cbe3e8534c02";
 const isSonic3 = (model) => model.startsWith("sonic-3");
 export {
+  AUDIO_ENCODING,
   TTSDefaultVoiceId,
   isSonic3
 };

package/dist/models.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/models.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport type TTSModels =\n  | 'sonic'\n  | 'sonic-2'\n  | 'sonic-3'\n  | 'sonic-lite'\n  | 'sonic-preview'\n  | 'sonic-turbo';\n\nexport type TTSLanguages = 'en' | 'es' | 'fr' | 'de' | 'pt' | 'zh' | 'ja';\n\nexport const TTSDefaultVoiceId = 'f786b574-daa5-4673-aa0c-cbe3e8534c02';\n\nexport const isSonic3 = (model: string): boolean => model.startsWith('sonic-3');\n\nexport type TTSVoiceSpeed = 'fastest' | 'fast' | 'normal' | 'slow' | 'slowest';\n\nexport type TTSVoiceEmotion =\n  | 'anger:lowest'\n  | 'anger:low'\n  | 'anger'\n  | 'anger:high'\n  | 'anger:highest'\n  | 'positivity:lowest'\n  | 'positivity:low'\n  | 'positivity'\n  | 'positivity:high'\n  | 'positivity:highest'\n  | 'surprise:lowest'\n  | 'surprise:low'\n  | 'surprise'\n  | 'surprise:high'\n  | 'surprise:highest'\n  | 'sadness:lowest'\n  | 'sadness:low'\n  | 'sadness'\n  | 'sadness:high'\n  | 'sadness:highest'\n  | 'curiosity:lowest'\n  | 'curiosity:low'\n  | 'curiosity'\n  | 'curiosity:high'\n  | 'curiosity:highest';\n\nexport type TTSEncoding =\n  // XXX(nbsp): not yet supported\n  // | 'pcm_f32le'\n  // | 'pcm_mulaw'\n  // | 'pcm_alaw'\n  'pcm_s16le';\n"],"mappings":"AAcO,MAAM,oBAAoB;AAE1B,MAAM,WAAW,CAAC,UAA2B,MAAM,WAAW,SAAS;","names":[]}
+{"version":3,"sources":["../src/models.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\n/**\n * LiveKit uses this encoding for all audio\n */\nexport const AUDIO_ENCODING = 'pcm_s16le';\n\n// ============================================================================\n//                                   TTS\n// ============================================================================\n\n/**\n * See [the docs](https://docs.cartesia.ai/build-with-cartesia/tts-models/latest) for all options.\n */\nexport type TTSModels =\n  | 'sonic'\n  | 'sonic-2'\n  | 'sonic-3'\n  | 'sonic-lite'\n  | 'sonic-preview'\n  | 'sonic-turbo';\n\n/**\n * See [the docs](https://docs.cartesia.ai/build-with-cartesia/tts-models/latest) for all options.\n */\nexport type TTSLanguages = 'en' | 'es' | 'fr' | 'de' | 'pt' | 'zh' | 'ja';\n\nexport const TTSDefaultVoiceId = 'f786b574-daa5-4673-aa0c-cbe3e8534c02';\n\nexport const isSonic3 = (model: string): boolean => model.startsWith('sonic-3');\n\nexport type TTSVoiceSpeed = 'fastest' | 'fast' | 'normal' | 'slow' | 'slowest';\n\nexport type TTSVoiceEmotion =\n  | 'anger:lowest'\n  | 'anger:low'\n  | 'anger'\n  | 'anger:high'\n  | 'anger:highest'\n  | 'positivity:lowest'\n  | 'positivity:low'\n  | 'positivity'\n  | 'positivity:high'\n  | 'positivity:highest'\n  | 'surprise:lowest'\n  | 'surprise:low'\n  | 'surprise'\n  | 'surprise:high'\n  | 'surprise:highest'\n  | 'sadness:lowest'\n  | 'sadness:low'\n  | 'sadness'\n  | 'sadness:high'\n  | 'sadness:highest'\n  | 'curiosity:lowest'\n  | 'curiosity:low'\n  | 'curiosity'\n  | 'curiosity:high'\n  | 'curiosity:highest';\n\n/**\n * @deprecated Encoding should not be parameterized. Only `pcm_s16le`is allowed. Prefer using {@link AUDIO_ENCODING}.\n */\nexport type TTSEncoding = 'pcm_s16le';\n\n// ============================================================================\n//                                   STT\n// ============================================================================\n\n/**\n * See [the docs](https://docs.cartesia.ai/build-with-cartesia/stt-models/latest) for all options.\n */\nexport type STTModel = 'ink-2';\n"],"mappings":"AAOO,MAAM,iBAAiB;AAsBvB,MAAM,oBAAoB;AAE1B,MAAM,WAAW,CAAC,UAA2B,MAAM,WAAW,SAAS;","names":[]}

package/dist/stt.cjs

@@ -0,0 +1,409 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var stt_exports = {};
+__export(stt_exports, {
+  STT: () => STT,
+  SpeechStream: () => SpeechStream
+});
+module.exports = __toCommonJS(stt_exports);
+var import_agents = require("@livekit/agents");
+var import_ws = require("ws");
+var import_models = require("./models.cjs");
+const AUTHORIZATION_HEADER = "X-API-Key";
+const VERSION_HEADER = "Cartesia-Version";
+const REQUEST_ID_HEADER = "cartesia-request-id";
+const API_VERSION = "2026-03-01";
+const DRAIN_TIMEOUT_MS = 5e3;
+const KEEPALIVE_INTERVAL_MS = 3e4;
+const defaultSTTOptions = {
+  model: "ink-2",
+  /** recommended default */
+  sampleRate: 16e3,
+  /** recommended default */
+  audioChunkDurationMS: 160,
+  baseUrl: "https://api.cartesia.ai"
+};
+function mergeSTTOptions(base, override) {
+  return {
+    apiKey: override.apiKey ?? base.apiKey,
+    baseUrl: override.baseUrl ?? base.baseUrl,
+    model: override.model ?? base.model,
+    sampleRate: override.sampleRate ?? base.sampleRate,
+    audioChunkDurationMS: override.audioChunkDurationMS ?? base.audioChunkDurationMS
+  };
+}
+class STT extends import_agents.stt.STT {
+  #opts;
+  constructor(opts = {}) {
+    super({
+      streaming: true,
+      interimResults: true,
+      alignedTranscript: false,
+      diarization: false
+    });
+    const apiKey = opts.apiKey ?? process.env.CARTESIA_API_KEY;
+    if (!apiKey) {
+      throw new Error(
+        "Cartesia API key is required, whether as an argument or as $CARTESIA_API_KEY"
+      );
+    }
+    this.#opts = mergeSTTOptions({ ...defaultSTTOptions, apiKey }, opts);
+  }
+  get label() {
+    return "cartesia.STT";
+  }
+  get model() {
+    return this.#opts.model;
+  }
+  get provider() {
+    return "Cartesia";
+  }
+  async _recognize() {
+    throw new Error("Cartesia STT does not support batch recognition, use stream() instead");
+  }
+  stream(options) {
+    return new SpeechStream(this, this.#opts, options == null ? void 0 : options.connOptions);
+  }
+}
+class SpeechStream extends import_agents.stt.SpeechStream {
+  #opts;
+  #logger = (0, import_agents.log)();
+  #ws = null;
+  #requestId;
+  #speaking = false;
+  #currentTranscript = "";
+  #speechDuration = 0;
+  #closingWs = false;
+  constructor(sttInstance, opts, connOptions) {
+    super(sttInstance, opts.sampleRate, connOptions);
+    this.#opts = { ...opts };
+  }
+  get label() {
+    return "cartesia.SpeechStream";
+  }
+  async run() {
+    var _a;
+    if (this.closed) {
+      this.close();
+      return;
+    }
+    this.#speaking = false;
+    this.#currentTranscript = "";
+    this.#speechDuration = 0;
+    this.#closingWs = false;
+    let keepaliveInterval;
+    const abortController = new AbortController();
+    try {
+      const url = this.#getCartesiaUrl();
+      this.#logger.debug(`Connecting to Cartesia STT: ${url}`);
+      const ws = new import_ws.WebSocket(url, {
+        headers: {
+          [AUTHORIZATION_HEADER]: this.#opts.apiKey,
+          [VERSION_HEADER]: API_VERSION
+        }
+      });
+      this.#ws = ws;
+      ws.once("upgrade", (response) => {
+        const headerValue = response.headers[REQUEST_ID_HEADER];
+        const requestId = Array.isArray(headerValue) ? headerValue[0] : headerValue;
+        if (requestId) {
+          this.#requestId = requestId;
+          this.#logger.debug({ cartesiaRequestId: requestId }, "Cartesia STT WebSocket connected");
+        }
+      });
+      await new Promise((resolve, reject) => {
+        const onOpen = () => {
+          ws.off("error", onError);
+          resolve();
+        };
+        const onError = (err) => {
+          ws.off("open", onOpen);
+          reject(new import_agents.APIConnectionError({ message: err.message, options: { retryable: true } }));
+        };
+        ws.once("open", onOpen);
+        ws.once("error", onError);
+      });
+      let firstError;
+      const stopPeer = (err) => {
+        if (firstError === void 0) firstError = err;
+        abortController.abort();
+        if (ws.readyState === import_ws.WebSocket.OPEN) ws.close();
+      };
+      const sendPromise = this.#sendTask(ws, abortController.signal).catch(stopPeer);
+      const recvPromise = this.#recvTask(ws).catch(stopPeer);
+      keepaliveInterval = setInterval(() => {
+        try {
+          if (ws.readyState === import_ws.WebSocket.OPEN) {
+            ws.ping();
+          }
+        } catch {
+        }
+      }, KEEPALIVE_INTERVAL_MS);
+      await Promise.all([sendPromise, recvPromise]);
+      if (firstError !== void 0) throw firstError;
+    } catch (error) {
+      this.#logger.error("Cartesia STT stream error", { error });
+      throw error;
+    } finally {
+      abortController.abort();
+      if (keepaliveInterval) clearInterval(keepaliveInterval);
+      if (((_a = this.#ws) == null ? void 0 : _a.readyState) === import_ws.WebSocket.OPEN) {
+        this.#ws.close();
+      }
+    }
+    this.close();
+  }
+  async #sendTask(ws, abortSignal) {
+    const samplesPerChunk = Math.floor(
+      this.#opts.sampleRate * this.#opts.audioChunkDurationMS / 1e3
+    );
+    const audioBstream = new import_agents.AudioByteStream(this.#opts.sampleRate, 1, samplesPerChunk);
+    let hasEnded = false;
+    const iterator = this.input[Symbol.asyncIterator]();
+    const abortPromise = (0, import_agents.waitForAbort)(abortSignal);
+    while (true) {
+      const result = await Promise.race([iterator.next(), abortPromise]);
+      if (result === void 0) return;
+      if (result.done) {
+        hasEnded = true;
+      } else {
+        const data = result.value;
+        const frames = [];
+        if (data === SpeechStream.FLUSH_SENTINEL) {
+          frames.push(...audioBstream.flush());
+        } else {
+          frames.push(...audioBstream.write(data.data));
+        }
+        for (const frame of frames) {
+          this.#speechDuration += (0, import_agents.calculateAudioDurationSeconds)(frame);
+          if (ws.readyState === import_ws.WebSocket.OPEN) {
+            ws.send(frame.data);
+          }
+        }
+      }
+      if (hasEnded) break;
+    }
+    if (ws.readyState === import_ws.WebSocket.OPEN) {
+      this.#closingWs = true;
+      this.#logger.debug("Sending close message to Cartesia STT");
+      ws.send(JSON.stringify({ type: "close" }));
+      const drainTimer = setTimeout(() => {
+        this.#logger.warn(
+          `Cartesia STT did not close within ${DRAIN_TIMEOUT_MS}ms after done; forcing close`
+        );
+        if (ws.readyState === import_ws.WebSocket.OPEN || ws.readyState === import_ws.WebSocket.CLOSING) {
+          ws.terminate();
+        }
+      }, DRAIN_TIMEOUT_MS);
+      ws.once("close", () => clearTimeout(drainTimer));
+    }
+  }
+  async #recvTask(ws) {
+    let pendingError = void 0;
+    return new Promise((resolve, reject) => {
+      const settle = (err) => {
+        if (err) reject(err);
+        else resolve();
+      };
+      ws.on("message", (data, isBinary) => {
+        if (isBinary) {
+          this.#logger.warn("Received unexpected binary message from Cartesia STT");
+          return;
+        }
+        let msg;
+        try {
+          msg = JSON.parse(data.toString());
+        } catch (error) {
+          this.#logger.error("Failed to parse Cartesia STT message", { error });
+          return;
+        }
+        try {
+          if (msg.type === "error") {
+            this.#logger.error("Cartesia sent an error", msg);
+            if (msg.status_code === void 0 || msg.status_code >= 500) {
+              pendingError = new import_agents.APIConnectionError({
+                message: msg.message || msg.title,
+                options: {
+                  retryable: true
+                }
+              });
+              ws.close();
+              return;
+            }
+          } else {
+            this.#processStreamEvent(msg);
+          }
+        } catch (error) {
+          this.#logger.error("Failed to process Cartesia STT message", { error });
+        }
+      });
+      ws.on("close", (code, reason) => {
+        if (this.#speaking) {
+          if (this.#currentTranscript) {
+            this.#sendTranscriptEvent(
+              import_agents.stt.SpeechEventType.FINAL_TRANSCRIPT,
+              this.#currentTranscript
+            );
+          }
+          if (!this.queue.closed)
+            this.queue.put({
+              type: import_agents.stt.SpeechEventType.END_OF_SPEECH,
+              requestId: this.#requestId
+            });
+          this.#logger.debug("Cartesia STT END_OF_SPEECH");
+          if (this.#speechDuration > 0 && !this.queue.closed) {
+            this.queue.put({
+              type: import_agents.stt.SpeechEventType.RECOGNITION_USAGE,
+              requestId: this.#requestId,
+              recognitionUsage: { audioDuration: this.#speechDuration }
+            });
+            this.#speechDuration = 0;
+          }
+          this.#speaking = false;
+          this.#currentTranscript = "";
+        }
+        if (pendingError) {
+          settle(pendingError);
+          return;
+        }
+        if (this.#closingWs || this.closed) {
+          settle();
+          return;
+        }
+        this.#logger.warn(`Cartesia STT WebSocket closed: ${code} ${reason.toString()}`);
+        settle(
+          new import_agents.APIConnectionError({
+            message: `Cartesia STT connection closed unexpectedly (code=${code})`,
+            options: { retryable: true }
+          })
+        );
+      });
+      ws.on("error", (err) => {
+        if (this.closed) {
+          settle();
+          return;
+        }
+        settle(new import_agents.APIConnectionError({ message: err.message, options: { retryable: true } }));
+      });
+    });
+  }
+  #processStreamEvent(data) {
+    if (data.request_id) {
+      this.#requestId = data.request_id;
+    }
+    switch (data.type) {
+      case "connected":
+        return;
+      case "turn.start": {
+        if (this.#speaking) return;
+        this.#speaking = true;
+        this.#currentTranscript = "";
+        if (!this.queue.closed)
+          this.queue.put({
+            type: import_agents.stt.SpeechEventType.START_OF_SPEECH,
+            requestId: this.#requestId
+          });
+        this.#logger.debug("Cartesia STT START_OF_SPEECH");
+        return;
+      }
+      case "turn.update": {
+        if (!this.#speaking) return;
+        const transcript = data.transcript;
+        if (!transcript) return;
+        if (this.#currentTranscript === transcript) return;
+        this.#currentTranscript = transcript;
+        this.#sendTranscriptEvent(import_agents.stt.SpeechEventType.INTERIM_TRANSCRIPT, transcript);
+        return;
+      }
+      case "turn.eager_end": {
+        if (!this.#speaking) return;
+        const transcript = data.transcript;
+        if (!transcript) return;
+        this.#currentTranscript = transcript;
+        this.#sendTranscriptEvent(import_agents.stt.SpeechEventType.PREFLIGHT_TRANSCRIPT, transcript);
+        return;
+      }
+      case "turn.resume":
+        if (!this.#speaking || !this.#currentTranscript) return;
+        this.#sendTranscriptEvent(import_agents.stt.SpeechEventType.INTERIM_TRANSCRIPT, this.#currentTranscript);
+        return;
+      case "turn.end": {
+        if (!this.#speaking) return;
+        const transcript = data.transcript;
+        this.#sendTranscriptEvent(import_agents.stt.SpeechEventType.FINAL_TRANSCRIPT, transcript);
+        this.#speaking = false;
+        if (!this.queue.closed)
+          this.queue.put({
+            type: import_agents.stt.SpeechEventType.END_OF_SPEECH,
+            requestId: this.#requestId
+          });
+        this.#logger.debug("Cartesia STT END_OF_SPEECH");
+        this.#currentTranscript = "";
+        if (this.#speechDuration > 0 && !this.queue.closed) {
+          this.queue.put({
+            type: import_agents.stt.SpeechEventType.RECOGNITION_USAGE,
+            requestId: this.#requestId,
+            recognitionUsage: { audioDuration: this.#speechDuration }
+          });
+          this.#speechDuration = 0;
+        }
+        return;
+      }
+      default:
+        this.#logger.warn("received unexpected message from Cartesia STT", { data });
+    }
+  }
+  #sendTranscriptEvent(eventType, transcript) {
+    if (this.queue.closed) return;
+    this.queue.put({
+      type: eventType,
+      requestId: this.#requestId,
+      alternatives: [
+        {
+          // Cartesia STT only supports English at this time.
+          language: (0, import_agents.asLanguageCode)("en"),
+          text: transcript,
+          startTime: 0,
+          endTime: 0,
+          confidence: 0
+        }
+      ]
+    });
+  }
+  #getCartesiaUrl() {
+    const params = new URLSearchParams({
+      model: this.#opts.model,
+      sample_rate: this.#opts.sampleRate.toString(),
+      encoding: import_models.AUDIO_ENCODING
+    });
+    const wsBase = this.#opts.baseUrl.replace(/^http/, "ws");
+    return `${wsBase}/stt/turns/websocket?${params.toString()}`;
+  }
+  close() {
+    var _a;
+    super.close();
+    (_a = this.#ws) == null ? void 0 : _a.close();
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  STT,
+  SpeechStream
+});
+//# sourceMappingURL=stt.cjs.map