@livekit/agents-plugin-baseten 1.0.31

package/dist/tts.js

@@ -0,0 +1,141 @@
+import {
+  AudioByteStream,
+  shortuuid,
+  tts,
+  waitForAbort
+} from "@livekit/agents";
+const defaultTTSOptions = {
+  voice: "tara",
+  language: "en",
+  temperature: 0.6
+};
+class TTS extends tts.TTS {
+  opts;
+  label = "baseten.TTS";
+  abortController = new AbortController();
+  constructor(opts = {}) {
+    super(24e3, 1, { streaming: false });
+    const apiKey = opts.apiKey ?? process.env.BASETEN_API_KEY;
+    const modelEndpoint = opts.modelEndpoint ?? process.env.BASETEN_MODEL_ENDPOINT;
+    if (!apiKey) {
+      throw new Error(
+        "Baseten API key is required, either pass it as `apiKey` or set $BASETEN_API_KEY"
+      );
+    }
+    if (!modelEndpoint) {
+      throw new Error(
+        "Baseten model endpoint is required, either pass it as `modelEndpoint` or set $BASETEN_MODEL_ENDPOINT"
+      );
+    }
+    this.opts = {
+      ...defaultTTSOptions,
+      ...opts,
+      apiKey,
+      modelEndpoint
+    };
+  }
+  updateOptions(opts) {
+    this.opts = {
+      ...this.opts,
+      ...opts
+    };
+  }
+  /**
+   * Synthesize speech for a given piece of text.  Returns a `ChunkedStream`
+   * which will asynchronously fetch audio from Baseten and push frames into
+   * LiveKit's playback pipeline.  If you need to cancel synthesis you can
+   * call {@link ChunkedStream.stop} on the returned object.
+   */
+  synthesize(text, connOptions, abortSignal) {
+    return new ChunkedStream(this, text, this.opts, connOptions, abortSignal);
+  }
+  stream() {
+    throw new Error("Streaming is not supported on Baseten TTS");
+  }
+  async close() {
+    this.abortController.abort();
+  }
+}
+class ChunkedStream extends tts.ChunkedStream {
+  label = "baseten.ChunkedStream";
+  opts;
+  constructor(tts2, text, opts, connOptions, abortSignal) {
+    super(text, tts2, connOptions, abortSignal);
+    this.opts = opts;
+  }
+  /**
+   * Execute the synthesis request.  This method is automatically invoked
+   * by the base class when the stream starts.  It performs a POST request
+   * to the configured `modelEndpoint` with the input text and optional
+   * parameters.  Audio chunks are streamed as they arrive and transformed
+   * into a sequence of `AudioFrame` objects that are enqueued immediately
+   * for playback.
+   */
+  async run() {
+    const { apiKey, modelEndpoint, voice, language, temperature, maxTokens } = this.opts;
+    const payload = {
+      prompt: this.inputText
+    };
+    if (voice) payload.voice = voice;
+    if (language) payload.language = language;
+    if (temperature !== void 0) payload.temperature = temperature;
+    if (maxTokens !== void 0) payload.max_tokens = maxTokens;
+    const headers = {
+      Authorization: `Api-Key ${apiKey}`,
+      "Content-Type": "application/json"
+    };
+    const response = await fetch(modelEndpoint, {
+      method: "POST",
+      headers,
+      body: JSON.stringify(payload),
+      signal: this.abortSignal
+    });
+    if (!response.ok) {
+      let errText;
+      try {
+        errText = await response.text();
+      } catch {
+        errText = response.statusText;
+      }
+      throw new Error(`Baseten TTS request failed: ${response.status} ${errText}`);
+    }
+    if (!response.body) {
+      throw new Error("Response body is not available for streaming");
+    }
+    const requestId = shortuuid();
+    const audioByteStream = new AudioByteStream(24e3, 1);
+    const reader = response.body.getReader();
+    try {
+      let lastFrame;
+      const sendLastFrame = (segmentId, final) => {
+        if (lastFrame) {
+          this.queue.put({ requestId, segmentId, frame: lastFrame, final });
+          lastFrame = void 0;
+        }
+      };
+      const abortPromise = waitForAbort(this.abortSignal);
+      while (!this.abortSignal.aborted) {
+        const result = await Promise.race([reader.read(), abortPromise]);
+        if (result === void 0) break;
+        const { done, value } = result;
+        if (done) {
+          break;
+        }
+        const frames = audioByteStream.write(value.buffer);
+        for (const frame of frames) {
+          sendLastFrame(requestId, false);
+          lastFrame = frame;
+        }
+      }
+      sendLastFrame(requestId, true);
+    } finally {
+      reader.releaseLock();
+      this.queue.close();
+    }
+  }
+}
+export {
+  ChunkedStream,
+  TTS
+};
+//# sourceMappingURL=tts.js.map

package/dist/tts.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/tts.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport {\n  type APIConnectOptions,\n  AudioByteStream,\n  shortuuid,\n  tts,\n  waitForAbort,\n} from '@livekit/agents';\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { BasetenTTSOptions } from './types.js';\n\nconst defaultTTSOptions: Partial<BasetenTTSOptions> = {\n  voice: 'tara',\n  language: 'en',\n  temperature: 0.6,\n};\n\n/**\n * Baseten TTS implementation (streaming, 24kHz mono)\n */\nexport class TTS extends tts.TTS {\n  private opts: BasetenTTSOptions;\n  label = 'baseten.TTS';\n  private abortController = new AbortController();\n  constructor(opts: Partial<BasetenTTSOptions> = {}) {\n    /**\n     * Baseten audio is 24kHz mono.\n     * The Orpheus model generates audio chunks that are processed as they arrive,\n     * which reduces latency and improves agent responsiveness.\n     */\n    super(24000, 1, { streaming: false });\n\n    // Apply defaults and environment fallbacks.\n    const apiKey = opts.apiKey ?? process.env.BASETEN_API_KEY;\n    const modelEndpoint = opts.modelEndpoint ?? process.env.BASETEN_MODEL_ENDPOINT;\n\n    if (!apiKey) {\n      throw new Error(\n        'Baseten API key is required, either pass it as `apiKey` or set $BASETEN_API_KEY',\n      );\n    }\n    if (!modelEndpoint) {\n      throw new Error(\n        'Baseten model endpoint is required, either pass it as `modelEndpoint` or set $BASETEN_MODEL_ENDPOINT',\n      );\n    }\n\n    this.opts = {\n      ...defaultTTSOptions,\n      ...opts,\n      apiKey,\n      modelEndpoint,\n    } as BasetenTTSOptions;\n  }\n\n  updateOptions(opts: Partial<Omit<BasetenTTSOptions, 'apiKey' | 'modelEndpoint'>>) {\n    this.opts = {\n      ...this.opts,\n      ...opts,\n    } as BasetenTTSOptions;\n  }\n\n  /**\n   * Synthesize speech for a given piece of text.  Returns a `ChunkedStream`\n   * which will asynchronously fetch audio from Baseten and push frames into\n   * LiveKit's playback pipeline.  If you need to cancel synthesis you can\n   * call {@link ChunkedStream.stop} on the returned object.\n   */\n  synthesize(\n    text: string,\n    connOptions?: APIConnectOptions,\n    abortSignal?: AbortSignal,\n  ): ChunkedStream {\n    return new ChunkedStream(this, text, this.opts, connOptions, abortSignal);\n  }\n\n  stream(): tts.SynthesizeStream {\n    throw new Error('Streaming is not supported on Baseten TTS');\n  }\n\n  async close(): Promise<void> {\n    this.abortController.abort();\n  }\n}\n\n/**\n * Internal helper that performs the actual HTTP request and converts the\n * response into audio frames.  It inherits from `tts.ChunkedStream` to\n * integrate with LiveKit's event and cancellation framework.\n *\n * This implementation streams audio chunks as they arrive from the Baseten\n * model endpoint, processing them incrementally instead of waiting for the\n * complete response.\n */\nexport class ChunkedStream extends tts.ChunkedStream {\n  label = 'baseten.ChunkedStream';\n  private readonly opts: BasetenTTSOptions;\n\n  constructor(\n    tts: TTS,\n    text: string,\n    opts: BasetenTTSOptions,\n    connOptions?: APIConnectOptions,\n    abortSignal?: AbortSignal,\n  ) {\n    super(text, tts, connOptions, abortSignal);\n    this.opts = opts;\n  }\n\n  /**\n   * Execute the synthesis request.  This method is automatically invoked\n   * by the base class when the stream starts.  It performs a POST request\n   * to the configured `modelEndpoint` with the input text and optional\n   * parameters.  Audio chunks are streamed as they arrive and transformed\n   * into a sequence of `AudioFrame` objects that are enqueued immediately\n   * for playback.\n   */\n  protected async run() {\n    const { apiKey, modelEndpoint, voice, language, temperature, maxTokens } = this.opts;\n    const payload: Record<string, unknown> = {\n      prompt: this.inputText,\n    };\n    if (voice) payload.voice = voice;\n    if (language) payload.language = language;\n    if (temperature !== undefined) payload.temperature = temperature;\n    if (maxTokens !== undefined) payload.max_tokens = maxTokens;\n\n    const headers: Record<string, string> = {\n      Authorization: `Api-Key ${apiKey}`,\n      'Content-Type': 'application/json',\n    };\n\n    const response = await fetch(modelEndpoint, {\n      method: 'POST',\n      headers,\n      body: JSON.stringify(payload),\n      signal: this.abortSignal,\n    });\n\n    if (!response.ok) {\n      let errText: string;\n      try {\n        errText = await response.text();\n      } catch {\n        errText = response.statusText;\n      }\n      throw new Error(`Baseten TTS request failed: ${response.status} ${errText}`);\n    }\n\n    // Stream the response body as chunks arrive\n    if (!response.body) {\n      throw new Error('Response body is not available for streaming');\n    }\n\n    const requestId = shortuuid();\n    const audioByteStream = new AudioByteStream(24000, 1);\n    const reader = response.body.getReader();\n\n    try {\n      let lastFrame: AudioFrame | undefined;\n      const sendLastFrame = (segmentId: string, final: boolean) => {\n        if (lastFrame) {\n          this.queue.put({ requestId, segmentId, frame: lastFrame, final });\n          lastFrame = undefined;\n        }\n      };\n\n      // waitForAbort internally sets up an abort listener on the abort signal\n      // we need to put it outside loop to avoid constant re-registration of the listener\n      const abortPromise = waitForAbort(this.abortSignal);\n\n      while (!this.abortSignal.aborted) {\n        const result = await Promise.race([reader.read(), abortPromise]);\n\n        if (result === undefined) break; // aborted\n\n        const { done, value } = result;\n\n        if (done) {\n          break;\n        }\n\n        // Process the chunk and convert to audio frames\n        // Convert Uint8Array to ArrayBuffer for AudioByteStream\n        const frames = audioByteStream.write(value.buffer);\n\n        for (const frame of frames) {\n          sendLastFrame(requestId, false);\n          lastFrame = frame;\n        }\n      }\n\n      // Send the final frame\n      sendLastFrame(requestId, true);\n    } finally {\n      reader.releaseLock();\n      this.queue.close();\n    }\n  }\n}\n"],"mappings":"AAGA;AAAA,EAEE;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;AAIP,MAAM,oBAAgD;AAAA,EACpD,OAAO;AAAA,EACP,UAAU;AAAA,EACV,aAAa;AACf;AAKO,MAAM,YAAY,IAAI,IAAI;AAAA,EACvB;AAAA,EACR,QAAQ;AAAA,EACA,kBAAkB,IAAI,gBAAgB;AAAA,EAC9C,YAAY,OAAmC,CAAC,GAAG;AAMjD,UAAM,MAAO,GAAG,EAAE,WAAW,MAAM,CAAC;AAGpC,UAAM,SAAS,KAAK,UAAU,QAAQ,IAAI;AAC1C,UAAM,gBAAgB,KAAK,iBAAiB,QAAQ,IAAI;AAExD,QAAI,CAAC,QAAQ;AACX,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AACA,QAAI,CAAC,eAAe;AAClB,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAEA,SAAK,OAAO;AAAA,MACV,GAAG;AAAA,MACH,GAAG;AAAA,MACH;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA,EAEA,cAAc,MAAoE;AAChF,SAAK,OAAO;AAAA,MACV,GAAG,KAAK;AAAA,MACR,GAAG;AAAA,IACL;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,WACE,MACA,aACA,aACe;AACf,WAAO,IAAI,cAAc,MAAM,MAAM,KAAK,MAAM,aAAa,WAAW;AAAA,EAC1E;AAAA,EAEA,SAA+B;AAC7B,UAAM,IAAI,MAAM,2CAA2C;AAAA,EAC7D;AAAA,EAEA,MAAM,QAAuB;AAC3B,SAAK,gBAAgB,MAAM;AAAA,EAC7B;AACF;AAWO,MAAM,sBAAsB,IAAI,cAAc;AAAA,EACnD,QAAQ;AAAA,EACS;AAAA,EAEjB,YACEA,MACA,MACA,MACA,aACA,aACA;AACA,UAAM,MAAMA,MAAK,aAAa,WAAW;AACzC,SAAK,OAAO;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAUA,MAAgB,MAAM;AACpB,UAAM,EAAE,QAAQ,eAAe,OAAO,UAAU,aAAa,UAAU,IAAI,KAAK;AAChF,UAAM,UAAmC;AAAA,MACvC,QAAQ,KAAK;AAAA,IACf;AACA,QAAI,MAAO,SAAQ,QAAQ;AAC3B,QAAI,SAAU,SAAQ,WAAW;AACjC,QAAI,gBAAgB,OAAW,SAAQ,cAAc;AACrD,QAAI,cAAc,OAAW,SAAQ,aAAa;AAElD,UAAM,UAAkC;AAAA,MACtC,eAAe,WAAW,MAAM;AAAA,MAChC,gBAAgB;AAAA,IAClB;AAEA,UAAM,WAAW,MAAM,MAAM,eAAe;AAAA,MAC1C,QAAQ;AAAA,MACR;AAAA,MACA,MAAM,KAAK,UAAU,OAAO;AAAA,MAC5B,QAAQ,KAAK;AAAA,IACf,CAAC;AAED,QAAI,CAAC,SAAS,IAAI;AAChB,UAAI;AACJ,UAAI;AACF,kBAAU,MAAM,SAAS,KAAK;AAAA,MAChC,QAAQ;AACN,kBAAU,SAAS;AAAA,MACrB;AACA,YAAM,IAAI,MAAM,+BAA+B,SAAS,MAAM,IAAI,OAAO,EAAE;AAAA,IAC7E;AAGA,QAAI,CAAC,SAAS,MAAM;AAClB,YAAM,IAAI,MAAM,8CAA8C;AAAA,IAChE;AAEA,UAAM,YAAY,UAAU;AAC5B,UAAM,kBAAkB,IAAI,gBAAgB,MAAO,CAAC;AACpD,UAAM,SAAS,SAAS,KAAK,UAAU;AAEvC,QAAI;AACF,UAAI;AACJ,YAAM,gBAAgB,CAAC,WAAmB,UAAmB;AAC3D,YAAI,WAAW;AACb,eAAK,MAAM,IAAI,EAAE,WAAW,WAAW,OAAO,WAAW,MAAM,CAAC;AAChE,sBAAY;AAAA,QACd;AAAA,MACF;AAIA,YAAM,eAAe,aAAa,KAAK,WAAW;AAElD,aAAO,CAAC,KAAK,YAAY,SAAS;AAChC,cAAM,SAAS,MAAM,QAAQ,KAAK,CAAC,OAAO,KAAK,GAAG,YAAY,CAAC;AAE/D,YAAI,WAAW,OAAW;AAE1B,cAAM,EAAE,MAAM,MAAM,IAAI;AAExB,YAAI,MAAM;AACR;AAAA,QACF;AAIA,cAAM,SAAS,gBAAgB,MAAM,MAAM,MAAM;AAEjD,mBAAW,SAAS,QAAQ;AAC1B,wBAAc,WAAW,KAAK;AAC9B,sBAAY;AAAA,QACd;AAAA,MACF;AAGA,oBAAc,WAAW,IAAI;AAAA,IAC/B,UAAE;AACA,aAAO,YAAY;AACnB,WAAK,MAAM,MAAM;AAAA,IACnB;AAAA,EACF;AACF;","names":["tts"]}

package/dist/tts.test.cjs

@@ -0,0 +1,9 @@
+"use strict";
+var import_agents_plugins_test = require("@livekit/agents-plugins-test");
+var import_vitest = require("vitest");
+var import_stt = require("./stt.cjs");
+var import_tts = require("./tts.cjs");
+(0, import_vitest.describe)("Baseten", async () => {
+  await (0, import_agents_plugins_test.tts)(new import_tts.TTS(), new import_stt.STT(), { streaming: false });
+});
+//# sourceMappingURL=tts.test.cjs.map

package/dist/tts.test.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/tts.test.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { tts } from '@livekit/agents-plugins-test';\nimport { describe } from 'vitest';\nimport { STT } from './stt.js';\nimport { TTS } from './tts.js';\n\ndescribe('Baseten', async () => {\n  await tts(new TTS(), new STT(), { streaming: false });\n});\n"],"mappings":";AAGA,iCAAoB;AACpB,oBAAyB;AACzB,iBAAoB;AACpB,iBAAoB;AAAA,IAEpB,wBAAS,WAAW,YAAY;AAC9B,YAAM,gCAAI,IAAI,eAAI,GAAG,IAAI,eAAI,GAAG,EAAE,WAAW,MAAM,CAAC;AACtD,CAAC;","names":[]}

package/dist/tts.test.d.cts

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=tts.test.d.ts.map

package/dist/tts.test.d.ts

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=tts.test.d.ts.map

package/dist/tts.test.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"tts.test.d.ts","sourceRoot":"","sources":["../src/tts.test.ts"],"names":[],"mappings":""}

package/dist/tts.test.js

@@ -0,0 +1,8 @@
+import { tts } from "@livekit/agents-plugins-test";
+import { describe } from "vitest";
+import { STT } from "./stt.js";
+import { TTS } from "./tts.js";
+describe("Baseten", async () => {
+  await tts(new TTS(), new STT(), { streaming: false });
+});
+//# sourceMappingURL=tts.test.js.map

package/dist/tts.test.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/tts.test.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { tts } from '@livekit/agents-plugins-test';\nimport { describe } from 'vitest';\nimport { STT } from './stt.js';\nimport { TTS } from './tts.js';\n\ndescribe('Baseten', async () => {\n  await tts(new TTS(), new STT(), { streaming: false });\n});\n"],"mappings":"AAGA,SAAS,WAAW;AACpB,SAAS,gBAAgB;AACzB,SAAS,WAAW;AACpB,SAAS,WAAW;AAEpB,SAAS,WAAW,YAAY;AAC9B,QAAM,IAAI,IAAI,IAAI,GAAG,IAAI,IAAI,GAAG,EAAE,WAAW,MAAM,CAAC;AACtD,CAAC;","names":[]}

package/dist/types.cjs

@@ -0,0 +1,17 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+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 types_exports = {};
+module.exports = __toCommonJS(types_exports);
+//# sourceMappingURL=types.cjs.map

package/dist/types.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/types.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\n/**\n * Baseten plugin types and interfaces\n */\n\n/**\n * Options for configuring the Baseten LLM\n * Since Baseten provides an OpenAI-compatible API, these options\n * map to standard OpenAI parameters.\n */\nexport interface BasetenLLMOptions {\n  apiKey?: string;\n  model: string;\n  temperature?: number;\n  maxTokens?: number;\n  user?: string;\n  toolChoice?: 'none' | 'auto' | 'required' | { type: 'function'; function: { name: string } };\n  parallelToolCalls?: boolean;\n}\n\n/**\n * Options for configuring the Baseten STT service\n */\nexport interface BasetenSttOptions {\n  apiKey: string;\n  modelId: string;\n  environment?: string;\n  encoding?: string;\n  sampleRate?: number;\n  bufferSizeSeconds?: number;\n  vadThreshold?: number;\n  vadMinSilenceDurationMs?: number;\n  vadSpeechPadMs?: number;\n  enablePartialTranscripts?: boolean;\n  partialTranscriptIntervalS?: number;\n  finalTranscriptMaxDurationS?: number;\n  audioLanguage?: string;\n  prompt?: string;\n  languageDetectionOnly?: boolean;\n}\n\n/**\n * Options for configuring the Baseten TTS service\n */\nexport interface BasetenTTSOptions {\n  apiKey: string;\n  modelEndpoint: string;\n  voice?: string;\n  language?: string;\n  temperature?: number;\n  maxTokens?: number;\n}\n"],"mappings":";;;;;;;;;;;;;;AAAA;AAAA;","names":[]}

package/dist/types.d.cts

@@ -0,0 +1,54 @@
+/**
+ * Baseten plugin types and interfaces
+ */
+/**
+ * Options for configuring the Baseten LLM
+ * Since Baseten provides an OpenAI-compatible API, these options
+ * map to standard OpenAI parameters.
+ */
+export interface BasetenLLMOptions {
+    apiKey?: string;
+    model: string;
+    temperature?: number;
+    maxTokens?: number;
+    user?: string;
+    toolChoice?: 'none' | 'auto' | 'required' | {
+        type: 'function';
+        function: {
+            name: string;
+        };
+    };
+    parallelToolCalls?: boolean;
+}
+/**
+ * Options for configuring the Baseten STT service
+ */
+export interface BasetenSttOptions {
+    apiKey: string;
+    modelId: string;
+    environment?: string;
+    encoding?: string;
+    sampleRate?: number;
+    bufferSizeSeconds?: number;
+    vadThreshold?: number;
+    vadMinSilenceDurationMs?: number;
+    vadSpeechPadMs?: number;
+    enablePartialTranscripts?: boolean;
+    partialTranscriptIntervalS?: number;
+    finalTranscriptMaxDurationS?: number;
+    audioLanguage?: string;
+    prompt?: string;
+    languageDetectionOnly?: boolean;
+}
+/**
+ * Options for configuring the Baseten TTS service
+ */
+export interface BasetenTTSOptions {
+    apiKey: string;
+    modelEndpoint: string;
+    voice?: string;
+    language?: string;
+    temperature?: number;
+    maxTokens?: number;
+}
+//# sourceMappingURL=types.d.ts.map

package/dist/types.d.ts

@@ -0,0 +1,54 @@
+/**
+ * Baseten plugin types and interfaces
+ */
+/**
+ * Options for configuring the Baseten LLM
+ * Since Baseten provides an OpenAI-compatible API, these options
+ * map to standard OpenAI parameters.
+ */
+export interface BasetenLLMOptions {
+    apiKey?: string;
+    model: string;
+    temperature?: number;
+    maxTokens?: number;
+    user?: string;
+    toolChoice?: 'none' | 'auto' | 'required' | {
+        type: 'function';
+        function: {
+            name: string;
+        };
+    };
+    parallelToolCalls?: boolean;
+}
+/**
+ * Options for configuring the Baseten STT service
+ */
+export interface BasetenSttOptions {
+    apiKey: string;
+    modelId: string;
+    environment?: string;
+    encoding?: string;
+    sampleRate?: number;
+    bufferSizeSeconds?: number;
+    vadThreshold?: number;
+    vadMinSilenceDurationMs?: number;
+    vadSpeechPadMs?: number;
+    enablePartialTranscripts?: boolean;
+    partialTranscriptIntervalS?: number;
+    finalTranscriptMaxDurationS?: number;
+    audioLanguage?: string;
+    prompt?: string;
+    languageDetectionOnly?: boolean;
+}
+/**
+ * Options for configuring the Baseten TTS service
+ */
+export interface BasetenTTSOptions {
+    apiKey: string;
+    modelEndpoint: string;
+    voice?: string;
+    language?: string;
+    temperature?: number;
+    maxTokens?: number;
+}
+//# sourceMappingURL=types.d.ts.map

package/dist/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAIA;;GAEG;AAEH;;;;GAIG;AACH,MAAM,WAAW,iBAAiB;IAChC,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,KAAK,EAAE,MAAM,CAAC;IACd,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,UAAU,CAAC,EAAE,MAAM,GAAG,MAAM,GAAG,UAAU,GAAG;QAAE,IAAI,EAAE,UAAU,CAAC;QAAC,QAAQ,EAAE;YAAE,IAAI,EAAE,MAAM,CAAA;SAAE,CAAA;KAAE,CAAC;IAC7F,iBAAiB,CAAC,EAAE,OAAO,CAAC;CAC7B;AAED;;GAEG;AACH,MAAM,WAAW,iBAAiB;IAChC,MAAM,EAAE,MAAM,CAAC;IACf,OAAO,EAAE,MAAM,CAAC;IAChB,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,iBAAiB,CAAC,EAAE,MAAM,CAAC;IAC3B,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,uBAAuB,CAAC,EAAE,MAAM,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,wBAAwB,CAAC,EAAE,OAAO,CAAC;IACnC,0BAA0B,CAAC,EAAE,MAAM,CAAC;IACpC,2BAA2B,CAAC,EAAE,MAAM,CAAC;IACrC,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,qBAAqB,CAAC,EAAE,OAAO,CAAC;CACjC;AAED;;GAEG;AACH,MAAM,WAAW,iBAAiB;IAChC,MAAM,EAAE,MAAM,CAAC;IACf,aAAa,EAAE,MAAM,CAAC;IACtB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,SAAS,CAAC,EAAE,MAAM,CAAC;CACpB"}

package/dist/types.js

@@ -0,0 +1 @@
+//# sourceMappingURL=types.js.map

package/dist/types.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/package.json

@@ -0,0 +1,68 @@
+{
+  "name": "@livekit/agents-plugin-baseten",
+  "version": "1.0.31",
+  "description": "Baseten plugin for LiveKit Node Agents",
+  "main": "dist/index.js",
+  "require": "dist/index.cjs",
+  "types": "dist/index.d.ts",
+  "exports": {
+    "import": {
+      "types": "./dist/index.d.ts",
+      "default": "./dist/index.js"
+    },
+    "require": {
+      "types": "./dist/index.d.cts",
+      "default": "./dist/index.cjs"
+    }
+  },
+  "author": "LiveKit",
+  "type": "module",
+  "repository": "git@github.com:livekit/agents-js.git",
+  "license": "Apache-2.0",
+  "files": [
+    "dist",
+    "src",
+    "README.md"
+  ],
+  "keywords": [
+    "livekit",
+    "agents",
+    "baseten",
+    "llm",
+    "stt",
+    "tts",
+    "voice-ai"
+  ],
+  "devDependencies": {
+    "@livekit/rtc-node": "^0.13.22",
+    "@microsoft/api-extractor": "^7.35.0",
+    "@types/node": "^22.18.11",
+    "@types/ws": "^8.5.8",
+    "tsx": "^4.7.0",
+    "typescript": "^5.9.3",
+    "@livekit/agents-plugin-silero": "1.0.31",
+    "@livekit/agents-plugins-test": "1.0.31",
+    "@livekit/agents": "1.0.31"
+  },
+  "dependencies": {
+    "dotenv": "^17.2.3",
+    "openai": "^6.8.1",
+    "ws": "^8.14.2"
+  },
+  "peerDependencies": {
+    "@livekit/rtc-node": "^0.13.22",
+    "@livekit/agents": "1.0.31"
+  },
+  "scripts": {
+    "build": "tsup --onSuccess \"pnpm build:types\"",
+    "build:types": "tsc --declaration --emitDeclarationOnly && node ../../scripts/copyDeclarationOutput.js",
+    "clean": "rm -rf dist",
+    "clean:build": "pnpm clean && pnpm build",
+    "dev": "tsc --watch",
+    "typecheck": "tsc --noEmit",
+    "lint": "eslint \"src/**/*.ts\"",
+    "lint:fix": "eslint --fix \"src/**/*.ts\"",
+    "api:check": "api-extractor run --typescript-compiler-folder ../../node_modules/typescript",
+    "api:update": "api-extractor run --local --typescript-compiler-folder ../../node_modules/typescript --verbose"
+  }
+}

package/src/index.ts

@@ -0,0 +1,20 @@
+// SPDX-FileCopyrightText: 2024 LiveKit, Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+import { Plugin } from '@livekit/agents';
+
+export { LLM } from './llm.js';
+export { STT } from './stt.js';
+export { TTS, ChunkedStream } from './tts.js';
+export type { BasetenLLMOptions, BasetenSttOptions, BasetenTTSOptions } from './types.js';
+class BasetenPlugin extends Plugin {
+  constructor() {
+    super({
+      title: 'baseten',
+      version: '1.0.0',
+      package: 'livekit-plugin-baseten',
+    });
+  }
+}
+
+Plugin.registerPlugin(new BasetenPlugin());

package/src/llm.test.ts

@@ -0,0 +1,16 @@
+// SPDX-FileCopyrightText: 2024 LiveKit, Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+import { llm } from '@livekit/agents-plugins-test';
+import { describe } from 'vitest';
+import { LLM } from './llm.js';
+
+describe('Baseten', async () => {
+  await llm(
+    new LLM({
+      model: 'openai/gpt-4o-mini',
+      temperature: 0,
+    }),
+    false,
+  );
+});

package/src/llm.ts

@@ -0,0 +1,172 @@
+// SPDX-FileCopyrightText: 2024 LiveKit, Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+
+/**
+ * Baseten LLM plugin for LiveKit Agents
+ * Configures the OpenAI plugin to work with Baseten's OpenAI-compatible API
+ */
+import type { APIConnectOptions } from '@livekit/agents';
+import { DEFAULT_API_CONNECT_OPTIONS, inference, llm } from '@livekit/agents';
+import { OpenAI } from 'openai';
+import type { BasetenLLMOptions } from './types.js';
+
+export interface LLMOptions {
+  model: string;
+  apiKey?: string;
+  baseURL?: string;
+  user?: string;
+  temperature?: number;
+  client?: OpenAI;
+  toolChoice?: llm.ToolChoice;
+  parallelToolCalls?: boolean;
+  metadata?: Record<string, string>;
+  maxCompletionTokens?: number;
+  serviceTier?: string;
+  store?: boolean;
+  strictToolSchema?: boolean;
+}
+
+const defaultLLMOptions: LLMOptions = {
+  model: 'openai/gpt-4o-mini',
+  apiKey: process.env.OPENAI_API_KEY,
+  parallelToolCalls: true,
+  strictToolSchema: false,
+};
+
+export class OpenAILLM extends llm.LLM {
+  #opts: LLMOptions;
+  #client: OpenAI;
+  #providerFmt: llm.ProviderFormat;
+
+  constructor(
+    opts: Partial<LLMOptions> = defaultLLMOptions,
+    providerFmt: llm.ProviderFormat = 'openai',
+  ) {
+    super();
+
+    this.#opts = { ...defaultLLMOptions, ...opts };
+    this.#providerFmt = providerFmt;
+    if (this.#opts.apiKey === undefined) {
+      throw new Error('OpenAI API key is required, whether as an argument or as $OPENAI_API_KEY');
+    }
+
+    this.#client =
+      this.#opts.client ||
+      new OpenAI({
+        baseURL: opts.baseURL,
+        apiKey: opts.apiKey,
+      });
+  }
+
+  label(): string {
+    return 'openai.LLM';
+  }
+
+  get model(): string {
+    return this.#opts.model;
+  }
+
+  chat({
+    chatCtx,
+    toolCtx,
+    connOptions = DEFAULT_API_CONNECT_OPTIONS,
+    parallelToolCalls,
+    toolChoice,
+    extraKwargs,
+  }: {
+    chatCtx: llm.ChatContext;
+    toolCtx?: llm.ToolContext;
+    connOptions?: APIConnectOptions;
+    parallelToolCalls?: boolean;
+    toolChoice?: llm.ToolChoice;
+    extraKwargs?: Record<string, unknown>;
+  }): inference.LLMStream {
+    const extras: Record<string, unknown> = { ...extraKwargs };
+
+    if (this.#opts.metadata) {
+      extras.metadata = this.#opts.metadata;
+    }
+
+    if (this.#opts.user) {
+      extras.user = this.#opts.user;
+    }
+
+    if (this.#opts.maxCompletionTokens) {
+      extras.max_completion_tokens = this.#opts.maxCompletionTokens;
+    }
+
+    if (this.#opts.temperature) {
+      extras.temperature = this.#opts.temperature;
+    }
+
+    if (this.#opts.serviceTier) {
+      extras.service_tier = this.#opts.serviceTier;
+    }
+
+    if (this.#opts.store !== undefined) {
+      extras.store = this.#opts.store;
+    }
+
+    parallelToolCalls =
+      parallelToolCalls !== undefined ? parallelToolCalls : this.#opts.parallelToolCalls;
+    if (toolCtx && Object.keys(toolCtx).length > 0 && parallelToolCalls !== undefined) {
+      extras.parallel_tool_calls = parallelToolCalls;
+    }
+
+    toolChoice = toolChoice !== undefined ? toolChoice : this.#opts.toolChoice;
+    if (toolChoice) {
+      extras.tool_choice = toolChoice;
+    }
+
+    return new LLMStream(this as unknown as inference.LLM, {
+      model: this.#opts.model,
+      providerFmt: this.#providerFmt,
+      // eslint-disable-next-line @typescript-eslint/no-explicit-any
+      client: this.#client as any,
+      chatCtx,
+      toolCtx,
+      connOptions,
+      modelOptions: extras,
+      strictToolSchema: this.#opts.strictToolSchema || false,
+      gatewayOptions: undefined, // OpenAI plugin doesn't use gateway authentication
+    });
+  }
+}
+
+export class LLMStream extends inference.LLMStream {}
+
+export class LLM extends OpenAILLM {
+  constructor(opts: BasetenLLMOptions) {
+    const apiKey = opts.apiKey ?? process.env.BASETEN_API_KEY;
+    if (!apiKey) {
+      throw new Error(
+        'Baseten API key is required. Set BASETEN_API_KEY environment variable or pass apiKey in options.',
+      );
+    }
+
+    if (!opts.model) {
+      throw new Error(
+        'Model is required. Please specify a model name (e.g., "openai/gpt-4o-mini").',
+      );
+    }
+
+    const model = opts.model;
+
+    // Configure the OpenAI plugin with Baseten's endpoint
+    super({
+      model,
+      apiKey,
+      baseURL: 'https://inference.baseten.co/v1',
+      temperature: opts.temperature,
+      user: opts.user,
+      maxCompletionTokens: opts.maxTokens,
+      toolChoice: opts.toolChoice,
+      parallelToolCalls: opts.parallelToolCalls,
+    });
+  }
+
+  label(): string {
+    return 'baseten.LLM';
+  }
+}

package/src/stt.test.ts

@@ -0,0 +1,11 @@
+// SPDX-FileCopyrightText: 2024 LiveKit, Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+import { VAD } from '@livekit/agents-plugin-silero';
+import { stt } from '@livekit/agents-plugins-test';
+import { describe } from 'vitest';
+import { STT } from './stt.js';
+
+describe('Baseten', async () => {
+  await stt(new STT(), await VAD.load(), { streaming: true });
+});