@livekit/agents 0.4.5 → 0.5.0

package/dist/plugin.js.map

@@ -1 +1 @@
-{"version":3,"file":"plugin.js","sourceRoot":"","sources":["../src/plugin.ts"],"names":[],"mappings":"AAAA,6CAA6C;AAC7C,EAAE;AACF,sCAAsC;AAEtC,MAAM,OAAgB,MAAM;IAC1B,iBAAiB,GAAa,EAAE,CAAC;IACjC,MAAM,CAAS;IACf,QAAQ,CAAS;IAEjB,YAAY,KAAa,EAAE,OAAe;QACxC,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC;QACpB,IAAI,CAAC,QAAQ,GAAG,OAAO,CAAC;IAC1B,CAAC;IAEM,MAAM,CAAC,eAAe,CAAC,MAAc;QAC1C,MAAM,CAAC,iBAAiB,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IACxC,CAAC;IAID,IAAI,KAAK;QACP,OAAO,IAAI,CAAC,MAAM,CAAC;IACrB,CAAC;IAED,IAAI,OAAO;QACT,OAAO,IAAI,CAAC,QAAQ,CAAC;IACvB,CAAC;CACF"}
+{"version":3,"sources":["../src/plugin.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport abstract class Plugin {\n  registeredPlugins: Plugin[] = [];\n  #title: string;\n  #version: string;\n\n  constructor(title: string, version: string) {\n    this.#title = title;\n    this.#version = version;\n  }\n\n  public static registerPlugins(plugin: Plugin) {\n    plugin.registeredPlugins.push(plugin);\n  }\n\n  abstract downloadFiles(): void;\n\n  get title(): string {\n    return this.#title;\n  }\n\n  get version(): string {\n    return this.#version;\n  }\n}\n"],"mappings":"AAIO,MAAe,OAAO;AAAA,EAC3B,oBAA8B,CAAC;AAAA,EAC/B;AAAA,EACA;AAAA,EAEA,YAAY,OAAe,SAAiB;AAC1C,SAAK,SAAS;AACd,SAAK,WAAW;AAAA,EAClB;AAAA,EAEA,OAAc,gBAAgB,QAAgB;AAC5C,WAAO,kBAAkB,KAAK,MAAM;AAAA,EACtC;AAAA,EAIA,IAAI,QAAgB;AAClB,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,IAAI,UAAkB;AACpB,WAAO,KAAK;AAAA,EACd;AACF;","names":[]}

package/dist/stt/index.cjs

@@ -0,0 +1,38 @@
+"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: () => import_stt.STT,
+  SpeechEventType: () => import_stt.SpeechEventType,
+  SpeechStream: () => import_stt.SpeechStream,
+  StreamAdapter: () => import_stream_adapter.StreamAdapter,
+  StreamAdapterWrapper: () => import_stream_adapter.StreamAdapterWrapper
+});
+module.exports = __toCommonJS(stt_exports);
+var import_stt = require("./stt.cjs");
+var import_stream_adapter = require("./stream_adapter.cjs");
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  STT,
+  SpeechEventType,
+  SpeechStream,
+  StreamAdapter,
+  StreamAdapterWrapper
+});
+//# sourceMappingURL=index.cjs.map

package/dist/stt/index.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../src/stt/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport {\n  type SpeechEvent,\n  type SpeechData,\n  type STTCapabilities,\n  SpeechEventType,\n  STT,\n  SpeechStream,\n} from './stt.js';\nexport { StreamAdapter, StreamAdapterWrapper } from './stream_adapter.js';\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAIA,iBAOO;AACP,4BAAoD;","names":[]}

package/dist/stt/index.js

@@ -1,6 +1,14 @@
-// SPDX-FileCopyrightText: 2024 LiveKit, Inc.
-//
-// SPDX-License-Identifier: Apache-2.0
-export { SpeechEventType, STT, SpeechStream, } from './stt.js';
-export { StreamAdapter, StreamAdapterWrapper } from './stream_adapter.js';
+import {
+  SpeechEventType,
+  STT,
+  SpeechStream
+} from "./stt.js";
+import { StreamAdapter, StreamAdapterWrapper } from "./stream_adapter.js";
+export {
+  STT,
+  SpeechEventType,
+  SpeechStream,
+  StreamAdapter,
+  StreamAdapterWrapper
+};
 //# sourceMappingURL=index.js.map

package/dist/stt/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/stt/index.ts"],"names":[],"mappings":"AAAA,6CAA6C;AAC7C,EAAE;AACF,sCAAsC;AAEtC,OAAO,EAIL,eAAe,EACf,GAAG,EACH,YAAY,GACb,MAAM,UAAU,CAAC;AAClB,OAAO,EAAE,aAAa,EAAE,oBAAoB,EAAE,MAAM,qBAAqB,CAAC"}
+{"version":3,"sources":["../../src/stt/index.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\n\nexport {\n  type SpeechEvent,\n  type SpeechData,\n  type STTCapabilities,\n  SpeechEventType,\n  STT,\n  SpeechStream,\n} from './stt.js';\nexport { StreamAdapter, StreamAdapterWrapper } from './stream_adapter.js';\n"],"mappings":"AAIA;AAAA,EAIE;AAAA,EACA;AAAA,EACA;AAAA,OACK;AACP,SAAS,eAAe,4BAA4B;","names":[]}

package/dist/stt/stream_adapter.cjs

@@ -0,0 +1,87 @@
+"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 stream_adapter_exports = {};
+__export(stream_adapter_exports, {
+  StreamAdapter: () => StreamAdapter,
+  StreamAdapterWrapper: () => StreamAdapterWrapper
+});
+module.exports = __toCommonJS(stream_adapter_exports);
+var import_vad = require("../vad.cjs");
+var import_stt = require("./stt.cjs");
+class StreamAdapter extends import_stt.STT {
+  #stt;
+  #vad;
+  constructor(stt, vad) {
+    super({ streaming: true, interimResults: false });
+    this.#stt = stt;
+    this.#vad = vad;
+  }
+  recognize(frame) {
+    return this.#stt.recognize(frame);
+  }
+  stream() {
+    return new StreamAdapterWrapper(this.#stt, this.#vad);
+  }
+}
+class StreamAdapterWrapper extends import_stt.SpeechStream {
+  #stt;
+  #vadStream;
+  constructor(stt, vad) {
+    super();
+    this.#stt = stt;
+    this.#vadStream = vad.stream();
+    this.#run();
+  }
+  async #run() {
+    const forwardInput = async () => {
+      for await (const input of this.input) {
+        if (input === import_stt.SpeechStream.FLUSH_SENTINEL) {
+          this.#vadStream.flush();
+        } else {
+          this.#vadStream.pushFrame(input);
+        }
+      }
+      this.#vadStream.endInput();
+    };
+    const recognize = async () => {
+      for await (const ev of this.#vadStream) {
+        switch (ev.type) {
+          case import_vad.VADEventType.START_OF_SPEECH:
+            this.queue.put({ type: import_stt.SpeechEventType.START_OF_SPEECH });
+            break;
+          case import_vad.VADEventType.END_OF_SPEECH:
+            this.queue.put({ type: import_stt.SpeechEventType.END_OF_SPEECH });
+            const event = await this.#stt.recognize(ev.frames);
+            if (!event.alternatives[0].text) {
+              continue;
+            }
+            this.queue.put(event);
+            break;
+        }
+      }
+    };
+    Promise.all([forwardInput(), recognize()]);
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  StreamAdapter,
+  StreamAdapterWrapper
+});
+//# sourceMappingURL=stream_adapter.cjs.map

package/dist/stt/stream_adapter.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../src/stt/stream_adapter.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { VAD, VADStream } from '../vad.js';\nimport { VADEventType } from '../vad.js';\nimport type { SpeechEvent } from './stt.js';\nimport { STT, SpeechEventType, SpeechStream } from './stt.js';\n\nexport class StreamAdapter extends STT {\n  #stt: STT;\n  #vad: VAD;\n\n  constructor(stt: STT, vad: VAD) {\n    super({ streaming: true, interimResults: false });\n    this.#stt = stt;\n    this.#vad = vad;\n  }\n\n  recognize(frame: AudioFrame): Promise<SpeechEvent> {\n    return this.#stt.recognize(frame);\n  }\n\n  stream(): StreamAdapterWrapper {\n    return new StreamAdapterWrapper(this.#stt, this.#vad);\n  }\n}\n\nexport class StreamAdapterWrapper extends SpeechStream {\n  #stt: STT;\n  #vadStream: VADStream;\n\n  constructor(stt: STT, vad: VAD) {\n    super();\n    this.#stt = stt;\n    this.#vadStream = vad.stream();\n\n    this.#run();\n  }\n\n  async #run() {\n    const forwardInput = async () => {\n      for await (const input of this.input) {\n        if (input === SpeechStream.FLUSH_SENTINEL) {\n          this.#vadStream.flush();\n        } else {\n          this.#vadStream.pushFrame(input);\n        }\n      }\n      this.#vadStream.endInput();\n    };\n\n    const recognize = async () => {\n      for await (const ev of this.#vadStream) {\n        switch (ev.type) {\n          case VADEventType.START_OF_SPEECH:\n            this.queue.put({ type: SpeechEventType.START_OF_SPEECH });\n            break;\n          case VADEventType.END_OF_SPEECH:\n            this.queue.put({ type: SpeechEventType.END_OF_SPEECH });\n\n            const event = await this.#stt.recognize(ev.frames);\n            if (!event.alternatives![0].text) {\n              continue;\n            }\n\n            this.queue.put(event);\n            break;\n        }\n      }\n    };\n\n    Promise.all([forwardInput(), recognize()]);\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAKA,iBAA6B;AAE7B,iBAAmD;AAE5C,MAAM,sBAAsB,eAAI;AAAA,EACrC;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,KAAU;AAC9B,UAAM,EAAE,WAAW,MAAM,gBAAgB,MAAM,CAAC;AAChD,SAAK,OAAO;AACZ,SAAK,OAAO;AAAA,EACd;AAAA,EAEA,UAAU,OAAyC;AACjD,WAAO,KAAK,KAAK,UAAU,KAAK;AAAA,EAClC;AAAA,EAEA,SAA+B;AAC7B,WAAO,IAAI,qBAAqB,KAAK,MAAM,KAAK,IAAI;AAAA,EACtD;AACF;AAEO,MAAM,6BAA6B,wBAAa;AAAA,EACrD;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,KAAU;AAC9B,UAAM;AACN,SAAK,OAAO;AACZ,SAAK,aAAa,IAAI,OAAO;AAE7B,SAAK,KAAK;AAAA,EACZ;AAAA,EAEA,MAAM,OAAO;AACX,UAAM,eAAe,YAAY;AAC/B,uBAAiB,SAAS,KAAK,OAAO;AACpC,YAAI,UAAU,wBAAa,gBAAgB;AACzC,eAAK,WAAW,MAAM;AAAA,QACxB,OAAO;AACL,eAAK,WAAW,UAAU,KAAK;AAAA,QACjC;AAAA,MACF;AACA,WAAK,WAAW,SAAS;AAAA,IAC3B;AAEA,UAAM,YAAY,YAAY;AAC5B,uBAAiB,MAAM,KAAK,YAAY;AACtC,gBAAQ,GAAG,MAAM;AAAA,UACf,KAAK,wBAAa;AAChB,iBAAK,MAAM,IAAI,EAAE,MAAM,2BAAgB,gBAAgB,CAAC;AACxD;AAAA,UACF,KAAK,wBAAa;AAChB,iBAAK,MAAM,IAAI,EAAE,MAAM,2BAAgB,cAAc,CAAC;AAEtD,kBAAM,QAAQ,MAAM,KAAK,KAAK,UAAU,GAAG,MAAM;AACjD,gBAAI,CAAC,MAAM,aAAc,CAAC,EAAE,MAAM;AAChC;AAAA,YACF;AAEA,iBAAK,MAAM,IAAI,KAAK;AACpB;AAAA,QACJ;AAAA,MACF;AAAA,IACF;AAEA,YAAQ,IAAI,CAAC,aAAa,GAAG,UAAU,CAAC,CAAC;AAAA,EAC3C;AACF;","names":[]}

package/dist/stt/stream_adapter.js

@@ -1,59 +1,62 @@
-import { VADEventType } from '../vad.js';
-import { STT, SpeechEventType, SpeechStream } from './stt.js';
-export class StreamAdapter extends STT {
-    #stt;
-    #vad;
-    constructor(stt, vad) {
-        super({ streaming: true, interimResults: false });
-        this.#stt = stt;
-        this.#vad = vad;
-    }
-    recognize(frame) {
-        return this.#stt.recognize(frame);
-    }
-    stream() {
-        return new StreamAdapterWrapper(this.#stt, this.#vad);
-    }
+import { VADEventType } from "../vad.js";
+import { STT, SpeechEventType, SpeechStream } from "./stt.js";
+class StreamAdapter extends STT {
+  #stt;
+  #vad;
+  constructor(stt, vad) {
+    super({ streaming: true, interimResults: false });
+    this.#stt = stt;
+    this.#vad = vad;
+  }
+  recognize(frame) {
+    return this.#stt.recognize(frame);
+  }
+  stream() {
+    return new StreamAdapterWrapper(this.#stt, this.#vad);
+  }
 }
-export class StreamAdapterWrapper extends SpeechStream {
-    #stt;
-    #vadStream;
-    constructor(stt, vad) {
-        super();
-        this.#stt = stt;
-        this.#vadStream = vad.stream();
-        this.#run();
-    }
-    async #run() {
-        const forwardInput = async () => {
-            for await (const input of this.input) {
-                if (input === SpeechStream.FLUSH_SENTINEL) {
-                    this.#vadStream.flush();
-                }
-                else {
-                    this.#vadStream.pushFrame(input);
-                }
+class StreamAdapterWrapper extends SpeechStream {
+  #stt;
+  #vadStream;
+  constructor(stt, vad) {
+    super();
+    this.#stt = stt;
+    this.#vadStream = vad.stream();
+    this.#run();
+  }
+  async #run() {
+    const forwardInput = async () => {
+      for await (const input of this.input) {
+        if (input === SpeechStream.FLUSH_SENTINEL) {
+          this.#vadStream.flush();
+        } else {
+          this.#vadStream.pushFrame(input);
+        }
+      }
+      this.#vadStream.endInput();
+    };
+    const recognize = async () => {
+      for await (const ev of this.#vadStream) {
+        switch (ev.type) {
+          case VADEventType.START_OF_SPEECH:
+            this.queue.put({ type: SpeechEventType.START_OF_SPEECH });
+            break;
+          case VADEventType.END_OF_SPEECH:
+            this.queue.put({ type: SpeechEventType.END_OF_SPEECH });
+            const event = await this.#stt.recognize(ev.frames);
+            if (!event.alternatives[0].text) {
+              continue;
             }
-            this.#vadStream.endInput();
-        };
-        const recognize = async () => {
-            for await (const ev of this.#vadStream) {
-                switch (ev.type) {
-                    case VADEventType.START_OF_SPEECH:
-                        this.queue.put({ type: SpeechEventType.START_OF_SPEECH });
-                        break;
-                    case VADEventType.END_OF_SPEECH:
-                        this.queue.put({ type: SpeechEventType.END_OF_SPEECH });
-                        const event = await this.#stt.recognize(ev.frames);
-                        if (!event.alternatives[0].text) {
-                            continue;
-                        }
-                        this.queue.put(event);
-                        break;
-                }
-            }
-        };
-        Promise.all([forwardInput(), recognize()]);
-    }
+            this.queue.put(event);
+            break;
+        }
+      }
+    };
+    Promise.all([forwardInput(), recognize()]);
+  }
 }
+export {
+  StreamAdapter,
+  StreamAdapterWrapper
+};
 //# sourceMappingURL=stream_adapter.js.map

package/dist/stt/stream_adapter.js.map

@@ -1 +1 @@
-{"version":3,"file":"stream_adapter.js","sourceRoot":"","sources":["../../src/stt/stream_adapter.ts"],"names":[],"mappings":"AAKA,OAAO,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AAEzC,OAAO,EAAE,GAAG,EAAE,eAAe,EAAE,YAAY,EAAE,MAAM,UAAU,CAAC;AAE9D,MAAM,OAAO,aAAc,SAAQ,GAAG;IACpC,IAAI,CAAM;IACV,IAAI,CAAM;IAEV,YAAY,GAAQ,EAAE,GAAQ;QAC5B,KAAK,CAAC,EAAE,SAAS,EAAE,IAAI,EAAE,cAAc,EAAE,KAAK,EAAE,CAAC,CAAC;QAClD,IAAI,CAAC,IAAI,GAAG,GAAG,CAAC;QAChB,IAAI,CAAC,IAAI,GAAG,GAAG,CAAC;IAClB,CAAC;IAED,SAAS,CAAC,KAAiB;QACzB,OAAO,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IACpC,CAAC;IAED,MAAM;QACJ,OAAO,IAAI,oBAAoB,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,CAAC,CAAC;IACxD,CAAC;CACF;AAED,MAAM,OAAO,oBAAqB,SAAQ,YAAY;IACpD,IAAI,CAAM;IACV,UAAU,CAAY;IAEtB,YAAY,GAAQ,EAAE,GAAQ;QAC5B,KAAK,EAAE,CAAC;QACR,IAAI,CAAC,IAAI,GAAG,GAAG,CAAC;QAChB,IAAI,CAAC,UAAU,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC;QAE/B,IAAI,CAAC,IAAI,EAAE,CAAC;IACd,CAAC;IAED,KAAK,CAAC,IAAI;QACR,MAAM,YAAY,GAAG,KAAK,IAAI,EAAE;YAC9B,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,EAAE,CAAC;gBACrC,IAAI,KAAK,KAAK,YAAY,CAAC,cAAc,EAAE,CAAC;oBAC1C,IAAI,CAAC,UAAU,CAAC,KAAK,EAAE,CAAC;gBAC1B,CAAC;qBAAM,CAAC;oBACN,IAAI,CAAC,UAAU,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;gBACnC,CAAC;YACH,CAAC;YACD,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC;QAC7B,CAAC,CAAC;QAEF,MAAM,SAAS,GAAG,KAAK,IAAI,EAAE;YAC3B,IAAI,KAAK,EAAE,MAAM,EAAE,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;gBACvC,QAAQ,EAAE,CAAC,IAAI,EAAE,CAAC;oBAChB,KAAK,YAAY,CAAC,eAAe;wBAC/B,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,IAAI,EAAE,eAAe,CAAC,eAAe,EAAE,CAAC,CAAC;wBAC1D,MAAM;oBACR,KAAK,YAAY,CAAC,aAAa;wBAC7B,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,IAAI,EAAE,eAAe,CAAC,aAAa,EAAE,CAAC,CAAC;wBAExD,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC,MAAM,CAAC,CAAC;wBACnD,IAAI,CAAC,KAAK,CAAC,YAAa,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;4BACjC,SAAS;wBACX,CAAC;wBAED,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;wBACtB,MAAM;gBACV,CAAC;YACH,CAAC;QACH,CAAC,CAAC;QAEF,OAAO,CAAC,GAAG,CAAC,CAAC,YAAY,EAAE,EAAE,SAAS,EAAE,CAAC,CAAC,CAAC;IAC7C,CAAC;CACF"}
+{"version":3,"sources":["../../src/stt/stream_adapter.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { VAD, VADStream } from '../vad.js';\nimport { VADEventType } from '../vad.js';\nimport type { SpeechEvent } from './stt.js';\nimport { STT, SpeechEventType, SpeechStream } from './stt.js';\n\nexport class StreamAdapter extends STT {\n  #stt: STT;\n  #vad: VAD;\n\n  constructor(stt: STT, vad: VAD) {\n    super({ streaming: true, interimResults: false });\n    this.#stt = stt;\n    this.#vad = vad;\n  }\n\n  recognize(frame: AudioFrame): Promise<SpeechEvent> {\n    return this.#stt.recognize(frame);\n  }\n\n  stream(): StreamAdapterWrapper {\n    return new StreamAdapterWrapper(this.#stt, this.#vad);\n  }\n}\n\nexport class StreamAdapterWrapper extends SpeechStream {\n  #stt: STT;\n  #vadStream: VADStream;\n\n  constructor(stt: STT, vad: VAD) {\n    super();\n    this.#stt = stt;\n    this.#vadStream = vad.stream();\n\n    this.#run();\n  }\n\n  async #run() {\n    const forwardInput = async () => {\n      for await (const input of this.input) {\n        if (input === SpeechStream.FLUSH_SENTINEL) {\n          this.#vadStream.flush();\n        } else {\n          this.#vadStream.pushFrame(input);\n        }\n      }\n      this.#vadStream.endInput();\n    };\n\n    const recognize = async () => {\n      for await (const ev of this.#vadStream) {\n        switch (ev.type) {\n          case VADEventType.START_OF_SPEECH:\n            this.queue.put({ type: SpeechEventType.START_OF_SPEECH });\n            break;\n          case VADEventType.END_OF_SPEECH:\n            this.queue.put({ type: SpeechEventType.END_OF_SPEECH });\n\n            const event = await this.#stt.recognize(ev.frames);\n            if (!event.alternatives![0].text) {\n              continue;\n            }\n\n            this.queue.put(event);\n            break;\n        }\n      }\n    };\n\n    Promise.all([forwardInput(), recognize()]);\n  }\n}\n"],"mappings":"AAKA,SAAS,oBAAoB;AAE7B,SAAS,KAAK,iBAAiB,oBAAoB;AAE5C,MAAM,sBAAsB,IAAI;AAAA,EACrC;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,KAAU;AAC9B,UAAM,EAAE,WAAW,MAAM,gBAAgB,MAAM,CAAC;AAChD,SAAK,OAAO;AACZ,SAAK,OAAO;AAAA,EACd;AAAA,EAEA,UAAU,OAAyC;AACjD,WAAO,KAAK,KAAK,UAAU,KAAK;AAAA,EAClC;AAAA,EAEA,SAA+B;AAC7B,WAAO,IAAI,qBAAqB,KAAK,MAAM,KAAK,IAAI;AAAA,EACtD;AACF;AAEO,MAAM,6BAA6B,aAAa;AAAA,EACrD;AAAA,EACA;AAAA,EAEA,YAAY,KAAU,KAAU;AAC9B,UAAM;AACN,SAAK,OAAO;AACZ,SAAK,aAAa,IAAI,OAAO;AAE7B,SAAK,KAAK;AAAA,EACZ;AAAA,EAEA,MAAM,OAAO;AACX,UAAM,eAAe,YAAY;AAC/B,uBAAiB,SAAS,KAAK,OAAO;AACpC,YAAI,UAAU,aAAa,gBAAgB;AACzC,eAAK,WAAW,MAAM;AAAA,QACxB,OAAO;AACL,eAAK,WAAW,UAAU,KAAK;AAAA,QACjC;AAAA,MACF;AACA,WAAK,WAAW,SAAS;AAAA,IAC3B;AAEA,UAAM,YAAY,YAAY;AAC5B,uBAAiB,MAAM,KAAK,YAAY;AACtC,gBAAQ,GAAG,MAAM;AAAA,UACf,KAAK,aAAa;AAChB,iBAAK,MAAM,IAAI,EAAE,MAAM,gBAAgB,gBAAgB,CAAC;AACxD;AAAA,UACF,KAAK,aAAa;AAChB,iBAAK,MAAM,IAAI,EAAE,MAAM,gBAAgB,cAAc,CAAC;AAEtD,kBAAM,QAAQ,MAAM,KAAK,KAAK,UAAU,GAAG,MAAM;AACjD,gBAAI,CAAC,MAAM,aAAc,CAAC,EAAE,MAAM;AAChC;AAAA,YACF;AAEA,iBAAK,MAAM,IAAI,KAAK;AACpB;AAAA,QACJ;AAAA,MACF;AAAA,IACF;AAEA,YAAQ,IAAI,CAAC,aAAa,GAAG,UAAU,CAAC,CAAC;AAAA,EAC3C;AACF;","names":[]}

package/dist/stt/stt.cjs

@@ -0,0 +1,98 @@
+"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,
+  SpeechEventType: () => SpeechEventType,
+  SpeechStream: () => SpeechStream
+});
+module.exports = __toCommonJS(stt_exports);
+var import_utils = require("../utils.cjs");
+var SpeechEventType = /* @__PURE__ */ ((SpeechEventType2) => {
+  SpeechEventType2[SpeechEventType2["START_OF_SPEECH"] = 0] = "START_OF_SPEECH";
+  SpeechEventType2[SpeechEventType2["INTERIM_TRANSCRIPT"] = 1] = "INTERIM_TRANSCRIPT";
+  SpeechEventType2[SpeechEventType2["FINAL_TRANSCRIPT"] = 2] = "FINAL_TRANSCRIPT";
+  SpeechEventType2[SpeechEventType2["END_OF_SPEECH"] = 3] = "END_OF_SPEECH";
+  return SpeechEventType2;
+})(SpeechEventType || {});
+class STT {
+  #capabilities;
+  constructor(capabilities) {
+    this.#capabilities = capabilities;
+  }
+  /** Returns this STT's capabilities */
+  get capabilities() {
+    return this.#capabilities;
+  }
+}
+class SpeechStream {
+  static FLUSH_SENTINEL = Symbol("FLUSH_SENTINEL");
+  input = new import_utils.AsyncIterableQueue();
+  queue = new import_utils.AsyncIterableQueue();
+  closed = false;
+  /** Push an audio frame to the STT */
+  pushFrame(frame) {
+    if (this.input.closed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.input.put(frame);
+  }
+  /** Flush the STT, causing it to process all pending text */
+  flush() {
+    if (this.input.closed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.input.put(SpeechStream.FLUSH_SENTINEL);
+  }
+  /** Mark the input as ended and forbid additional pushes */
+  endInput() {
+    if (this.input.closed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.input.close();
+  }
+  next() {
+    return this.queue.next();
+  }
+  /** Close both the input and output of the STT stream */
+  close() {
+    this.input.close();
+    this.queue.close();
+    this.closed = true;
+  }
+  [Symbol.asyncIterator]() {
+    return this;
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  STT,
+  SpeechEventType,
+  SpeechStream
+});
+//# sourceMappingURL=stt.cjs.map

package/dist/stt/stt.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../src/stt/stt.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { AudioBuffer } from '../utils.js';\nimport { AsyncIterableQueue } from '../utils.js';\n\n/** Indicates start/middle/end of speech */\nexport enum SpeechEventType {\n  /**\n   * Indicate the start of speech.\n   * If the STT doesn't support this event, this will be emitted at the same time\n   * as the first INTERIM_TRANSCRIPT.\n   */\n  START_OF_SPEECH = 0,\n  /**\n   * Interim transcript, useful for real-time transcription.\n   */\n  INTERIM_TRANSCRIPT = 1,\n  /**\n   * Final transcript, emitted when the STT is confident enough that a certain\n   * portion of the speech will not change.\n   */\n  FINAL_TRANSCRIPT = 2,\n  /**\n   * Indicate the end of speech, emitted when the user stops speaking.\n   * The first alternative is a combination of all the previous FINAL_TRANSCRIPT events.\n   */\n  END_OF_SPEECH = 3,\n}\n\n/** SpeechData contains metadata about this {@link SpeechEvent}. */\nexport interface SpeechData {\n  language: string;\n  text: string;\n  startTime: number;\n  endTime: number;\n  confidence: number;\n}\n\n/** SpeechEvent is a packet of speech-to-text data. */\nexport interface SpeechEvent {\n  type: SpeechEventType;\n  alternatives?: [SpeechData, ...SpeechData[]];\n}\n\n/**\n * Describes the capabilities of the STT provider.\n *\n * @remarks\n * At present, the framework only supports providers that have a streaming endpoint.\n */\nexport interface STTCapabilities {\n  streaming: boolean;\n  interimResults: boolean;\n}\n\n/**\n * An instance of a speech-to-text adapter.\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child STT class, which inherits this class's methods.\n */\nexport abstract class STT {\n  #capabilities: STTCapabilities;\n\n  constructor(capabilities: STTCapabilities) {\n    this.#capabilities = capabilities;\n  }\n\n  /** Returns this STT's capabilities */\n  get capabilities(): STTCapabilities {\n    return this.#capabilities;\n  }\n\n  /** Receives an audio buffer and returns transcription in the form of a {@link SpeechEvent} */\n  abstract recognize(frame: AudioBuffer): Promise<SpeechEvent>;\n\n  /**\n   * Returns a {@link SpeechStream} that can be used to push audio frames and receive\n   * transcriptions\n   */\n  abstract stream(): SpeechStream;\n}\n\n/**\n * An instance of a speech-to-text stream, as an asynchronous iterable iterator.\n *\n * @example Looping through frames\n * ```ts\n * for await (const event of stream) {\n *   if (event.type === SpeechEventType.FINAL_TRANSCRIPT) {\n *     console.log(event.alternatives[0].text)\n *   }\n * }\n * ```\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child SpeechStream class, which inherits this class's methods.\n */\nexport abstract class SpeechStream implements AsyncIterableIterator<SpeechEvent> {\n  protected static readonly FLUSH_SENTINEL = Symbol('FLUSH_SENTINEL');\n  protected input = new AsyncIterableQueue<AudioFrame | typeof SpeechStream.FLUSH_SENTINEL>();\n  protected queue = new AsyncIterableQueue<SpeechEvent>();\n  protected closed = false;\n\n  /** Push an audio frame to the STT */\n  pushFrame(frame: AudioFrame) {\n    if (this.input.closed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.input.put(frame);\n  }\n\n  /** Flush the STT, causing it to process all pending text */\n  flush() {\n    if (this.input.closed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.input.put(SpeechStream.FLUSH_SENTINEL);\n  }\n\n  /** Mark the input as ended and forbid additional pushes */\n  endInput() {\n    if (this.input.closed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.input.close();\n  }\n\n  next(): Promise<IteratorResult<SpeechEvent>> {\n    return this.queue.next();\n  }\n\n  /** Close both the input and output of the STT stream */\n  close() {\n    this.input.close();\n    this.queue.close();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): SpeechStream {\n    return this;\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAKA,mBAAmC;AAG5B,IAAK,kBAAL,kBAAKA,qBAAL;AAML,EAAAA,kCAAA,qBAAkB,KAAlB;AAIA,EAAAA,kCAAA,wBAAqB,KAArB;AAKA,EAAAA,kCAAA,sBAAmB,KAAnB;AAKA,EAAAA,kCAAA,mBAAgB,KAAhB;AApBU,SAAAA;AAAA,GAAA;AAwDL,MAAe,IAAI;AAAA,EACxB;AAAA,EAEA,YAAY,cAA+B;AACzC,SAAK,gBAAgB;AAAA,EACvB;AAAA;AAAA,EAGA,IAAI,eAAgC;AAClC,WAAO,KAAK;AAAA,EACd;AAUF;AAkBO,MAAe,aAA2D;AAAA,EAC/E,OAA0B,iBAAiB,OAAO,gBAAgB;AAAA,EACxD,QAAQ,IAAI,gCAAoE;AAAA,EAChF,QAAQ,IAAI,gCAAgC;AAAA,EAC5C,SAAS;AAAA;AAAA,EAGnB,UAAU,OAAmB;AAC3B,QAAI,KAAK,MAAM,QAAQ;AACrB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM,IAAI,KAAK;AAAA,EACtB;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,KAAK,MAAM,QAAQ;AACrB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM,IAAI,aAAa,cAAc;AAAA,EAC5C;AAAA;AAAA,EAGA,WAAW;AACT,QAAI,KAAK,MAAM,QAAQ;AACrB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM,MAAM;AAAA,EACnB;AAAA,EAEA,OAA6C;AAC3C,WAAO,KAAK,MAAM,KAAK;AAAA,EACzB;AAAA;AAAA,EAGA,QAAQ;AACN,SAAK,MAAM,MAAM;AACjB,SAAK,MAAM,MAAM;AACjB,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAkB;AACrC,WAAO;AAAA,EACT;AACF;","names":["SpeechEventType"]}

package/dist/stt/stt.js

@@ -1,107 +1,72 @@
-import { AsyncIterableQueue } from '../utils.js';
-/** Indicates start/middle/end of speech */
-export var SpeechEventType;
-(function (SpeechEventType) {
-    /**
-     * Indicate the start of speech.
-     * If the STT doesn't support this event, this will be emitted at the same time
-     * as the first INTERIM_TRANSCRIPT.
-     */
-    SpeechEventType[SpeechEventType["START_OF_SPEECH"] = 0] = "START_OF_SPEECH";
-    /**
-     * Interim transcript, useful for real-time transcription.
-     */
-    SpeechEventType[SpeechEventType["INTERIM_TRANSCRIPT"] = 1] = "INTERIM_TRANSCRIPT";
-    /**
-     * Final transcript, emitted when the STT is confident enough that a certain
-     * portion of the speech will not change.
-     */
-    SpeechEventType[SpeechEventType["FINAL_TRANSCRIPT"] = 2] = "FINAL_TRANSCRIPT";
-    /**
-     * Indicate the end of speech, emitted when the user stops speaking.
-     * The first alternative is a combination of all the previous FINAL_TRANSCRIPT events.
-     */
-    SpeechEventType[SpeechEventType["END_OF_SPEECH"] = 3] = "END_OF_SPEECH";
-})(SpeechEventType || (SpeechEventType = {}));
-/**
- * An instance of a speech-to-text adapter.
- *
- * @remarks
- * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that
- * exports its own child STT class, which inherits this class's methods.
- */
-export class STT {
-    #capabilities;
-    constructor(capabilities) {
-        this.#capabilities = capabilities;
-    }
-    /** Returns this STT's capabilities */
-    get capabilities() {
-        return this.#capabilities;
-    }
+import { AsyncIterableQueue } from "../utils.js";
+var SpeechEventType = /* @__PURE__ */ ((SpeechEventType2) => {
+  SpeechEventType2[SpeechEventType2["START_OF_SPEECH"] = 0] = "START_OF_SPEECH";
+  SpeechEventType2[SpeechEventType2["INTERIM_TRANSCRIPT"] = 1] = "INTERIM_TRANSCRIPT";
+  SpeechEventType2[SpeechEventType2["FINAL_TRANSCRIPT"] = 2] = "FINAL_TRANSCRIPT";
+  SpeechEventType2[SpeechEventType2["END_OF_SPEECH"] = 3] = "END_OF_SPEECH";
+  return SpeechEventType2;
+})(SpeechEventType || {});
+class STT {
+  #capabilities;
+  constructor(capabilities) {
+    this.#capabilities = capabilities;
+  }
+  /** Returns this STT's capabilities */
+  get capabilities() {
+    return this.#capabilities;
+  }
 }
-/**
- * An instance of a speech-to-text stream, as an asynchronous iterable iterator.
- *
- * @example Looping through frames
- * ```ts
- * for await (const event of stream) {
- *   if (event.type === SpeechEventType.FINAL_TRANSCRIPT) {
- *     console.log(event.alternatives[0].text)
- *   }
- * }
- * ```
- *
- * @remarks
- * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that
- * exports its own child SpeechStream class, which inherits this class's methods.
- */
-export class SpeechStream {
-    static FLUSH_SENTINEL = Symbol('FLUSH_SENTINEL');
-    input = new AsyncIterableQueue();
-    queue = new AsyncIterableQueue();
-    closed = false;
-    /** Push an audio frame to the STT */
-    pushFrame(frame) {
-        if (this.input.closed) {
-            throw new Error('Input is closed');
-        }
-        if (this.closed) {
-            throw new Error('Stream is closed');
-        }
-        this.input.put(frame);
+class SpeechStream {
+  static FLUSH_SENTINEL = Symbol("FLUSH_SENTINEL");
+  input = new AsyncIterableQueue();
+  queue = new AsyncIterableQueue();
+  closed = false;
+  /** Push an audio frame to the STT */
+  pushFrame(frame) {
+    if (this.input.closed) {
+      throw new Error("Input is closed");
     }
-    /** Flush the STT, causing it to process all pending text */
-    flush() {
-        if (this.input.closed) {
-            throw new Error('Input is closed');
-        }
-        if (this.closed) {
-            throw new Error('Stream is closed');
-        }
-        this.input.put(SpeechStream.FLUSH_SENTINEL);
+    if (this.closed) {
+      throw new Error("Stream is closed");
     }
-    /** Mark the input as ended and forbid additional pushes */
-    endInput() {
-        if (this.input.closed) {
-            throw new Error('Input is closed');
-        }
-        if (this.closed) {
-            throw new Error('Stream is closed');
-        }
-        this.input.close();
+    this.input.put(frame);
+  }
+  /** Flush the STT, causing it to process all pending text */
+  flush() {
+    if (this.input.closed) {
+      throw new Error("Input is closed");
     }
-    next() {
-        return this.queue.next();
+    if (this.closed) {
+      throw new Error("Stream is closed");
     }
-    /** Close both the input and output of the STT stream */
-    close() {
-        this.input.close();
-        this.queue.close();
-        this.closed = true;
+    this.input.put(SpeechStream.FLUSH_SENTINEL);
+  }
+  /** Mark the input as ended and forbid additional pushes */
+  endInput() {
+    if (this.input.closed) {
+      throw new Error("Input is closed");
     }
-    [Symbol.asyncIterator]() {
-        return this;
+    if (this.closed) {
+      throw new Error("Stream is closed");
     }
+    this.input.close();
+  }
+  next() {
+    return this.queue.next();
+  }
+  /** Close both the input and output of the STT stream */
+  close() {
+    this.input.close();
+    this.queue.close();
+    this.closed = true;
+  }
+  [Symbol.asyncIterator]() {
+    return this;
+  }
 }
+export {
+  STT,
+  SpeechEventType,
+  SpeechStream
+};
 //# sourceMappingURL=stt.js.map

package/dist/stt/stt.js.map

@@ -1 +1 @@
-{"version":3,"file":"stt.js","sourceRoot":"","sources":["../../src/stt/stt.ts"],"names":[],"mappings":"AAKA,OAAO,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAC;AAEjD,2CAA2C;AAC3C,MAAM,CAAN,IAAY,eAqBX;AArBD,WAAY,eAAe;IACzB;;;;OAIG;IACH,2EAAmB,CAAA;IACnB;;OAEG;IACH,iFAAsB,CAAA;IACtB;;;OAGG;IACH,6EAAoB,CAAA;IACpB;;;OAGG;IACH,uEAAiB,CAAA;AACnB,CAAC,EArBW,eAAe,KAAf,eAAe,QAqB1B;AA4BD;;;;;;GAMG;AACH,MAAM,OAAgB,GAAG;IACvB,aAAa,CAAkB;IAE/B,YAAY,YAA6B;QACvC,IAAI,CAAC,aAAa,GAAG,YAAY,CAAC;IACpC,CAAC;IAED,sCAAsC;IACtC,IAAI,YAAY;QACd,OAAO,IAAI,CAAC,aAAa,CAAC;IAC5B,CAAC;CAUF;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,OAAgB,YAAY;IACtB,MAAM,CAAU,cAAc,GAAG,MAAM,CAAC,gBAAgB,CAAC,CAAC;IAC1D,KAAK,GAAG,IAAI,kBAAkB,EAAmD,CAAC;IAClF,KAAK,GAAG,IAAI,kBAAkB,EAAe,CAAC;IAC9C,MAAM,GAAG,KAAK,CAAC;IAEzB,qCAAqC;IACrC,SAAS,CAAC,KAAiB;QACzB,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;YACtB,MAAM,IAAI,KAAK,CAAC,iBAAiB,CAAC,CAAC;QACrC,CAAC;QACD,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;YAChB,MAAM,IAAI,KAAK,CAAC,kBAAkB,CAAC,CAAC;QACtC,CAAC;QACD,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IACxB,CAAC;IAED,4DAA4D;IAC5D,KAAK;QACH,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;YACtB,MAAM,IAAI,KAAK,CAAC,iBAAiB,CAAC,CAAC;QACrC,CAAC;QACD,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;YAChB,MAAM,IAAI,KAAK,CAAC,kBAAkB,CAAC,CAAC;QACtC,CAAC;QACD,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,YAAY,CAAC,cAAc,CAAC,CAAC;IAC9C,CAAC;IAED,2DAA2D;IAC3D,QAAQ;QACN,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;YACtB,MAAM,IAAI,KAAK,CAAC,iBAAiB,CAAC,CAAC;QACrC,CAAC;QACD,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;YAChB,MAAM,IAAI,KAAK,CAAC,kBAAkB,CAAC,CAAC;QACtC,CAAC;QACD,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;IACrB,CAAC;IAED,IAAI;QACF,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;IAC3B,CAAC;IAED,wDAAwD;IACxD,KAAK;QACH,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;QACnB,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;QACnB,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;IACrB,CAAC;IAED,CAAC,MAAM,CAAC,aAAa,CAAC;QACpB,OAAO,IAAI,CAAC;IACd,CAAC"}
+{"version":3,"sources":["../../src/stt/stt.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { AudioBuffer } from '../utils.js';\nimport { AsyncIterableQueue } from '../utils.js';\n\n/** Indicates start/middle/end of speech */\nexport enum SpeechEventType {\n  /**\n   * Indicate the start of speech.\n   * If the STT doesn't support this event, this will be emitted at the same time\n   * as the first INTERIM_TRANSCRIPT.\n   */\n  START_OF_SPEECH = 0,\n  /**\n   * Interim transcript, useful for real-time transcription.\n   */\n  INTERIM_TRANSCRIPT = 1,\n  /**\n   * Final transcript, emitted when the STT is confident enough that a certain\n   * portion of the speech will not change.\n   */\n  FINAL_TRANSCRIPT = 2,\n  /**\n   * Indicate the end of speech, emitted when the user stops speaking.\n   * The first alternative is a combination of all the previous FINAL_TRANSCRIPT events.\n   */\n  END_OF_SPEECH = 3,\n}\n\n/** SpeechData contains metadata about this {@link SpeechEvent}. */\nexport interface SpeechData {\n  language: string;\n  text: string;\n  startTime: number;\n  endTime: number;\n  confidence: number;\n}\n\n/** SpeechEvent is a packet of speech-to-text data. */\nexport interface SpeechEvent {\n  type: SpeechEventType;\n  alternatives?: [SpeechData, ...SpeechData[]];\n}\n\n/**\n * Describes the capabilities of the STT provider.\n *\n * @remarks\n * At present, the framework only supports providers that have a streaming endpoint.\n */\nexport interface STTCapabilities {\n  streaming: boolean;\n  interimResults: boolean;\n}\n\n/**\n * An instance of a speech-to-text adapter.\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child STT class, which inherits this class's methods.\n */\nexport abstract class STT {\n  #capabilities: STTCapabilities;\n\n  constructor(capabilities: STTCapabilities) {\n    this.#capabilities = capabilities;\n  }\n\n  /** Returns this STT's capabilities */\n  get capabilities(): STTCapabilities {\n    return this.#capabilities;\n  }\n\n  /** Receives an audio buffer and returns transcription in the form of a {@link SpeechEvent} */\n  abstract recognize(frame: AudioBuffer): Promise<SpeechEvent>;\n\n  /**\n   * Returns a {@link SpeechStream} that can be used to push audio frames and receive\n   * transcriptions\n   */\n  abstract stream(): SpeechStream;\n}\n\n/**\n * An instance of a speech-to-text stream, as an asynchronous iterable iterator.\n *\n * @example Looping through frames\n * ```ts\n * for await (const event of stream) {\n *   if (event.type === SpeechEventType.FINAL_TRANSCRIPT) {\n *     console.log(event.alternatives[0].text)\n *   }\n * }\n * ```\n *\n * @remarks\n * This class is abstract, and as such cannot be used directly. Instead, use a provider plugin that\n * exports its own child SpeechStream class, which inherits this class's methods.\n */\nexport abstract class SpeechStream implements AsyncIterableIterator<SpeechEvent> {\n  protected static readonly FLUSH_SENTINEL = Symbol('FLUSH_SENTINEL');\n  protected input = new AsyncIterableQueue<AudioFrame | typeof SpeechStream.FLUSH_SENTINEL>();\n  protected queue = new AsyncIterableQueue<SpeechEvent>();\n  protected closed = false;\n\n  /** Push an audio frame to the STT */\n  pushFrame(frame: AudioFrame) {\n    if (this.input.closed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.input.put(frame);\n  }\n\n  /** Flush the STT, causing it to process all pending text */\n  flush() {\n    if (this.input.closed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.input.put(SpeechStream.FLUSH_SENTINEL);\n  }\n\n  /** Mark the input as ended and forbid additional pushes */\n  endInput() {\n    if (this.input.closed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.input.close();\n  }\n\n  next(): Promise<IteratorResult<SpeechEvent>> {\n    return this.queue.next();\n  }\n\n  /** Close both the input and output of the STT stream */\n  close() {\n    this.input.close();\n    this.queue.close();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): SpeechStream {\n    return this;\n  }\n}\n"],"mappings":"AAKA,SAAS,0BAA0B;AAG5B,IAAK,kBAAL,kBAAKA,qBAAL;AAML,EAAAA,kCAAA,qBAAkB,KAAlB;AAIA,EAAAA,kCAAA,wBAAqB,KAArB;AAKA,EAAAA,kCAAA,sBAAmB,KAAnB;AAKA,EAAAA,kCAAA,mBAAgB,KAAhB;AApBU,SAAAA;AAAA,GAAA;AAwDL,MAAe,IAAI;AAAA,EACxB;AAAA,EAEA,YAAY,cAA+B;AACzC,SAAK,gBAAgB;AAAA,EACvB;AAAA;AAAA,EAGA,IAAI,eAAgC;AAClC,WAAO,KAAK;AAAA,EACd;AAUF;AAkBO,MAAe,aAA2D;AAAA,EAC/E,OAA0B,iBAAiB,OAAO,gBAAgB;AAAA,EACxD,QAAQ,IAAI,mBAAoE;AAAA,EAChF,QAAQ,IAAI,mBAAgC;AAAA,EAC5C,SAAS;AAAA;AAAA,EAGnB,UAAU,OAAmB;AAC3B,QAAI,KAAK,MAAM,QAAQ;AACrB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM,IAAI,KAAK;AAAA,EACtB;AAAA;AAAA,EAGA,QAAQ;AACN,QAAI,KAAK,MAAM,QAAQ;AACrB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM,IAAI,aAAa,cAAc;AAAA,EAC5C;AAAA;AAAA,EAGA,WAAW;AACT,QAAI,KAAK,MAAM,QAAQ;AACrB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,MAAM,MAAM;AAAA,EACnB;AAAA,EAEA,OAA6C;AAC3C,WAAO,KAAK,MAAM,KAAK;AAAA,EACzB;AAAA;AAAA,EAGA,QAAQ;AACN,SAAK,MAAM,MAAM;AACjB,SAAK,MAAM,MAAM;AACjB,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAkB;AACrC,WAAO;AAAA,EACT;AACF;","names":["SpeechEventType"]}