@fonoster/autopilot 0.9.40 → 0.9.43

package/dist/Autopilot.d.ts

@@ -2,8 +2,7 @@ import { AutopilotParams } from "./types";
 declare class Autopilot {
     private readonly params;
     private readonly actor;
-    private readonly vadWorker;
-    private vadWorkerReady;
+    private vad;
     constructor(params: AutopilotParams);
     start(): Promise<void>;
     stop(): void;

package/dist/Autopilot.js

@@ -23,29 +23,15 @@ exports.Autopilot = void 0;
  * limitations under the License.
  */
 const path_1 = __importDefault(require("path"));
-const worker_threads_1 = require("worker_threads");
 const logger_1 = require("@fonoster/logger");
 const xstate_1 = require("xstate");
 const machine_1 = require("./machine/machine");
+const vad_1 = require("./vad");
 const logger = (0, logger_1.getLogger)({ service: "autopilot", filePath: __filename });
 class Autopilot {
     constructor(params) {
         this.params = params;
         const { voice, languageModel, conversationSettings } = this.params;
-        const vadWorkerPath = path_1.default.resolve(__dirname, "../dist", "./vadWorker");
-        this.vadWorker = new worker_threads_1.Worker(vadWorkerPath, {
-            workerData: conversationSettings.vad
-        });
-        this.vadWorkerReady = new Promise((resolve, reject) => {
-            logger.verbose("waiting for vad worker to be ready");
-            this.vadWorker.once("message", (message) => {
-                if (message === "VAD_READY") {
-                    logger.verbose("vad worker is ready");
-                    resolve();
-                }
-            });
-            this.vadWorker.once("error", reject);
-        });
         this.actor = (0, xstate_1.createActor)(machine_1.machine, {
             input: {
                 conversationSettings,
@@ -55,22 +41,23 @@ class Autopilot {
         });
     }
     async start() {
-        // Wait for all the streams and vad worker to be ready before proceeding starting the actor
-        await this.vadWorkerReady;
+        const vadParams = this.params.conversationSettings.vad;
+        const sileroVad = new vad_1.SileroVad({
+            pathToModel: vadParams.pathToModel ||
+                path_1.default.resolve(__dirname, "..", "silero_vad_v5.onnx"),
+            activationThreshold: vadParams.activationThreshold,
+            deactivationThreshold: vadParams.deactivationThreshold,
+            debounceFrames: vadParams.debounceFrames
+        });
+        await sileroVad.init();
+        this.vad = sileroVad;
         await this.setupVoiceStream();
         await this.setupSpeechGathering();
         this.actor.start();
+        logger.verbose("autopilot is ready");
         this.actor.subscribe((state) => {
             logger.verbose("actor's new state is", { state: state.value });
         });
-        this.vadWorker.on("error", (err) => {
-            logger.error("vad worker error", err);
-        });
-        this.vadWorker.on("exit", (code) => {
-            if (code !== 0) {
-                logger.error("vad worker stopped with exit code", { code });
-            }
-        });
     }
     stop() {
         logger.verbose("stopping autopilot");
@@ -80,19 +67,23 @@ class Autopilot {
         const { voice } = this.params;
         const stream = await voice.stream();
         stream.onData(this.handleVoicePayload.bind(this));
-        this.vadWorker.on("message", (event) => {
-            logger.verbose("received speech event from vad", { event });
-            if (event === "SPEECH_START") {
-                this.actor.send({ type: "SPEECH_START" });
-            }
-            else if (event === "SPEECH_END") {
-                this.actor.send({ type: "SPEECH_END" });
-            }
-        });
     }
     handleVoicePayload(chunk) {
         try {
-            this.vadWorker.postMessage(chunk);
+            if (!this.vad) {
+                logger.error("VAD not initialized");
+                return;
+            }
+            // Process the audio chunk with the VAD directly
+            this.vad.processChunk(chunk, (event) => {
+                logger.verbose("received speech event from vad", { event });
+                if (event === "SPEECH_START") {
+                    this.actor.send({ type: "SPEECH_START" });
+                }
+                else if (event === "SPEECH_END") {
+                    this.actor.send({ type: "SPEECH_END" });
+                }
+            });
         }
         catch (err) {
             logger.error("an error occurred while processing vad", err);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@fonoster/autopilot",
-  "version": "0.9.40",
+  "version": "0.9.43",
   "description": "Voice AI for the Fonoster platform",
   "author": "Pedro Sanders <psanders@fonoster.com>",
   "homepage": "https://github.com/fonoster/fonoster#readme",
@@ -33,11 +33,11 @@
   },
   "dependencies": {
     "@aws-sdk/client-s3": "^3.712.0",
-    "@fonoster/common": "^0.9.35",
-    "@fonoster/logger": "^0.9.30",
-    "@fonoster/sdk": "^0.9.37",
-    "@fonoster/types": "^0.9.30",
-    "@fonoster/voice": "^0.9.35",
+    "@fonoster/common": "^0.9.42",
+    "@fonoster/logger": "^0.9.42",
+    "@fonoster/sdk": "^0.9.42",
+    "@fonoster/types": "^0.9.42",
+    "@fonoster/voice": "^0.9.42",
     "@langchain/community": "^0.3.32",
     "@langchain/core": "^0.3.40",
     "@langchain/groq": "^0.1.3",
@@ -56,5 +56,5 @@
     "xstate": "^5.17.3",
     "zod": "^3.23.8"
   },
-  "gitHead": "b5150ccfba8a96468d4e478796ac8e84fe57ff07"
+  "gitHead": "7fa2745908c3ced8a5401a48b26cf26c8fc09c2c"
 }