@seawork/server 1.0.0

package/dist/server/server/speech/providers/local/sherpa/sherpa-stt.js

@@ -0,0 +1,138 @@
+import { EventEmitter } from "node:events";
+import { v4 as uuidv4 } from "uuid";
+import { Pcm16MonoResampler } from "../../../../agent/pcm16-resampler.js";
+import { parsePcm16MonoWav, parsePcmRateFromFormat, pcm16lePeakAbs, pcm16leToFloat32, } from "../../../audio.js";
+export class SherpaOnnxSTT {
+    constructor(config, logger) {
+        this.id = "local";
+        this.engine = config.engine;
+        this.silencePeakThreshold = config.silencePeakThreshold ?? 300;
+        this.tailPaddingMs = config.tailPaddingMs ?? 500;
+        this.logger = logger.child({ module: "speech", provider: "local", component: "stt" });
+    }
+    createSession(params) {
+        const emitter = new EventEmitter();
+        void params;
+        const requiredSampleRate = this.engine.sampleRate;
+        let connected = false;
+        let segmentId = uuidv4();
+        let previousSegmentId = null;
+        let pcm16 = Buffer.alloc(0);
+        return {
+            requiredSampleRate,
+            async connect() {
+                connected = true;
+            },
+            appendPcm16(chunk) {
+                if (!connected) {
+                    emitter.emit("error", new Error("STT session not connected"));
+                    return;
+                }
+                pcm16 = pcm16.length === 0 ? chunk : Buffer.concat([pcm16, chunk]);
+            },
+            commit: () => {
+                if (!connected) {
+                    emitter.emit("error", new Error("STT session not connected"));
+                    return;
+                }
+                const committedId = segmentId;
+                const prev = previousSegmentId;
+                emitter.emit("committed", { segmentId: committedId, previousSegmentId: prev });
+                void (async () => {
+                    try {
+                        const rt = await this.transcribeAudio(pcm16, `audio/pcm;rate=${requiredSampleRate}`);
+                        emitter.emit("transcript", {
+                            segmentId: committedId,
+                            transcript: rt.text,
+                            isFinal: true,
+                            language: rt.language,
+                            logprobs: rt.logprobs,
+                            avgLogprob: rt.avgLogprob,
+                            isLowConfidence: rt.isLowConfidence,
+                        });
+                    }
+                    catch (err) {
+                        emitter.emit("error", err);
+                    }
+                    finally {
+                        previousSegmentId = committedId;
+                        segmentId = uuidv4();
+                        pcm16 = Buffer.alloc(0);
+                    }
+                })();
+            },
+            clear() {
+                pcm16 = Buffer.alloc(0);
+                segmentId = uuidv4();
+            },
+            close() {
+                connected = false;
+                pcm16 = Buffer.alloc(0);
+            },
+            on(event, handler) {
+                emitter.on(event, handler);
+                return undefined;
+            },
+        };
+    }
+    async transcribeAudio(audioBuffer, format) {
+        const start = Date.now();
+        let inputRate;
+        let pcm16;
+        if (format.toLowerCase().includes("audio/wav")) {
+            const parsed = parsePcm16MonoWav(audioBuffer);
+            inputRate = parsed.sampleRate;
+            pcm16 = parsed.pcm16;
+        }
+        else if (format.toLowerCase().includes("audio/pcm")) {
+            inputRate = parsePcmRateFromFormat(format, this.engine.sampleRate) ?? this.engine.sampleRate;
+            pcm16 = audioBuffer;
+        }
+        else {
+            throw new Error(`Unsupported audio format for sherpa STT: ${format}`);
+        }
+        const peak = pcm16lePeakAbs(pcm16);
+        if (peak < this.silencePeakThreshold) {
+            return { text: "", duration: Date.now() - start, isLowConfidence: true };
+        }
+        let pcmForModel = pcm16;
+        if (inputRate !== this.engine.sampleRate) {
+            const resampler = new Pcm16MonoResampler({ inputRate, outputRate: this.engine.sampleRate });
+            pcmForModel = resampler.processChunk(pcm16);
+            inputRate = this.engine.sampleRate;
+        }
+        const peakForModel = pcm16lePeakAbs(pcmForModel);
+        const peakFloat = peakForModel / 32768.0;
+        const targetPeak = 0.6;
+        const maxGain = 50;
+        const gain = peakFloat > 0 && peakFloat < targetPeak ? Math.min(maxGain, targetPeak / peakFloat) : 1;
+        const stream = this.engine.createStream();
+        try {
+            const floatSamples = pcm16leToFloat32(pcmForModel, gain);
+            stream.acceptWaveform(inputRate, floatSamples);
+            while (this.engine.recognizer.isReady(stream)) {
+                this.engine.recognizer.decode(stream);
+            }
+            const padSamples = Math.max(0, Math.round((this.engine.sampleRate * this.tailPaddingMs) / 1000));
+            if (padSamples > 0) {
+                stream.acceptWaveform(this.engine.sampleRate, new Float32Array(padSamples));
+            }
+            while (this.engine.recognizer.isReady(stream)) {
+                this.engine.recognizer.decode(stream);
+            }
+            const text = String(this.engine.recognizer.getResult(stream)?.text ?? "").trim();
+            const duration = Date.now() - start;
+            this.logger.debug({ duration, textLength: text.length }, "Sherpa transcription complete");
+            return { text, duration, ...(text.length === 0 ? { isLowConfidence: true } : {}) };
+        }
+        finally {
+            try {
+                stream.free?.();
+            }
+            catch {
+                // ignore
+            }
+        }
+    }
+}
+//# sourceMappingURL=sherpa-stt.js.map

package/dist/server/server/speech/providers/local/sherpa/sherpa-stt.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sherpa-stt.js","sourceRoot":"","sources":["../../../../../../../src/server/speech/providers/local/sherpa/sherpa-stt.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAC3C,OAAO,EAAE,EAAE,IAAI,MAAM,EAAE,MAAM,MAAM,CAAC;AAQpC,OAAO,EAAE,kBAAkB,EAAE,MAAM,sCAAsC,CAAC;AAC1E,OAAO,EACL,iBAAiB,EACjB,sBAAsB,EACtB,cAAc,EACd,gBAAgB,GACjB,MAAM,mBAAmB,CAAC;AAS3B,MAAM,OAAO,aAAa;IAOxB,YAAY,MAAuB,EAAE,MAAmB;QAFxC,OAAE,GAAG,OAAgB,CAAC;QAGpC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC;QAC5B,IAAI,CAAC,oBAAoB,GAAG,MAAM,CAAC,oBAAoB,IAAI,GAAG,CAAC;QAC/D,IAAI,CAAC,aAAa,GAAG,MAAM,CAAC,aAAa,IAAI,GAAG,CAAC;QACjD,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC,CAAC;IACxF,CAAC;IAEM,aAAa,CAAC,MAIpB;QACC,MAAM,OAAO,GAAG,IAAI,YAAY,EAAE,CAAC;QACnC,KAAK,MAAM,CAAC;QACZ,MAAM,kBAAkB,GAAG,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC;QAClD,IAAI,SAAS,GAAG,KAAK,CAAC;QACtB,IAAI,SAAS,GAAG,MAAM,EAAE,CAAC;QACzB,IAAI,iBAAiB,GAAkB,IAAI,CAAC;QAC5C,IAAI,KAAK,GAAW,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QAEpC,OAAO;YACL,kBAAkB;YAClB,KAAK,CAAC,OAAO;gBACX,SAAS,GAAG,IAAI,CAAC;YACnB,CAAC;YACD,WAAW,CAAC,KAAa;gBACvB,IAAI,CAAC,SAAS,EAAE,CAAC;oBACd,OAAe,CAAC,IAAI,CAAC,OAAO,EAAE,IAAI,KAAK,CAAC,2BAA2B,CAAC,CAAC,CAAC;oBACvE,OAAO;gBACT,CAAC;gBACD,KAAK,GAAG,KAAK,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;YACrE,CAAC;YACD,MAAM,EAAE,GAAG,EAAE;gBACX,IAAI,CAAC,SAAS,EAAE,CAAC;oBACd,OAAe,CAAC,IAAI,CAAC,OAAO,EAAE,IAAI,KAAK,CAAC,2BAA2B,CAAC,CAAC,CAAC;oBACvE,OAAO;gBACT,CAAC;gBAED,MAAM,WAAW,GAAG,SAAS,CAAC;gBAC9B,MAAM,IAAI,GAAG,iBAAiB,CAAC;gBAC9B,OAAe,CAAC,IAAI,CAAC,WAAW,EAAE,EAAE,SAAS,EAAE,WAAW,EAAE,iBAAiB,EAAE,IAAI,EAAE,CAAC,CAAC;gBAExF,KAAK,CAAC,KAAK,IAAI,EAAE;oBACf,IAAI,CAAC;wBACH,MAAM,EAAE,GAAG,MAAM,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE,kBAAkB,kBAAkB,EAAE,CAAC,CAAC;wBACpF,OAAe,CAAC,IAAI,CAAC,YAAY,EAAE;4BAClC,SAAS,EAAE,WAAW;4BACtB,UAAU,EAAE,EAAE,CAAC,IAAI;4BACnB,OAAO,EAAE,IAAI;4BACb,QAAQ,EAAE,EAAE,CAAC,QAAQ;4BACrB,QAAQ,EAAE,EAAE,CAAC,QAAQ;4BACrB,UAAU,EAAE,EAAE,CAAC,UAAU;4BACzB,eAAe,EAAE,EAAE,CAAC,eAAe;yBACpC,CAAC,CAAC;oBACL,CAAC;oBAAC,OAAO,GAAG,EAAE,CAAC;wBACZ,OAAe,CAAC,IAAI,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC;oBACtC,CAAC;4BAAS,CAAC;wBACT,iBAAiB,GAAG,WAAW,CAAC;wBAChC,SAAS,GAAG,MAAM,EAAE,CAAC;wBACrB,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;oBAC1B,CAAC;gBACH,CAAC,CAAC,EAAE,CAAC;YACP,CAAC;YACD,KAAK;gBACH,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;gBACxB,SAAS,GAAG,MAAM,EAAE,CAAC;YACvB,CAAC;YACD,KAAK;gBACH,SAAS,GAAG,KAAK,CAAC;gBAClB,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;YAC1B,CAAC;YACD,EAAE,CAAC,KAAU,EAAE,OAAY;gBACzB,OAAO,CAAC,EAAE,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;gBAC3B,OAAO,SAAS,CAAC;YACnB,CAAC;SACF,CAAC;IACJ,CAAC;IAEM,KAAK,CAAC,eAAe,CAAC,WAAmB,EAAE,MAAc;QAC9D,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAEzB,IAAI,SAAiB,CAAC;QACtB,IAAI,KAAa,CAAC;QAElB,IAAI,MAAM,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;YAC/C,MAAM,MAAM,GAAG,iBAAiB,CAAC,WAAW,CAAC,CAAC;YAC9C,SAAS,GAAG,MAAM,CAAC,UAAU,CAAC;YAC9B,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QACvB,CAAC;aAAM,IAAI,MAAM,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;YACtD,SAAS,GAAG,sBAAsB,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC,IAAI,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC;YAC7F,KAAK,GAAG,WAAW,CAAC;QACtB,CAAC;aAAM,CAAC;YACN,MAAM,IAAI,KAAK,CAAC,4CAA4C,MAAM,EAAE,CAAC,CAAC;QACxE,CAAC;QAED,MAAM,IAAI,GAAG,cAAc,CAAC,KAAK,CAAC,CAAC;QACnC,IAAI,IAAI,GAAG,IAAI,CAAC,oBAAoB,EAAE,CAAC;YACrC,OAAO,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,KAAK,EAAE,eAAe,EAAE,IAAI,EAAE,CAAC;QAC3E,CAAC;QAED,IAAI,WAAW,GAAG,KAAK,CAAC;QACxB,IAAI,SAAS,KAAK,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,CAAC;YACzC,MAAM,SAAS,GAAG,IAAI,kBAAkB,CAAC,EAAE,SAAS,EAAE,UAAU,EAAE,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,CAAC,CAAC;YAC5F,WAAW,GAAG,SAAS,CAAC,YAAY,CAAC,KAAK,CAAC,CAAC;YAC5C,SAAS,GAAG,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC;QACrC,CAAC;QAED,MAAM,YAAY,GAAG,cAAc,CAAC,WAAW,CAAC,CAAC;QACjD,MAAM,SAAS,GAAG,YAAY,GAAG,OAAO,CAAC;QACzC,MAAM,UAAU,GAAG,GAAG,CAAC;QACvB,MAAM,OAAO,GAAG,EAAE,CAAC;QACnB,MAAM,IAAI,GACR,SAAS,GAAG,CAAC,IAAI,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,EAAE,UAAU,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAE1F,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,YAAY,EAAE,CAAC;QAC1C,IAAI,CAAC;YACH,MAAM,YAAY,GAAG,gBAAgB,CAAC,WAAW,EAAE,IAAI,CAAC,CAAC;YACzD,MAAM,CAAC,cAAc,CAAC,SAAS,EAAE,YAAY,CAAC,CAAC;YAE/C,OAAO,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC9C,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;YACxC,CAAC;YAED,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,CACzB,CAAC,EACD,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,UAAU,GAAG,IAAI,CAAC,aAAa,CAAC,GAAG,IAAI,CAAC,CACjE,CAAC;YACF,IAAI,UAAU,GAAG,CAAC,EAAE,CAAC;gBACnB,MAAM,CAAC,cAAc,CAAC,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,IAAI,YAAY,CAAC,UAAU,CAAC,CAAC,CAAC;YAC9E,CAAC;YAED,OAAO,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC9C,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;YACxC,CAAC;YAED,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC,SAAS,CAAC,MAAM,CAAC,EAAE,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;YACjF,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,KAAK,CAAC;YACpC,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,QAAQ,EAAE,UAAU,EAAE,IAAI,CAAC,MAAM,EAAE,EAAE,+BAA+B,CAAC,CAAC;YAC1F,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,CAAC,IAAI,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,eAAe,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC;QACrF,CAAC;gBAAS,CAAC;YACT,IAAI,CAAC;gBACH,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC;YAClB,CAAC;YAAC,MAAM,CAAC;gBACP,SAAS;YACX,CAAC;QACH,CAAC;IACH,CAAC;CACF"}

package/dist/server/server/speech/providers/local/sherpa/sherpa-tts.d.ts

@@ -0,0 +1,21 @@
+import type pino from "pino";
+import type { SpeechStreamResult, TextToSpeechProvider } from "../../../speech-provider.js";
+export type SherpaTtsPreset = "kokoro-en-v0_19" | "kitten-nano-en-v0_1-fp16";
+export type SherpaTtsConfig = {
+    preset: SherpaTtsPreset;
+    modelDir: string;
+    speakerId?: number;
+    speed?: number;
+    lengthScale?: number;
+    numThreads?: number;
+};
+export declare class SherpaOnnxTTS implements TextToSpeechProvider {
+    private readonly tts;
+    private readonly speakerId;
+    private readonly speed;
+    private readonly logger;
+    constructor(config: SherpaTtsConfig, logger: pino.Logger);
+    synthesizeSpeech(text: string): Promise<SpeechStreamResult>;
+    free(): void;
+}
+//# sourceMappingURL=sherpa-tts.d.ts.map

package/dist/server/server/speech/providers/local/sherpa/sherpa-tts.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"sherpa-tts.d.ts","sourceRoot":"","sources":["../../../../../../../src/server/speech/providers/local/sherpa/sherpa-tts.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,IAAI,MAAM,MAAM,CAAC;AAI7B,OAAO,KAAK,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,6BAA6B,CAAC;AAI5F,MAAM,MAAM,eAAe,GAAG,iBAAiB,GAAG,0BAA0B,CAAC;AAE7E,MAAM,MAAM,eAAe,GAAG;IAC5B,MAAM,EAAE,eAAe,CAAC;IACxB,QAAQ,EAAE,MAAM,CAAC;IACjB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,UAAU,CAAC,EAAE,MAAM,CAAC;CACrB,CAAC;AAQF,qBAAa,aAAc,YAAW,oBAAoB;IACxD,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAM;IAC1B,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAS;IACnC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAS;IAC/B,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAc;gBAEzB,MAAM,EAAE,eAAe,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM;IA2DlD,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,kBAAkB,CAAC;IA+CjE,IAAI,IAAI,IAAI;CAOb"}

package/dist/server/server/speech/providers/local/sherpa/sherpa-tts.js

@@ -0,0 +1,108 @@
+import { Readable } from "node:stream";
+import { existsSync } from "node:fs";
+import { chunkBuffer, float32ToPcm16le } from "../../../audio.js";
+import { loadSherpaOnnxNode } from "./sherpa-onnx-node-loader.js";
+function assertFileExists(filePath, label) {
+    if (!existsSync(filePath)) {
+        throw new Error(`Missing ${label}: ${filePath}`);
+    }
+}
+export class SherpaOnnxTTS {
+    constructor(config, logger) {
+        if (config.preset !== "kokoro-en-v0_19" && config.preset !== "kitten-nano-en-v0_1-fp16") {
+            throw new Error(`Unsupported Sherpa TTS preset: ${config.preset}`);
+        }
+        this.logger = logger.child({ module: "speech", provider: "local", component: "tts" });
+        this.speakerId = config.speakerId ?? 0;
+        this.speed = config.speed ?? 1.0;
+        const sherpa = loadSherpaOnnxNode();
+        if (typeof sherpa.OfflineTts !== "function") {
+            throw new Error("sherpa-onnx-node OfflineTts is unavailable");
+        }
+        const modelFile = config.preset === "kokoro-en-v0_19" ? "model.onnx" : "model.fp16.onnx";
+        const modelPath = `${config.modelDir}/${modelFile}`;
+        const voicesPath = `${config.modelDir}/voices.bin`;
+        const tokensPath = `${config.modelDir}/tokens.txt`;
+        const dataDir = `${config.modelDir}/espeak-ng-data`;
+        assertFileExists(modelPath, "TTS model");
+        assertFileExists(voicesPath, "TTS voices");
+        assertFileExists(tokensPath, "TTS tokens");
+        assertFileExists(dataDir, "TTS espeak-ng dataDir");
+        const modelConfig = config.preset === "kokoro-en-v0_19"
+            ? {
+                kokoro: {
+                    model: modelPath,
+                    voices: voicesPath,
+                    tokens: tokensPath,
+                    dataDir,
+                    lengthScale: config.lengthScale ?? 1.0,
+                },
+            }
+            : {
+                kitten: {
+                    model: modelPath,
+                    voices: voicesPath,
+                    tokens: tokensPath,
+                    dataDir,
+                    lengthScale: config.lengthScale ?? 1.0,
+                },
+            };
+        const offlineTtsConfig = {
+            model: modelConfig,
+            numThreads: config.numThreads ?? 2,
+            provider: "cpu",
+            maxNumSentences: 1,
+        };
+        this.tts = new sherpa.OfflineTts(offlineTtsConfig);
+        this.logger.info({ preset: config.preset, modelDir: config.modelDir }, "Sherpa offline TTS initialized");
+    }
+    async synthesizeSpeech(text) {
+        const trimmed = text.trim();
+        if (!trimmed) {
+            throw new Error("Cannot synthesize empty text");
+        }
+        const audio = this.tts.generate({
+            text: trimmed,
+            sid: this.speakerId,
+            speed: this.speed,
+            // Electron rejects native external-backed typed arrays. Request a copied buffer
+            // from sherpa itself instead of trying to clone after generate() returns.
+            enableExternalBuffer: false,
+        });
+        const rawSamples = audio && audio.samples instanceof Float32Array
+            ? audio.samples
+            : audio && Array.isArray(audio.samples)
+                ? Float32Array.from(audio.samples)
+                : null;
+        // Copy to avoid "External buffers are not allowed" when sherpa-onnx
+        // returns a Float32Array backed by native memory.
+        const samples = rawSamples ? Float32Array.from(rawSamples) : null;
+        const sampleRate = audio &&
+            typeof audio.sampleRate === "number" &&
+            Number.isFinite(audio.sampleRate) &&
+            audio.sampleRate > 0
+            ? audio.sampleRate
+            : typeof this.tts.sampleRate === "number"
+                ? this.tts.sampleRate
+                : 24000;
+        if (!samples) {
+            throw new Error("Unexpected sherpa TTS output: missing Float32 samples");
+        }
+        const pcm16 = float32ToPcm16le(samples);
+        const chunkBytes = Math.max(2, Math.round(sampleRate * 0.05) * 2); // ~50ms
+        const chunks = chunkBuffer(pcm16, chunkBytes);
+        return {
+            stream: Readable.from(chunks),
+            format: `pcm;rate=${sampleRate}`,
+        };
+    }
+    free() {
+        try {
+            this.tts?.free?.();
+        }
+        catch {
+            // ignore
+        }
+    }
+}
+//# sourceMappingURL=sherpa-tts.js.map

package/dist/server/server/speech/providers/local/sherpa/sherpa-tts.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sherpa-tts.js","sourceRoot":"","sources":["../../../../../../../src/server/speech/providers/local/sherpa/sherpa-tts.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AACvC,OAAO,EAAE,UAAU,EAAE,MAAM,SAAS,CAAC;AAGrC,OAAO,EAAE,WAAW,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAC;AAClE,OAAO,EAAE,kBAAkB,EAAE,MAAM,8BAA8B,CAAC;AAalE,SAAS,gBAAgB,CAAC,QAAgB,EAAE,KAAa;IACvD,IAAI,CAAC,UAAU,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC1B,MAAM,IAAI,KAAK,CAAC,WAAW,KAAK,KAAK,QAAQ,EAAE,CAAC,CAAC;IACnD,CAAC;AACH,CAAC;AAED,MAAM,OAAO,aAAa;IAMxB,YAAY,MAAuB,EAAE,MAAmB;QACtD,IAAI,MAAM,CAAC,MAAM,KAAK,iBAAiB,IAAI,MAAM,CAAC,MAAM,KAAK,0BAA0B,EAAE,CAAC;YACxF,MAAM,IAAI,KAAK,CAAC,kCAAkC,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC;QACrE,CAAC;QACD,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC,CAAC;QACtF,IAAI,CAAC,SAAS,GAAG,MAAM,CAAC,SAAS,IAAI,CAAC,CAAC;QACvC,IAAI,CAAC,KAAK,GAAG,MAAM,CAAC,KAAK,IAAI,GAAG,CAAC;QAEjC,MAAM,MAAM,GAAG,kBAAkB,EAAE,CAAC;QACpC,IAAI,OAAO,MAAM,CAAC,UAAU,KAAK,UAAU,EAAE,CAAC;YAC5C,MAAM,IAAI,KAAK,CAAC,4CAA4C,CAAC,CAAC;QAChE,CAAC;QAED,MAAM,SAAS,GAAG,MAAM,CAAC,MAAM,KAAK,iBAAiB,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,iBAAiB,CAAC;QACzF,MAAM,SAAS,GAAG,GAAG,MAAM,CAAC,QAAQ,IAAI,SAAS,EAAE,CAAC;QACpD,MAAM,UAAU,GAAG,GAAG,MAAM,CAAC,QAAQ,aAAa,CAAC;QACnD,MAAM,UAAU,GAAG,GAAG,MAAM,CAAC,QAAQ,aAAa,CAAC;QACnD,MAAM,OAAO,GAAG,GAAG,MAAM,CAAC,QAAQ,iBAAiB,CAAC;QAEpD,gBAAgB,CAAC,SAAS,EAAE,WAAW,CAAC,CAAC;QACzC,gBAAgB,CAAC,UAAU,EAAE,YAAY,CAAC,CAAC;QAC3C,gBAAgB,CAAC,UAAU,EAAE,YAAY,CAAC,CAAC;QAC3C,gBAAgB,CAAC,OAAO,EAAE,uBAAuB,CAAC,CAAC;QAEnD,MAAM,WAAW,GACf,MAAM,CAAC,MAAM,KAAK,iBAAiB;YACjC,CAAC,CAAC;gBACE,MAAM,EAAE;oBACN,KAAK,EAAE,SAAS;oBAChB,MAAM,EAAE,UAAU;oBAClB,MAAM,EAAE,UAAU;oBAClB,OAAO;oBACP,WAAW,EAAE,MAAM,CAAC,WAAW,IAAI,GAAG;iBACvC;aACF;YACH,CAAC,CAAC;gBACE,MAAM,EAAE;oBACN,KAAK,EAAE,SAAS;oBAChB,MAAM,EAAE,UAAU;oBAClB,MAAM,EAAE,UAAU;oBAClB,OAAO;oBACP,WAAW,EAAE,MAAM,CAAC,WAAW,IAAI,GAAG;iBACvC;aACF,CAAC;QAER,MAAM,gBAAgB,GAAG;YACvB,KAAK,EAAE,WAAW;YAClB,UAAU,EAAE,MAAM,CAAC,UAAU,IAAI,CAAC;YAClC,QAAQ,EAAE,KAAK;YACf,eAAe,EAAE,CAAC;SACnB,CAAC;QAEF,IAAI,CAAC,GAAG,GAAG,IAAI,MAAM,CAAC,UAAU,CAAC,gBAAgB,CAAC,CAAC;QACnD,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,EAAE,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,QAAQ,EAAE,MAAM,CAAC,QAAQ,EAAE,EACpD,gCAAgC,CACjC,CAAC;IACJ,CAAC;IAED,KAAK,CAAC,gBAAgB,CAAC,IAAY;QACjC,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;QAC5B,IAAI,CAAC,OAAO,EAAE,CAAC;YACb,MAAM,IAAI,KAAK,CAAC,8BAA8B,CAAC,CAAC;QAClD,CAAC;QAED,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC;YAC9B,IAAI,EAAE,OAAO;YACb,GAAG,EAAE,IAAI,CAAC,SAAS;YACnB,KAAK,EAAE,IAAI,CAAC,KAAK;YACjB,gFAAgF;YAChF,0EAA0E;YAC1E,oBAAoB,EAAE,KAAK;SAC5B,CAAC,CAAC;QACH,MAAM,UAAU,GACd,KAAK,IAAI,KAAK,CAAC,OAAO,YAAY,YAAY;YAC5C,CAAC,CAAC,KAAK,CAAC,OAAO;YACf,CAAC,CAAC,KAAK,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,OAAO,CAAC;gBACrC,CAAC,CAAC,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,OAAmB,CAAC;gBAC9C,CAAC,CAAC,IAAI,CAAC;QACb,oEAAoE;QACpE,kDAAkD;QAClD,MAAM,OAAO,GAAG,UAAU,CAAC,CAAC,CAAC,YAAY,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;QAClE,MAAM,UAAU,GACd,KAAK;YACL,OAAO,KAAK,CAAC,UAAU,KAAK,QAAQ;YACpC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,UAAU,CAAC;YACjC,KAAK,CAAC,UAAU,GAAG,CAAC;YAClB,CAAC,CAAC,KAAK,CAAC,UAAU;YAClB,CAAC,CAAC,OAAO,IAAI,CAAC,GAAG,CAAC,UAAU,KAAK,QAAQ;gBACvC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,UAAU;gBACrB,CAAC,CAAC,KAAK,CAAC;QAEd,IAAI,CAAC,OAAO,EAAE,CAAC;YACb,MAAM,IAAI,KAAK,CAAC,uDAAuD,CAAC,CAAC;QAC3E,CAAC;QAED,MAAM,KAAK,GAAG,gBAAgB,CAAC,OAAO,CAAC,CAAC;QACxC,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ;QAC3E,MAAM,MAAM,GAAG,WAAW,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC;QAE9C,OAAO;YACL,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC;YAC7B,MAAM,EAAE,YAAY,UAAU,EAAE;SACjC,CAAC;IACJ,CAAC;IAED,IAAI;QACF,IAAI,CAAC;YACH,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,CAAC;QACrB,CAAC;QAAC,MAAM,CAAC;YACP,SAAS;QACX,CAAC;IACH,CAAC;CACF"}

package/dist/server/server/speech/providers/local/sherpa/silero-vad-provider.d.ts

@@ -0,0 +1,19 @@
+import type { Logger } from "pino";
+import type { TurnDetectionProvider, TurnDetectionSession } from "../../../turn-detection-provider.js";
+import { type SherpaSileroVadSessionConfig } from "./silero-vad-session.js";
+/**
+ * Ensure the Silero VAD ONNX model exists in modelsDir where native code can
+ * read it. The bundled asset lives inside Electron's app.asar which native C++
+ * cannot open, so we copy it out on first run using Node.js fs (asar-aware).
+ */
+export declare function ensureSileroVadModel(modelsDir: string, logger: Logger): Promise<string>;
+export declare class SherpaSileroTurnDetectionProvider implements TurnDetectionProvider {
+    readonly id: "local";
+    private readonly config;
+    private readonly logger;
+    constructor(config: SherpaSileroVadSessionConfig, logger: Logger);
+    createSession(params: {
+        logger: Logger;
+    }): TurnDetectionSession;
+}
+//# sourceMappingURL=silero-vad-provider.d.ts.map

package/dist/server/server/speech/providers/local/sherpa/silero-vad-provider.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"silero-vad-provider.d.ts","sourceRoot":"","sources":["../../../../../../../src/server/speech/providers/local/sherpa/silero-vad-provider.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,MAAM,CAAC;AAEnC,OAAO,KAAK,EACV,qBAAqB,EACrB,oBAAoB,EACrB,MAAM,qCAAqC,CAAC;AAC7C,OAAO,EAGL,KAAK,4BAA4B,EAClC,MAAM,yBAAyB,CAAC;AAKjC;;;;GAIG;AACH,wBAAsB,oBAAoB,CAAC,SAAS,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAgB7F;AAED,qBAAa,iCAAkC,YAAW,qBAAqB;IAC7E,SAAgB,EAAE,EAAG,OAAO,CAAU;IAEtC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAA+B;IACtD,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;gBAEpB,MAAM,EAAE,4BAA4B,EAAE,MAAM,EAAE,MAAM;IAShE,aAAa,CAAC,MAAM,EAAE;QAAE,MAAM,EAAE,MAAM,CAAA;KAAE,GAAG,oBAAoB;CAahE"}

package/dist/server/server/speech/providers/local/sherpa/silero-vad-provider.js

@@ -0,0 +1,49 @@
+import { copyFile, mkdir, stat } from "node:fs/promises";
+import path from "node:path";
+import { resolveBundledSileroVadModelPath, SherpaSileroVadSession, } from "./silero-vad-session.js";
+const SILERO_VAD_DIR = "silero-vad";
+const SILERO_VAD_FILE = "silero_vad.onnx";
+/**
+ * Ensure the Silero VAD ONNX model exists in modelsDir where native code can
+ * read it. The bundled asset lives inside Electron's app.asar which native C++
+ * cannot open, so we copy it out on first run using Node.js fs (asar-aware).
+ */
+export async function ensureSileroVadModel(modelsDir, logger) {
+    const destDir = path.join(modelsDir, SILERO_VAD_DIR);
+    const destPath = path.join(destDir, SILERO_VAD_FILE);
+    try {
+        const s = await stat(destPath);
+        if (s.isFile() && s.size > 0)
+            return destPath;
+    }
+    catch {
+        // not present yet
+    }
+    const bundledPath = resolveBundledSileroVadModelPath();
+    await mkdir(destDir, { recursive: true });
+    await copyFile(bundledPath, destPath);
+    logger.info({ destPath }, "Copied Silero VAD model to models directory");
+    return destPath;
+}
+export class SherpaSileroTurnDetectionProvider {
+    constructor(config, logger) {
+        this.id = "local";
+        this.config = config;
+        this.logger = logger.child({
+            module: "speech",
+            provider: "local",
+            component: "silero-vad",
+        });
+    }
+    createSession(params) {
+        this.logger.debug({ sampleRate: this.config.sampleRate, modelPath: this.config.modelPath }, "Creating Silero VAD turn-detection session");
+        return new SherpaSileroVadSession({
+            logger: params.logger.child({
+                provider: "local",
+                component: "silero-vad-session",
+            }),
+            config: this.config,
+        });
+    }
+}
+//# sourceMappingURL=silero-vad-provider.js.map

package/dist/server/server/speech/providers/local/sherpa/silero-vad-provider.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"silero-vad-provider.js","sourceRoot":"","sources":["../../../../../../../src/server/speech/providers/local/sherpa/silero-vad-provider.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,kBAAkB,CAAC;AACzD,OAAO,IAAI,MAAM,WAAW,CAAC;AAQ7B,OAAO,EACL,gCAAgC,EAChC,sBAAsB,GAEvB,MAAM,yBAAyB,CAAC;AAEjC,MAAM,cAAc,GAAG,YAAY,CAAC;AACpC,MAAM,eAAe,GAAG,iBAAiB,CAAC;AAE1C;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,oBAAoB,CAAC,SAAiB,EAAE,MAAc;IAC1E,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,cAAc,CAAC,CAAC;IACrD,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,eAAe,CAAC,CAAC;IAErD,IAAI,CAAC;QACH,MAAM,CAAC,GAAG,MAAM,IAAI,CAAC,QAAQ,CAAC,CAAC;QAC/B,IAAI,CAAC,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,IAAI,GAAG,CAAC;YAAE,OAAO,QAAQ,CAAC;IAChD,CAAC;IAAC,MAAM,CAAC;QACP,kBAAkB;IACpB,CAAC;IAED,MAAM,WAAW,GAAG,gCAAgC,EAAE,CAAC;IACvD,MAAM,KAAK,CAAC,OAAO,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;IAC1C,MAAM,QAAQ,CAAC,WAAW,EAAE,QAAQ,CAAC,CAAC;IACtC,MAAM,CAAC,IAAI,CAAC,EAAE,QAAQ,EAAE,EAAE,6CAA6C,CAAC,CAAC;IACzE,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED,MAAM,OAAO,iCAAiC;IAM5C,YAAY,MAAoC,EAAE,MAAc;QALhD,OAAE,GAAG,OAAgB,CAAC;QAMpC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC;YACzB,MAAM,EAAE,QAAQ;YAChB,QAAQ,EAAE,OAAO;YACjB,SAAS,EAAE,YAAY;SACxB,CAAC,CAAC;IACL,CAAC;IAED,aAAa,CAAC,MAA0B;QACtC,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,EAAE,UAAU,EAAE,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,SAAS,EAAE,IAAI,CAAC,MAAM,CAAC,SAAS,EAAE,EACxE,4CAA4C,CAC7C,CAAC;QACF,OAAO,IAAI,sBAAsB,CAAC;YAChC,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC;gBAC1B,QAAQ,EAAE,OAAO;gBACjB,SAAS,EAAE,oBAAoB;aAChC,CAAC;YACF,MAAM,EAAE,IAAI,CAAC,MAAM;SACpB,CAAC,CAAC;IACL,CAAC;CACF"}

package/dist/server/server/speech/providers/local/sherpa/silero-vad-session.d.ts

@@ -0,0 +1,38 @@
+import { EventEmitter } from "node:events";
+import type { TurnDetectionSession } from "../../../turn-detection-provider.js";
+export declare function resolveBundledSileroVadModelPath(): string;
+export interface SherpaSileroVadSessionConfig {
+    modelPath?: string;
+    sampleRate?: number;
+    threshold?: number;
+    windowSize?: number;
+    bufferSizeInSeconds?: number;
+    confirmMs?: number;
+    silenceMs?: number;
+}
+export declare class SherpaSileroVadSession extends EventEmitter implements TurnDetectionSession {
+    readonly requiredSampleRate: number;
+    private readonly vad;
+    private readonly inputBuffer;
+    private readonly windowSize;
+    private readonly msPerWindow;
+    private readonly confirmMs;
+    private readonly silenceMs;
+    private connected;
+    private phase;
+    private windowTimestamp;
+    private readonly logger;
+    constructor(params: {
+        logger: {
+            debug: (...args: unknown[]) => void;
+        };
+        config?: SherpaSileroVadSessionConfig;
+    });
+    connect(): Promise<void>;
+    appendPcm16(pcm16le: Buffer): void;
+    flush(): void;
+    reset(): void;
+    close(): void;
+    private stepStateMachine;
+}
+//# sourceMappingURL=silero-vad-session.d.ts.map

package/dist/server/server/speech/providers/local/sherpa/silero-vad-session.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"silero-vad-session.d.ts","sourceRoot":"","sources":["../../../../../../../src/server/speech/providers/local/sherpa/silero-vad-session.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAI3C,OAAO,KAAK,EAAE,oBAAoB,EAAE,MAAM,qCAAqC,CAAC;AAuChF,wBAAgB,gCAAgC,IAAI,MAAM,CAEzD;AAED,MAAM,WAAW,4BAA4B;IAC3C,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,SAAS,CAAC,EAAE,MAAM,CAAC;CACpB;AAQD,qBAAa,sBAAuB,SAAQ,YAAa,YAAW,oBAAoB;IACtF,SAAgB,kBAAkB,EAAE,MAAM,CAAC;IAE3C,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAkB;IACtC,OAAO,CAAC,QAAQ,CAAC,WAAW,CAA6B;IACzD,OAAO,CAAC,QAAQ,CAAC,UAAU,CAAS;IACpC,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAS;IACrC,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAS;IACnC,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAS;IACnC,OAAO,CAAC,SAAS,CAAS;IAC1B,OAAO,CAAC,KAAK,CAA+B;IAC5C,OAAO,CAAC,eAAe,CAAK;IAC5B,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAC;gBAEZ,MAAM,EAAE;QAClB,MAAM,EAAE;YAAE,KAAK,EAAE,CAAC,GAAG,IAAI,EAAE,OAAO,EAAE,KAAK,IAAI,CAAA;SAAE,CAAC;QAChD,MAAM,CAAC,EAAE,4BAA4B,CAAC;KACvC;IAgDK,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC;IAI9B,WAAW,CAAC,OAAO,EAAE,MAAM,GAAG,IAAI;IA0BlC,KAAK,IAAI,IAAI;IAsBb,KAAK,IAAI,IAAI;IAWb,KAAK,IAAI,IAAI;IAMb,OAAO,CAAC,gBAAgB;CAiEzB"}

package/dist/server/server/speech/providers/local/sherpa/silero-vad-session.js

@@ -0,0 +1,176 @@
+import { EventEmitter } from "node:events";
+import { fileURLToPath } from "node:url";
+import { pcm16leToFloat32 } from "../../../audio.js";
+import { loadSherpaOnnxNode } from "./sherpa-onnx-node-loader.js";
+const DEFAULT_SAMPLE_RATE = 16000;
+const DEFAULT_BUFFER_SIZE_SECONDS = 60;
+const DEFAULT_SILERO_THRESHOLD = 0.5;
+const DEFAULT_WINDOW_SIZE = 512;
+// Silero internal durations — kept low so isDetected() tracks actual sound.
+// Our own state machine handles speech confirmation and end-of-speech detection.
+const SILERO_MIN_SILENCE_DURATION = 0.2;
+const SILERO_MIN_SPEECH_DURATION = 0.1;
+// Our boundary detection thresholds (in milliseconds).
+const DEFAULT_CONFIRM_MS = 800;
+const DEFAULT_SILENCE_MS = 1000;
+export function resolveBundledSileroVadModelPath() {
+    return fileURLToPath(new URL("./assets/silero_vad.onnx", import.meta.url));
+}
+export class SherpaSileroVadSession extends EventEmitter {
+    constructor(params) {
+        super();
+        this.connected = false;
+        this.phase = { state: "idle" };
+        this.windowTimestamp = 0;
+        this.logger = params.logger;
+        const config = params.config ?? {};
+        this.requiredSampleRate = config.sampleRate ?? DEFAULT_SAMPLE_RATE;
+        this.windowSize = config.windowSize ?? DEFAULT_WINDOW_SIZE;
+        this.msPerWindow = (this.windowSize / this.requiredSampleRate) * 1000;
+        this.confirmMs = config.confirmMs ?? DEFAULT_CONFIRM_MS;
+        this.silenceMs = config.silenceMs ?? DEFAULT_SILENCE_MS;
+        const threshold = config.threshold ?? DEFAULT_SILERO_THRESHOLD;
+        this.logger.debug({
+            threshold,
+            sileroMinSilenceDuration: SILERO_MIN_SILENCE_DURATION,
+            sileroMinSpeechDuration: SILERO_MIN_SPEECH_DURATION,
+            confirmMs: this.confirmMs,
+            silenceMs: this.silenceMs,
+            windowSize: this.windowSize,
+            msPerWindow: this.msPerWindow,
+            sampleRate: this.requiredSampleRate,
+        }, "[VAD] Initializing Silero VAD session");
+        const sherpa = loadSherpaOnnxNode();
+        this.vad = new sherpa.Vad({
+            sileroVad: {
+                model: config.modelPath ?? resolveBundledSileroVadModelPath(),
+                threshold,
+                minSilenceDuration: SILERO_MIN_SILENCE_DURATION,
+                minSpeechDuration: SILERO_MIN_SPEECH_DURATION,
+                windowSize: this.windowSize,
+            },
+            sampleRate: this.requiredSampleRate,
+            numThreads: 1,
+            provider: "cpu",
+            debug: 0,
+        }, config.bufferSizeInSeconds ?? DEFAULT_BUFFER_SIZE_SECONDS);
+        this.inputBuffer = new sherpa.CircularBuffer((config.bufferSizeInSeconds ?? DEFAULT_BUFFER_SIZE_SECONDS) * this.requiredSampleRate);
+    }
+    async connect() {
+        this.connected = true;
+    }
+    appendPcm16(pcm16le) {
+        if (!this.connected) {
+            this.emit("error", new Error("Turn detection session not connected"));
+            return;
+        }
+        if (pcm16le.length === 0) {
+            return;
+        }
+        try {
+            const samples = pcm16leToFloat32(pcm16le, 1);
+            this.inputBuffer.push(samples);
+            let windowsProcessed = 0;
+            while (this.inputBuffer.size() > this.windowSize) {
+                const window = this.inputBuffer.get(this.inputBuffer.head(), this.windowSize, false);
+                this.inputBuffer.pop(this.windowSize);
+                this.vad.acceptWaveform(window);
+                this.windowTimestamp += this.msPerWindow;
+                windowsProcessed++;
+                this.stepStateMachine();
+            }
+        }
+        catch (error) {
+            this.emit("error", error instanceof Error ? error : new Error(String(error)));
+        }
+    }
+    flush() {
+        if (!this.connected) {
+            return;
+        }
+        try {
+            this.logger.debug({ phase: this.phase.state }, "[VAD] Flushing remaining audio");
+            this.vad.flush();
+            this.stepStateMachine();
+            if (this.phase.state === "speaking" || this.phase.state === "ending") {
+                this.logger.debug("[VAD] Forcing speech_stopped after flush");
+                this.phase = { state: "idle" };
+                this.emit("speech_stopped");
+            }
+            else if (this.phase.state === "confirming") {
+                this.logger.debug("[VAD] Discarding unconfirmed speech on flush");
+                this.phase = { state: "idle" };
+            }
+        }
+        catch (error) {
+            this.emit("error", error instanceof Error ? error : new Error(String(error)));
+        }
+    }
+    reset() {
+        try {
+            this.vad.reset();
+            this.inputBuffer.reset();
+        }
+        catch {
+            // ignore native cleanup failures
+        }
+        finally {
+            this.phase = { state: "idle" };
+        }
+    }
+    close() {
+        this.reset();
+        this.connected = false;
+        this.windowTimestamp = 0;
+    }
+    stepStateMachine() {
+        const detected = this.vad.isDetected();
+        const now = this.windowTimestamp;
+        switch (this.phase.state) {
+            case "idle": {
+                if (detected) {
+                    this.logger.debug({ now }, "[VAD] idle → confirming (detection started)");
+                    this.phase = { state: "confirming", startedAt: now };
+                }
+                break;
+            }
+            case "confirming": {
+                if (!detected) {
+                    const elapsed = now - this.phase.startedAt;
+                    this.logger.debug({ elapsed, confirmMs: this.confirmMs }, "[VAD] confirming → idle (detection dropped before confirmation)");
+                    this.phase = { state: "idle" };
+                    break;
+                }
+                const elapsed = now - this.phase.startedAt;
+                if (elapsed >= this.confirmMs) {
+                    this.logger.debug({ elapsed, confirmMs: this.confirmMs }, "[VAD] confirming → speaking (speech confirmed)");
+                    this.phase = { state: "speaking" };
+                    this.emit("speech_started");
+                }
+                break;
+            }
+            case "speaking": {
+                if (!detected) {
+                    this.logger.debug({ now }, "[VAD] speaking → ending (silence started)");
+                    this.phase = { state: "ending", startedAt: now };
+                }
+                break;
+            }
+            case "ending": {
+                if (detected) {
+                    this.logger.debug({ elapsed: now - this.phase.startedAt }, "[VAD] ending → speaking (speech resumed)");
+                    this.phase = { state: "speaking" };
+                    break;
+                }
+                const elapsed = now - this.phase.startedAt;
+                if (elapsed >= this.silenceMs) {
+                    this.logger.debug({ elapsed, silenceMs: this.silenceMs }, "[VAD] ending → idle (speech stopped)");
+                    this.phase = { state: "idle" };
+                    this.emit("speech_stopped");
+                }
+                break;
+            }
+        }
+    }
+}
+//# sourceMappingURL=silero-vad-session.js.map

package/dist/server/server/speech/providers/local/sherpa/silero-vad-session.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"silero-vad-session.js","sourceRoot":"","sources":["../../../../../../../src/server/speech/providers/local/sherpa/silero-vad-session.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAC3C,OAAO,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AAEzC,OAAO,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAC;AAErD,OAAO,EAAE,kBAAkB,EAAE,MAAM,8BAA8B,CAAC;AAElE,MAAM,mBAAmB,GAAG,KAAK,CAAC;AAClC,MAAM,2BAA2B,GAAG,EAAE,CAAC;AACvC,MAAM,wBAAwB,GAAG,GAAG,CAAC;AACrC,MAAM,mBAAmB,GAAG,GAAG,CAAC;AAEhC,4EAA4E;AAC5E,iFAAiF;AACjF,MAAM,2BAA2B,GAAG,GAAG,CAAC;AACxC,MAAM,0BAA0B,GAAG,GAAG,CAAC;AAEvC,uDAAuD;AACvD,MAAM,kBAAkB,GAAG,GAAG,CAAC;AAC/B,MAAM,kBAAkB,GAAG,IAAI,CAAC;AAwBhC,MAAM,UAAU,gCAAgC;IAC9C,OAAO,aAAa,CAAC,IAAI,GAAG,CAAC,0BAA0B,EAAE,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;AAC7E,CAAC;AAkBD,MAAM,OAAO,sBAAuB,SAAQ,YAAY;IActD,YAAY,MAGX;QACC,KAAK,EAAE,CAAC;QATF,cAAS,GAAG,KAAK,CAAC;QAClB,UAAK,GAAa,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;QACpC,oBAAe,GAAG,CAAC,CAAC;QAQ1B,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC;QAC5B,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,EAAE,CAAC;QACnC,IAAI,CAAC,kBAAkB,GAAG,MAAM,CAAC,UAAU,IAAI,mBAAmB,CAAC;QACnE,IAAI,CAAC,UAAU,GAAG,MAAM,CAAC,UAAU,IAAI,mBAAmB,CAAC;QAC3D,IAAI,CAAC,WAAW,GAAG,CAAC,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,kBAAkB,CAAC,GAAG,IAAI,CAAC;QACtE,IAAI,CAAC,SAAS,GAAG,MAAM,CAAC,SAAS,IAAI,kBAAkB,CAAC;QACxD,IAAI,CAAC,SAAS,GAAG,MAAM,CAAC,SAAS,IAAI,kBAAkB,CAAC;QAExD,MAAM,SAAS,GAAG,MAAM,CAAC,SAAS,IAAI,wBAAwB,CAAC;QAE/D,IAAI,CAAC,MAAM,CAAC,KAAK,CACf;YACE,SAAS;YACT,wBAAwB,EAAE,2BAA2B;YACrD,uBAAuB,EAAE,0BAA0B;YACnD,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,UAAU,EAAE,IAAI,CAAC,UAAU;YAC3B,WAAW,EAAE,IAAI,CAAC,WAAW;YAC7B,UAAU,EAAE,IAAI,CAAC,kBAAkB;SACpC,EACD,uCAAuC,CACxC,CAAC;QAEF,MAAM,MAAM,GAAG,kBAAkB,EAAgC,CAAC;QAClE,IAAI,CAAC,GAAG,GAAG,IAAI,MAAM,CAAC,GAAG,CACvB;YACE,SAAS,EAAE;gBACT,KAAK,EAAE,MAAM,CAAC,SAAS,IAAI,gCAAgC,EAAE;gBAC7D,SAAS;gBACT,kBAAkB,EAAE,2BAA2B;gBAC/C,iBAAiB,EAAE,0BAA0B;gBAC7C,UAAU,EAAE,IAAI,CAAC,UAAU;aAC5B;YACD,UAAU,EAAE,IAAI,CAAC,kBAAkB;YACnC,UAAU,EAAE,CAAC;YACb,QAAQ,EAAE,KAAK;YACf,KAAK,EAAE,CAAC;SACT,EACD,MAAM,CAAC,mBAAmB,IAAI,2BAA2B,CAC1D,CAAC;QACF,IAAI,CAAC,WAAW,GAAG,IAAI,MAAM,CAAC,cAAc,CAC1C,CAAC,MAAM,CAAC,mBAAmB,IAAI,2BAA2B,CAAC,GAAG,IAAI,CAAC,kBAAkB,CACtF,CAAC;IACJ,CAAC;IAED,KAAK,CAAC,OAAO;QACX,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC;IACxB,CAAC;IAED,WAAW,CAAC,OAAe;QACzB,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,CAAC;YACpB,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC,CAAC;YACtE,OAAO;QACT,CAAC;QACD,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACzB,OAAO;QACT,CAAC;QAED,IAAI,CAAC;YACH,MAAM,OAAO,GAAG,gBAAgB,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;YAC7C,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YAC/B,IAAI,gBAAgB,GAAG,CAAC,CAAC;YACzB,OAAO,IAAI,CAAC,WAAW,CAAC,IAAI,EAAE,GAAG,IAAI,CAAC,UAAU,EAAE,CAAC;gBACjD,MAAM,MAAM,GAAG,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,CAAC,IAAI,EAAE,EAAE,IAAI,CAAC,UAAU,EAAE,KAAK,CAAC,CAAC;gBACrF,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;gBACtC,IAAI,CAAC,GAAG,CAAC,cAAc,CAAC,MAAM,CAAC,CAAC;gBAChC,IAAI,CAAC,eAAe,IAAI,IAAI,CAAC,WAAW,CAAC;gBACzC,gBAAgB,EAAE,CAAC;gBACnB,IAAI,CAAC,gBAAgB,EAAE,CAAC;YAC1B,CAAC;QACH,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QAChF,CAAC;IACH,CAAC;IAED,KAAK;QACH,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,CAAC;YACpB,OAAO;QACT,CAAC;QAED,IAAI,CAAC;YACH,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,EAAE,gCAAgC,CAAC,CAAC;YACjF,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC;YACjB,IAAI,CAAC,gBAAgB,EAAE,CAAC;YACxB,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,KAAK,UAAU,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,KAAK,QAAQ,EAAE,CAAC;gBACrE,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,0CAA0C,CAAC,CAAC;gBAC9D,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;gBAC/B,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;YAC9B,CAAC;iBAAM,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,KAAK,YAAY,EAAE,CAAC;gBAC7C,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,8CAA8C,CAAC,CAAC;gBAClE,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;YACjC,CAAC;QACH,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QAChF,CAAC;IACH,CAAC;IAED,KAAK;QACH,IAAI,CAAC;YACH,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC;YACjB,IAAI,CAAC,WAAW,CAAC,KAAK,EAAE,CAAC;QAC3B,CAAC;QAAC,MAAM,CAAC;YACP,iCAAiC;QACnC,CAAC;gBAAS,CAAC;YACT,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;QACjC,CAAC;IACH,CAAC;IAED,KAAK;QACH,IAAI,CAAC,KAAK,EAAE,CAAC;QACb,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC;QACvB,IAAI,CAAC,eAAe,GAAG,CAAC,CAAC;IAC3B,CAAC;IAEO,gBAAgB;QACtB,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,UAAU,EAAE,CAAC;QACvC,MAAM,GAAG,GAAG,IAAI,CAAC,eAAe,CAAC;QAEjC,QAAQ,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;YACzB,KAAK,MAAM,CAAC,CAAC,CAAC;gBACZ,IAAI,QAAQ,EAAE,CAAC;oBACb,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,EAAE,EAAE,6CAA6C,CAAC,CAAC;oBAC1E,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,YAAY,EAAE,SAAS,EAAE,GAAG,EAAE,CAAC;gBACvD,CAAC;gBACD,MAAM;YACR,CAAC;YAED,KAAK,YAAY,CAAC,CAAC,CAAC;gBAClB,IAAI,CAAC,QAAQ,EAAE,CAAC;oBACd,MAAM,OAAO,GAAG,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC;oBAC3C,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,EAAE,OAAO,EAAE,SAAS,EAAE,IAAI,CAAC,SAAS,EAAE,EACtC,iEAAiE,CAClE,CAAC;oBACF,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;oBAC/B,MAAM;gBACR,CAAC;gBACD,MAAM,OAAO,GAAG,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC;gBAC3C,IAAI,OAAO,IAAI,IAAI,CAAC,SAAS,EAAE,CAAC;oBAC9B,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,EAAE,OAAO,EAAE,SAAS,EAAE,IAAI,CAAC,SAAS,EAAE,EACtC,gDAAgD,CACjD,CAAC;oBACF,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC;oBACnC,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;gBAC9B,CAAC;gBACD,MAAM;YACR,CAAC;YAED,KAAK,UAAU,CAAC,CAAC,CAAC;gBAChB,IAAI,CAAC,QAAQ,EAAE,CAAC;oBACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,EAAE,EAAE,2CAA2C,CAAC,CAAC;oBACxE,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,QAAQ,EAAE,SAAS,EAAE,GAAG,EAAE,CAAC;gBACnD,CAAC;gBACD,MAAM;YACR,CAAC;YAED,KAAK,QAAQ,CAAC,CAAC,CAAC;gBACd,IAAI,QAAQ,EAAE,CAAC;oBACb,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,EAAE,OAAO,EAAE,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,EAAE,EACvC,0CAA0C,CAC3C,CAAC;oBACF,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC;oBACnC,MAAM;gBACR,CAAC;gBACD,MAAM,OAAO,GAAG,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC;gBAC3C,IAAI,OAAO,IAAI,IAAI,CAAC,SAAS,EAAE,CAAC;oBAC9B,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,EAAE,OAAO,EAAE,SAAS,EAAE,IAAI,CAAC,SAAS,EAAE,EACtC,sCAAsC,CACvC,CAAC;oBACF,IAAI,CAAC,KAAK,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;oBAC/B,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;gBAC9B,CAAC;gBACD,MAAM;YACR,CAAC;QACH,CAAC;IACH,CAAC;CACF"}

package/dist/server/server/speech/providers/openai/config.d.ts

@@ -0,0 +1,22 @@
+import type { PersistedConfig } from "../../../persisted-config.js";
+import type { RequestedSpeechProviders } from "../../speech-types.js";
+import type { STTConfig } from "./stt.js";
+import type { TTSConfig } from "./tts.js";
+export declare const DEFAULT_OPENAI_REALTIME_TRANSCRIPTION_MODEL = "gpt-4o-transcribe";
+export declare const DEFAULT_OPENAI_TTS_MODEL = "tts-1";
+export type OpenAiSpeechProviderConfig = {
+    apiKey?: string;
+    stt?: Partial<STTConfig> & {
+        apiKey?: string;
+    };
+    tts?: Partial<TTSConfig> & {
+        apiKey?: string;
+    };
+    realtimeTranscriptionModel?: string;
+};
+export declare function resolveOpenAiSpeechConfig(params: {
+    env: NodeJS.ProcessEnv;
+    persisted: PersistedConfig;
+    providers: RequestedSpeechProviders;
+}): OpenAiSpeechProviderConfig | undefined;
+//# sourceMappingURL=config.d.ts.map

package/dist/server/server/speech/providers/openai/config.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"config.d.ts","sourceRoot":"","sources":["../../../../../../src/server/speech/providers/openai/config.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,8BAA8B,CAAC;AACpE,OAAO,KAAK,EAAE,wBAAwB,EAAE,MAAM,uBAAuB,CAAC;AACtE,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,UAAU,CAAC;AAC1C,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,UAAU,CAAC;AAE1C,eAAO,MAAM,2CAA2C,sBAAsB,CAAC;AAC/E,eAAO,MAAM,wBAAwB,UAAU,CAAC;AAEhD,MAAM,MAAM,0BAA0B,GAAG;IACvC,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,GAAG,CAAC,EAAE,OAAO,CAAC,SAAS,CAAC,GAAG;QAAE,MAAM,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC;IAC/C,GAAG,CAAC,EAAE,OAAO,CAAC,SAAS,CAAC,GAAG;QAAE,MAAM,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC;IAC/C,0BAA0B,CAAC,EAAE,MAAM,CAAC;CACrC,CAAC;AAgCF,wBAAgB,yBAAyB,CAAC,MAAM,EAAE;IAChD,GAAG,EAAE,MAAM,CAAC,UAAU,CAAC;IACvB,SAAS,EAAE,eAAe,CAAC;IAC3B,SAAS,EAAE,wBAAwB,CAAC;CACrC,GAAG,0BAA0B,GAAG,SAAS,CA4DzC"}