daemora 1.0.0-alpha.0

package/dist/voice/stt/GroqSTT.js

@@ -0,0 +1,169 @@
+/**
+ * GroqSTT — batch speech-to-text via Groq's OpenAI-compatible API.
+ *
+ * NOT streaming. Collects PCM16 audio chunks into a buffer, then
+ * transcribes on flush() or when silence is detected (energy-based).
+ * Good as a fallback when a streaming provider is unavailable.
+ */
+import { createLogger } from "../../util/logger.js";
+import { STTAdapter, } from "./STTAdapter.js";
+const log = createLogger("voice:stt:groq");
+const GROQ_URL = "https://api.groq.com/openai/v1/audio/transcriptions";
+const SILENCE_THRESHOLD_MS = 1000; // 1s silence = user done talking
+const ENERGY_FLOOR = 200; // RMS threshold for "silence"
+export class GroqSTT extends STTAdapter {
+    id = "groq";
+    streaming = false;
+    config = null;
+    chunks = [];
+    totalBytes = 0;
+    silenceStart = null;
+    sampleRate = 16000;
+    flushing = false;
+    closed = false;
+    speaking = false;
+    // ── Connect / Disconnect ──────────────────────────────────────────
+    async connect(config) {
+        this.config = config;
+        this.sampleRate = config.sampleRate ?? 16000;
+        this.closed = false;
+        this.chunks = [];
+        this.totalBytes = 0;
+        this.speaking = false;
+        this.silenceStart = null;
+        log.info("ready (batch mode)");
+    }
+    async disconnect() {
+        this.closed = true;
+        // Flush remaining audio before closing
+        if (this.totalBytes > 0) {
+            await this.flush();
+        }
+        this.chunks = [];
+        this.totalBytes = 0;
+        this.emit("close");
+    }
+    // ── Audio input ───────────────────────────────────────────────────
+    sendAudio(pcm16) {
+        if (this.closed)
+            return;
+        this.chunks.push(pcm16);
+        this.totalBytes += pcm16.length;
+        // Simple energy-based silence detection
+        const rms = this.computeRMS(pcm16);
+        if (rms < ENERGY_FLOOR) {
+            if (!this.silenceStart) {
+                this.silenceStart = Date.now();
+            }
+            else if (Date.now() - this.silenceStart >= SILENCE_THRESHOLD_MS && this.totalBytes > 0) {
+                // Silence exceeded threshold — auto-flush
+                this.silenceStart = null;
+                this.flush().catch((err) => {
+                    log.error({ err }, "auto-flush failed");
+                });
+            }
+        }
+        else {
+            // Voice activity
+            if (this.silenceStart) {
+                this.emit("speech_started");
+            }
+            this.silenceStart = null;
+        }
+    }
+    // ── Flush (trigger transcription) ────────────────────────────────
+    async flush() {
+        if (this.flushing || this.totalBytes === 0)
+            return;
+        this.flushing = true;
+        const audioBuffer = Buffer.concat(this.chunks);
+        this.chunks = [];
+        this.totalBytes = 0;
+        try {
+            const wav = this.pcmToWav(audioBuffer, this.sampleRate);
+            const transcript = await this.transcribe(wav);
+            if (transcript) {
+                const event = {
+                    text: transcript,
+                    isFinal: true,
+                    confidence: 1.0, // Groq doesn't return confidence
+                    speechFinal: true,
+                };
+                this.emit("transcript", event);
+                this.emit("speech_ended");
+                this.emit("utterance_end");
+            }
+        }
+        catch (err) {
+            const error = err instanceof Error ? err : new Error(String(err));
+            log.error({ err: error }, "transcription failed");
+            this.emit("error", error);
+        }
+        finally {
+            this.flushing = false;
+        }
+    }
+    // ── HTTP transcription ────────────────────────────────────────────
+    async transcribe(wav) {
+        const apiKey = this.config.apiKey;
+        // Use native FormData (Node 22+) — no manual multipart needed
+        const formData = new FormData();
+        formData.append("file", new Blob([wav], { type: "audio/wav" }), "audio.wav");
+        formData.append("model", "whisper-large-v3-turbo");
+        const lang = this.config?.language === "multi" ? "en" : this.config?.language;
+        if (lang)
+            formData.append("language", lang);
+        const res = await fetch(GROQ_URL, {
+            method: "POST",
+            headers: { Authorization: `Bearer ${apiKey}` },
+            body: formData,
+            signal: AbortSignal.timeout(15_000),
+        });
+        if (!res.ok) {
+            const text = await res.text();
+            throw new Error(`Groq STT ${res.status}: ${text}`);
+        }
+        const json = (await res.json());
+        return (json.text ?? "").trim();
+    }
+    // ── PCM16 → WAV conversion ───────────────────────────────────────
+    pcmToWav(pcm, sampleRate) {
+        const numChannels = 1;
+        const bitsPerSample = 16;
+        const byteRate = sampleRate * numChannels * (bitsPerSample / 8);
+        const blockAlign = numChannels * (bitsPerSample / 8);
+        const dataSize = pcm.length;
+        const headerSize = 44;
+        const header = Buffer.alloc(headerSize);
+        // RIFF header
+        header.write("RIFF", 0);
+        header.writeUInt32LE(dataSize + headerSize - 8, 4);
+        header.write("WAVE", 8);
+        // fmt sub-chunk
+        header.write("fmt ", 12);
+        header.writeUInt32LE(16, 16); // sub-chunk size
+        header.writeUInt16LE(1, 20); // PCM format
+        header.writeUInt16LE(numChannels, 22);
+        header.writeUInt32LE(sampleRate, 24);
+        header.writeUInt32LE(byteRate, 28);
+        header.writeUInt16LE(blockAlign, 32);
+        header.writeUInt16LE(bitsPerSample, 34);
+        // data sub-chunk
+        header.write("data", 36);
+        header.writeUInt32LE(dataSize, 40);
+        return Buffer.concat([header, pcm]);
+    }
+    // ── Energy detection ──────────────────────────────────────────────
+    computeRMS(pcm16) {
+        if (pcm16.length < 2)
+            return 0;
+        let sumSquares = 0;
+        const sampleCount = Math.floor(pcm16.length / 2);
+        for (let i = 0; i < sampleCount; i++) {
+            const sample = pcm16.readInt16LE(i * 2);
+            sumSquares += sample * sample;
+        }
+        return Math.sqrt(sumSquares / sampleCount);
+    }
+}
+//# sourceMappingURL=GroqSTT.js.map

package/dist/voice/stt/GroqSTT.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"GroqSTT.js","sourceRoot":"","sources":["../../../src/voice/stt/GroqSTT.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAC;AACpD,OAAO,EACL,UAAU,GAGX,MAAM,iBAAiB,CAAC;AAEzB,MAAM,GAAG,GAAG,YAAY,CAAC,gBAAgB,CAAC,CAAC;AAE3C,MAAM,QAAQ,GAAG,qDAAqD,CAAC;AACvE,MAAM,oBAAoB,GAAG,IAAI,CAAC,CAAC,iCAAiC;AACpE,MAAM,YAAY,GAAG,GAAG,CAAC,CAAC,8BAA8B;AAExD,MAAM,OAAO,OAAQ,SAAQ,UAAU;IAC5B,EAAE,GAAG,MAAM,CAAC;IACZ,SAAS,GAAG,KAAK,CAAC;IAEnB,MAAM,GAAqB,IAAI,CAAC;IAChC,MAAM,GAAa,EAAE,CAAC;IACtB,UAAU,GAAG,CAAC,CAAC;IACf,YAAY,GAAkB,IAAI,CAAC;IACnC,UAAU,GAAG,KAAK,CAAC;IACnB,QAAQ,GAAG,KAAK,CAAC;IACjB,MAAM,GAAG,KAAK,CAAC;IACf,QAAQ,GAAG,KAAK,CAAC;IAEzB,qEAAqE;IAErE,KAAK,CAAC,OAAO,CAAC,MAAiB;QAC7B,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,UAAU,GAAG,MAAM,CAAC,UAAU,IAAI,KAAK,CAAC;QAC7C,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC;QACpB,IAAI,CAAC,MAAM,GAAG,EAAE,CAAC;QACjB,IAAI,CAAC,UAAU,GAAG,CAAC,CAAC;QACpB,IAAI,CAAC,QAAQ,GAAG,KAAK,CAAC;QACtB,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC;QACzB,GAAG,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;IACjC,CAAC;IAED,KAAK,CAAC,UAAU;QACd,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;QACnB,uCAAuC;QACvC,IAAI,IAAI,CAAC,UAAU,GAAG,CAAC,EAAE,CAAC;YACxB,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;QACrB,CAAC;QACD,IAAI,CAAC,MAAM,GAAG,EAAE,CAAC;QACjB,IAAI,CAAC,UAAU,GAAG,CAAC,CAAC;QACpB,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACrB,CAAC;IAED,qEAAqE;IAErE,SAAS,CAAC,KAAa;QACrB,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO;QAExB,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACxB,IAAI,CAAC,UAAU,IAAI,KAAK,CAAC,MAAM,CAAC;QAEhC,wCAAwC;QACxC,MAAM,GAAG,GAAG,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;QACnC,IAAI,GAAG,GAAG,YAAY,EAAE,CAAC;YACvB,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,CAAC;gBACvB,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;YACjC,CAAC;iBAAM,IAAI,IAAI,CAAC,GAAG,EAAE,GAAG,IAAI,CAAC,YAAY,IAAI,oBAAoB,IAAI,IAAI,CAAC,UAAU,GAAG,CAAC,EAAE,CAAC;gBACzF,0CAA0C;gBAC1C,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC;gBACzB,IAAI,CAAC,KAAK,EAAE,CAAC,KAAK,CAAC,CAAC,GAAY,EAAE,EAAE;oBAClC,GAAG,CAAC,KAAK,CAAC,EAAE,GAAG,EAAE,EAAE,mBAAmB,CAAC,CAAC;gBAC1C,CAAC,CAAC,CAAC;YACL,CAAC;QACH,CAAC;aAAM,CAAC;YACN,iBAAiB;YACjB,IAAI,IAAI,CAAC,YAAY,EAAE,CAAC;gBACtB,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;YAC9B,CAAC;YACD,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC;QAC3B,CAAC;IACH,CAAC;IAED,oEAAoE;IAEpE,KAAK,CAAC,KAAK;QACT,IAAI,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,UAAU,KAAK,CAAC;YAAE,OAAO;QACnD,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC;QAErB,MAAM,WAAW,GAAG,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAC/C,IAAI,CAAC,MAAM,GAAG,EAAE,CAAC;QACjB,IAAI,CAAC,UAAU,GAAG,CAAC,CAAC;QAEpB,IAAI,CAAC;YACH,MAAM,GAAG,GAAG,IAAI,CAAC,QAAQ,CAAC,WAAW,EAAE,IAAI,CAAC,UAAU,CAAC,CAAC;YACxD,MAAM,UAAU,GAAG,MAAM,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC;YAE9C,IAAI,UAAU,EAAE,CAAC;gBACf,MAAM,KAAK,GAAoB;oBAC7B,IAAI,EAAE,UAAU;oBAChB,OAAO,EAAE,IAAI;oBACb,UAAU,EAAE,GAAG,EAAE,iCAAiC;oBAClD,WAAW,EAAE,IAAI;iBAClB,CAAC;gBACF,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,KAAK,CAAC,CAAC;gBAC/B,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;gBAC1B,IAAI,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;YAC7B,CAAC;QACH,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,MAAM,KAAK,GAAG,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC;YAClE,GAAG,CAAC,KAAK,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,sBAAsB,CAAC,CAAC;YAClD,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;QAC5B,CAAC;gBAAS,CAAC;YACT,IAAI,CAAC,QAAQ,GAAG,KAAK,CAAC;QACxB,CAAC;IACH,CAAC;IAED,qEAAqE;IAE7D,KAAK,CAAC,UAAU,CAAC,GAAW;QAClC,MAAM,MAAM,GAAG,IAAI,CAAC,MAAO,CAAC,MAAM,CAAC;QAEnC,8DAA8D;QAC9D,MAAM,QAAQ,GAAG,IAAI,QAAQ,EAAE,CAAC;QAChC,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,IAAI,CAAC,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,EAAE,WAAW,CAAC,CAAC;QAC7E,QAAQ,CAAC,MAAM,CAAC,OAAO,EAAE,wBAAwB,CAAC,CAAC;QACnD,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,EAAE,QAAQ,KAAK,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,EAAE,QAAQ,CAAC;QAC9E,IAAI,IAAI;YAAE,QAAQ,CAAC,MAAM,CAAC,UAAU,EAAE,IAAI,CAAC,CAAC;QAE5C,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,QAAQ,EAAE;YAChC,MAAM,EAAE,MAAM;YACd,OAAO,EAAE,EAAE,aAAa,EAAE,UAAU,MAAM,EAAE,EAAE;YAC9C,IAAI,EAAE,QAAQ;YACd,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,MAAM,CAAC;SACpC,CAAC,CAAC;QAEH,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC;YACZ,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,IAAI,EAAE,CAAC;YAC9B,MAAM,IAAI,KAAK,CAAC,YAAY,GAAG,CAAC,MAAM,KAAK,IAAI,EAAE,CAAC,CAAC;QACrD,CAAC;QAED,MAAM,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAAsB,CAAC;QACrD,OAAO,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;IAClC,CAAC;IAED,oEAAoE;IAE5D,QAAQ,CAAC,GAAW,EAAE,UAAkB;QAC9C,MAAM,WAAW,GAAG,CAAC,CAAC;QACtB,MAAM,aAAa,GAAG,EAAE,CAAC;QACzB,MAAM,QAAQ,GAAG,UAAU,GAAG,WAAW,GAAG,CAAC,aAAa,GAAG,CAAC,CAAC,CAAC;QAChE,MAAM,UAAU,GAAG,WAAW,GAAG,CAAC,aAAa,GAAG,CAAC,CAAC,CAAC;QACrD,MAAM,QAAQ,GAAG,GAAG,CAAC,MAAM,CAAC;QAC5B,MAAM,UAAU,GAAG,EAAE,CAAC;QAEtB,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;QACxC,cAAc;QACd,MAAM,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QACxB,MAAM,CAAC,aAAa,CAAC,QAAQ,GAAG,UAAU,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC;QACnD,MAAM,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QACxB,gBAAgB;QAChB,MAAM,CAAC,KAAK,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;QACzB,MAAM,CAAC,aAAa,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,iBAAiB;QAC/C,MAAM,CAAC,aAAa,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,aAAa;QAC1C,MAAM,CAAC,aAAa,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC;QACtC,MAAM,CAAC,aAAa,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;QACrC,MAAM,CAAC,aAAa,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QACnC,MAAM,CAAC,aAAa,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;QACrC,MAAM,CAAC,aAAa,CAAC,aAAa,EAAE,EAAE,CAAC,CAAC;QACxC,iBAAiB;QACjB,MAAM,CAAC,KAAK,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;QACzB,MAAM,CAAC,aAAa,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QAEnC,OAAO,MAAM,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,CAAC;IACtC,CAAC;IAED,qEAAqE;IAE7D,UAAU,CAAC,KAAa;QAC9B,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,CAAC,CAAC;QAC/B,IAAI,UAAU,GAAG,CAAC,CAAC;QACnB,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACjD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,WAAW,EAAE,CAAC,EAAE,EAAE,CAAC;YACrC,MAAM,MAAM,GAAG,KAAK,CAAC,WAAW,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YACxC,UAAU,IAAI,MAAM,GAAG,MAAM,CAAC;QAChC,CAAC;QACD,OAAO,IAAI,CAAC,IAAI,CAAC,UAAU,GAAG,WAAW,CAAC,CAAC;IAC7C,CAAC;CACF"}

package/dist/voice/stt/STTAdapter.d.ts

@@ -0,0 +1,47 @@
+/**
+ * STTAdapter — provider-agnostic speech-to-text interface.
+ *
+ * All STT providers (streaming or batch) extend this class and emit
+ * a uniform set of events so the voice pipeline can swap providers
+ * without changing any upstream logic.
+ */
+import { EventEmitter } from "node:events";
+export interface STTConfig {
+    apiKey: string;
+    language?: string;
+    sampleRate?: number;
+    encoding?: string;
+    /** Silence duration (ms) to detect end of speech. */
+    endpointingMs?: number;
+    interimResults?: boolean;
+}
+export interface TranscriptEvent {
+    text: string;
+    isFinal: boolean;
+    confidence: number;
+    /** true = user finished speaking (endpointing triggered). */
+    speechFinal: boolean;
+}
+export interface STTEvents {
+    transcript: [event: TranscriptEvent];
+    speech_started: [];
+    speech_ended: [];
+    utterance_end: [];
+    error: [error: Error];
+    close: [code?: number, reason?: string];
+}
+export declare abstract class STTAdapter extends EventEmitter {
+    abstract readonly id: string;
+    abstract readonly streaming: boolean;
+    constructor();
+    /** Connect to the provider and prepare for audio. */
+    abstract connect(config: STTConfig): Promise<void>;
+    /** Gracefully disconnect and release resources. */
+    abstract disconnect(): Promise<void>;
+    /** Feed raw PCM16 LE audio into the recogniser. */
+    abstract sendAudio(pcm16: Buffer): void;
+    emit<K extends keyof STTEvents>(event: K, ...args: STTEvents[K]): boolean;
+    on<K extends keyof STTEvents>(event: K, listener: (...args: STTEvents[K]) => void): this;
+    once<K extends keyof STTEvents>(event: K, listener: (...args: STTEvents[K]) => void): this;
+}
+//# sourceMappingURL=STTAdapter.d.ts.map

package/dist/voice/stt/STTAdapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"STTAdapter.d.ts","sourceRoot":"","sources":["../../../src/voice/stt/STTAdapter.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAI3C,MAAM,WAAW,SAAS;IACxB,MAAM,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,qDAAqD;IACrD,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,cAAc,CAAC,EAAE,OAAO,CAAC;CAC1B;AAID,MAAM,WAAW,eAAe;IAC9B,IAAI,EAAE,MAAM,CAAC;IACb,OAAO,EAAE,OAAO,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC;IACnB,6DAA6D;IAC7D,WAAW,EAAE,OAAO,CAAC;CACtB;AAID,MAAM,WAAW,SAAS;IACxB,UAAU,EAAE,CAAC,KAAK,EAAE,eAAe,CAAC,CAAC;IACrC,cAAc,EAAE,EAAE,CAAC;IACnB,YAAY,EAAE,EAAE,CAAC;IACjB,aAAa,EAAE,EAAE,CAAC;IAClB,KAAK,EAAE,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC;IACtB,KAAK,EAAE,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,MAAM,CAAC,CAAC;CACzC;AAID,8BAAsB,UAAW,SAAQ,YAAY;IACnD,QAAQ,CAAC,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,QAAQ,CAAC,SAAS,EAAE,OAAO,CAAC;;IAQrC,qDAAqD;IACrD,QAAQ,CAAC,OAAO,CAAC,MAAM,EAAE,SAAS,GAAG,OAAO,CAAC,IAAI,CAAC;IAElD,mDAAmD;IACnD,QAAQ,CAAC,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;IAEpC,mDAAmD;IACnD,QAAQ,CAAC,SAAS,CAAC,KAAK,EAAE,MAAM,GAAG,IAAI;IAI9B,IAAI,CAAC,CAAC,SAAS,MAAM,SAAS,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,IAAI,EAAE,SAAS,CAAC,CAAC,CAAC,GAAG,OAAO;IAIzE,EAAE,CAAC,CAAC,SAAS,MAAM,SAAS,EAAE,KAAK,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,GAAG,IAAI,EAAE,SAAS,CAAC,CAAC,CAAC,KAAK,IAAI,GAAG,IAAI;IAIxF,IAAI,CAAC,CAAC,SAAS,MAAM,SAAS,EAAE,KAAK,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,GAAG,IAAI,EAAE,SAAS,CAAC,CAAC,CAAC,KAAK,IAAI,GAAG,IAAI;CAGpG"}

package/dist/voice/stt/STTAdapter.js

@@ -0,0 +1,27 @@
+/**
+ * STTAdapter — provider-agnostic speech-to-text interface.
+ *
+ * All STT providers (streaming or batch) extend this class and emit
+ * a uniform set of events so the voice pipeline can swap providers
+ * without changing any upstream logic.
+ */
+import { EventEmitter } from "node:events";
+// ── Abstract base ───────────────────────────────────────────────────
+export class STTAdapter extends EventEmitter {
+    constructor() {
+        super();
+        // Prevent unhandled error events from crashing the process
+        this.on("error", () => { });
+    }
+    // ── Typed emit / on helpers ─────────────────────────────────────
+    emit(event, ...args) {
+        return super.emit(event, ...args);
+    }
+    on(event, listener) {
+        return super.on(event, listener);
+    }
+    once(event, listener) {
+        return super.once(event, listener);
+    }
+}
+//# sourceMappingURL=STTAdapter.js.map

package/dist/voice/stt/STTAdapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"STTAdapter.js","sourceRoot":"","sources":["../../../src/voice/stt/STTAdapter.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAmC3C,uEAAuE;AAEvE,MAAM,OAAgB,UAAW,SAAQ,YAAY;IAInD;QACE,KAAK,EAAE,CAAC;QACR,2DAA2D;QAC3D,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,GAAG,EAAE,GAAE,CAAC,CAAC,CAAC;IAC7B,CAAC;IAWD,mEAAmE;IAE1D,IAAI,CAA4B,KAAQ,EAAE,GAAG,IAAkB;QACtE,OAAO,KAAK,CAAC,IAAI,CAAC,KAAe,EAAE,GAAG,IAAI,CAAC,CAAC;IAC9C,CAAC;IAEQ,EAAE,CAA4B,KAAQ,EAAE,QAAyC;QACxF,OAAO,KAAK,CAAC,EAAE,CAAC,KAAe,EAAE,QAAwC,CAAC,CAAC;IAC7E,CAAC;IAEQ,IAAI,CAA4B,KAAQ,EAAE,QAAyC;QAC1F,OAAO,KAAK,CAAC,IAAI,CAAC,KAAe,EAAE,QAAwC,CAAC,CAAC;IAC/E,CAAC;CACF"}

package/dist/voice/transcribe.d.ts

@@ -0,0 +1,33 @@
+/**
+ * transcribeAudioFile — shared Whisper transcription helper.
+ *
+ * Used by:
+ *   - The `transcribe_audio` tool (agent-exposed).
+ *   - Channel integrations that need to turn inbound voice messages
+ *     into text before handing them to the agent loop (Telegram voice,
+ *     WhatsApp voice notes, iMessage audio, etc.).
+ *
+ * Prefers Groq Whisper (free tier, fast) and falls back to OpenAI.
+ * Throws ProviderUnavailableError when neither key is set so callers
+ * can decide whether to retry, notify the user, or skip the message.
+ */
+import type { ConfigManager } from "../config/ConfigManager.js";
+export interface TranscribeOptions {
+    /** ISO 639-1 hint (e.g. "en"). Optional — Whisper auto-detects. */
+    readonly language?: string;
+    /** Short context phrase that biases recognition (names, jargon). */
+    readonly prompt?: string;
+    /** Override the per-provider default model. */
+    readonly model?: string;
+    /** Hard timeout. Defaults to 60 s — enough for ~10 min of audio. */
+    readonly timeoutMs?: number;
+    /** Abort from upstream (e.g. channel stop). */
+    readonly signal?: AbortSignal;
+}
+export interface TranscribeResult {
+    readonly text: string;
+    readonly provider: "groq" | "openai";
+    readonly model: string;
+}
+export declare function transcribeAudioFile(cfg: ConfigManager, filePath: string, opts?: TranscribeOptions): Promise<TranscribeResult>;
+//# sourceMappingURL=transcribe.d.ts.map

package/dist/voice/transcribe.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"transcribe.d.ts","sourceRoot":"","sources":["../../src/voice/transcribe.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAKH,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,4BAA4B,CAAC;AAGhE,MAAM,WAAW,iBAAiB;IAChC,mEAAmE;IACnE,QAAQ,CAAC,QAAQ,CAAC,EAAE,MAAM,CAAC;IAC3B,oEAAoE;IACpE,QAAQ,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IACzB,+CAA+C;IAC/C,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IACxB,oEAAoE;IACpE,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAC5B,+CAA+C;IAC/C,QAAQ,CAAC,MAAM,CAAC,EAAE,WAAW,CAAC;CAC/B;AAED,MAAM,WAAW,gBAAgB;IAC/B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,QAAQ,EAAE,MAAM,GAAG,QAAQ,CAAC;IACrC,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;CACxB;AAED,wBAAsB,mBAAmB,CACvC,GAAG,EAAE,aAAa,EAClB,QAAQ,EAAE,MAAM,EAChB,IAAI,GAAE,iBAAsB,GAC3B,OAAO,CAAC,gBAAgB,CAAC,CA6C3B"}

package/dist/voice/transcribe.js

@@ -0,0 +1,73 @@
+/**
+ * transcribeAudioFile — shared Whisper transcription helper.
+ *
+ * Used by:
+ *   - The `transcribe_audio` tool (agent-exposed).
+ *   - Channel integrations that need to turn inbound voice messages
+ *     into text before handing them to the agent loop (Telegram voice,
+ *     WhatsApp voice notes, iMessage audio, etc.).
+ *
+ * Prefers Groq Whisper (free tier, fast) and falls back to OpenAI.
+ * Throws ProviderUnavailableError when neither key is set so callers
+ * can decide whether to retry, notify the user, or skip the message.
+ */
+import { createReadStream } from "node:fs";
+import { basename } from "node:path";
+import { ProviderError, ProviderUnavailableError } from "../util/errors.js";
+export async function transcribeAudioFile(cfg, filePath, opts = {}) {
+    const groqKey = cfg.vault.get("GROQ_API_KEY")?.reveal();
+    const openaiKey = cfg.vault.get("OPENAI_API_KEY")?.reveal();
+    if (!groqKey && !openaiKey) {
+        throw new ProviderUnavailableError("Whisper STT", "GROQ_API_KEY or OPENAI_API_KEY");
+    }
+    const isGroq = !!groqKey;
+    const apiKey = isGroq ? groqKey : openaiKey;
+    const baseUrl = isGroq ? "https://api.groq.com/openai/v1" : "https://api.openai.com/v1";
+    const model = opts.model ?? (isGroq ? "whisper-large-v3-turbo" : "whisper-1");
+    const form = new FormData();
+    const stream = createReadStream(filePath);
+    const chunks = [];
+    for await (const chunk of stream)
+        chunks.push(chunk);
+    form.append("file", new Blob([Buffer.concat(chunks)]), basename(filePath));
+    form.append("model", model);
+    if (opts.prompt)
+        form.append("prompt", opts.prompt);
+    if (opts.language)
+        form.append("language", opts.language);
+    const timeoutSignal = AbortSignal.timeout(opts.timeoutMs ?? 60_000);
+    const signal = opts.signal
+        ? composeSignals([opts.signal, timeoutSignal])
+        : timeoutSignal;
+    const res = await fetch(`${baseUrl}/audio/transcriptions`, {
+        method: "POST",
+        headers: { Authorization: `Bearer ${apiKey}` },
+        body: form,
+        signal,
+    });
+    if (!res.ok) {
+        const body = await res.text().catch(() => "");
+        throw new ProviderError(`Whisper ${res.status}: ${body.slice(0, 400)}`, isGroq ? "groq" : "openai");
+    }
+    const data = (await res.json());
+    return {
+        text: (data.text ?? "").trim(),
+        provider: isGroq ? "groq" : "openai",
+        model,
+    };
+}
+function composeSignals(signals) {
+    const native = AbortSignal.any;
+    if (typeof native === "function")
+        return native(Array.from(signals));
+    const ctrl = new AbortController();
+    for (const s of signals) {
+        if (s.aborted) {
+            ctrl.abort(s.reason);
+            break;
+        }
+        s.addEventListener("abort", () => ctrl.abort(s.reason), { once: true });
+    }
+    return ctrl.signal;
+}
+//# sourceMappingURL=transcribe.js.map

package/dist/voice/transcribe.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"transcribe.js","sourceRoot":"","sources":["../../src/voice/transcribe.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,SAAS,CAAC;AAC3C,OAAO,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAC;AAGrC,OAAO,EAAE,aAAa,EAAE,wBAAwB,EAAE,MAAM,mBAAmB,CAAC;AAqB5E,MAAM,CAAC,KAAK,UAAU,mBAAmB,CACvC,GAAkB,EAClB,QAAgB,EAChB,OAA0B,EAAE;IAE5B,MAAM,OAAO,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,cAAc,CAAC,EAAE,MAAM,EAAE,CAAC;IACxD,MAAM,SAAS,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,gBAAgB,CAAC,EAAE,MAAM,EAAE,CAAC;IAC5D,IAAI,CAAC,OAAO,IAAI,CAAC,SAAS,EAAE,CAAC;QAC3B,MAAM,IAAI,wBAAwB,CAAC,aAAa,EAAE,gCAAgC,CAAC,CAAC;IACtF,CAAC;IAED,MAAM,MAAM,GAAG,CAAC,CAAC,OAAO,CAAC;IACzB,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,OAAQ,CAAC,CAAC,CAAC,SAAU,CAAC;IAC9C,MAAM,OAAO,GAAG,MAAM,CAAC,CAAC,CAAC,gCAAgC,CAAC,CAAC,CAAC,2BAA2B,CAAC;IACxF,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,wBAAwB,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC;IAE9E,MAAM,IAAI,GAAG,IAAI,QAAQ,EAAE,CAAC;IAC5B,MAAM,MAAM,GAAG,gBAAgB,CAAC,QAAQ,CAAC,CAAC;IAC1C,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,MAAM;QAAE,MAAM,CAAC,IAAI,CAAC,KAAe,CAAC,CAAC;IAC/D,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,IAAI,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC;IAC3E,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;IAC5B,IAAI,IAAI,CAAC,MAAM;QAAE,IAAI,CAAC,MAAM,CAAC,QAAQ,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;IACpD,IAAI,IAAI,CAAC,QAAQ;QAAE,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;IAE1D,MAAM,aAAa,GAAG,WAAW,CAAC,OAAO,CAAC,IAAI,CAAC,SAAS,IAAI,MAAM,CAAC,CAAC;IACpE,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM;QACxB,CAAC,CAAC,cAAc,CAAC,CAAC,IAAI,CAAC,MAAM,EAAE,aAAa,CAAC,CAAC;QAC9C,CAAC,CAAC,aAAa,CAAC;IAElB,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,GAAG,OAAO,uBAAuB,EAAE;QACzD,MAAM,EAAE,MAAM;QACd,OAAO,EAAE,EAAE,aAAa,EAAE,UAAU,MAAM,EAAE,EAAE;QAC9C,IAAI,EAAE,IAAI;QACV,MAAM;KACP,CAAC,CAAC;IACH,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC;QACZ,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC;QAC9C,MAAM,IAAI,aAAa,CACrB,WAAW,GAAG,CAAC,MAAM,KAAK,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,EAC9C,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,CAC3B,CAAC;IACJ,CAAC;IACD,MAAM,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAAsB,CAAC;IACrD,OAAO;QACL,IAAI,EAAE,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE;QAC9B,QAAQ,EAAE,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ;QACpC,KAAK;KACN,CAAC;AACJ,CAAC;AAED,SAAS,cAAc,CAAC,OAA+B;IACrD,MAAM,MAAM,GAAI,WAAsE,CAAC,GAAG,CAAC;IAC3F,IAAI,OAAO,MAAM,KAAK,UAAU;QAAE,OAAO,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;IACrE,MAAM,IAAI,GAAG,IAAI,eAAe,EAAE,CAAC;IACnC,KAAK,MAAM,CAAC,IAAI,OAAO,EAAE,CAAC;QACxB,IAAI,CAAC,CAAC,OAAO,EAAE,CAAC;YAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC;YAAC,MAAM;QAAC,CAAC;QAC/C,CAAC,CAAC,gBAAgB,CAAC,OAAO,EAAE,GAAG,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC;IAC1E,CAAC;IACD,OAAO,IAAI,CAAC,MAAM,CAAC;AACrB,CAAC"}

package/dist/voice/transcriptFilter.d.ts

@@ -0,0 +1,16 @@
+/**
+ * Whisper / Groq STT hallucination filter.
+ *
+ * These phrases are training artifacts — Whisper was trained on YouTube
+ * audio where creators say "Thanks for watching, subscribe, thank you"
+ * at video ends. When the model is given silence or low-energy noise,
+ * the decoder falls into that training bias and produces the same
+ * phrases. Dropping them before they reach the agent is mandatory;
+ * otherwise the shared session fills up with phantom turns.
+ *
+ * This is a belt-and-braces layer — the primary defence is keeping
+ * silent PCM out of the STT (`voiceEnergyGate.ts`). Anything that
+ * still sneaks through lands here.
+ */
+export declare function isHallucinatedTranscript(raw: string): boolean;
+//# sourceMappingURL=transcriptFilter.d.ts.map

package/dist/voice/transcriptFilter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"transcriptFilter.d.ts","sourceRoot":"","sources":["../../src/voice/transcriptFilter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAgBH,wBAAgB,wBAAwB,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAK7D"}

package/dist/voice/transcriptFilter.js

@@ -0,0 +1,36 @@
+/**
+ * Whisper / Groq STT hallucination filter.
+ *
+ * These phrases are training artifacts — Whisper was trained on YouTube
+ * audio where creators say "Thanks for watching, subscribe, thank you"
+ * at video ends. When the model is given silence or low-energy noise,
+ * the decoder falls into that training bias and produces the same
+ * phrases. Dropping them before they reach the agent is mandatory;
+ * otherwise the shared session fills up with phantom turns.
+ *
+ * This is a belt-and-braces layer — the primary defence is keeping
+ * silent PCM out of the STT (`voiceEnergyGate.ts`). Anything that
+ * still sneaks through lands here.
+ */
+const PATTERNS = [
+    /^thank\s*you\.?$/i,
+    /^thanks?(\s+for\s+(watching|listening))?\.?$/i,
+    /^you\.?$/i,
+    /^bye\.?$/i,
+    /^okay\.?$/i,
+    /^ok\.?$/i,
+    /^please\s+subscribe\.?$/i,
+    /^\.+$/,
+    /^\s*$/,
+    /^(thank\s*you[.,!\s]*){2,}/i,
+    /^(thanks[.,!\s]*){2,}/i,
+];
+export function isHallucinatedTranscript(raw) {
+    const trimmed = raw.trim().replace(/^[.,!\s]+/, "").trim();
+    if (!trimmed)
+        return true;
+    if (trimmed.length < 3)
+        return true;
+    return PATTERNS.some((re) => re.test(trimmed));
+}
+//# sourceMappingURL=transcriptFilter.js.map

package/dist/voice/transcriptFilter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"transcriptFilter.js","sourceRoot":"","sources":["../../src/voice/transcriptFilter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,MAAM,QAAQ,GAAsB;IAClC,mBAAmB;IACnB,+CAA+C;IAC/C,WAAW;IACX,WAAW;IACX,YAAY;IACZ,UAAU;IACV,0BAA0B;IAC1B,OAAO;IACP,OAAO;IACP,6BAA6B;IAC7B,wBAAwB;CACzB,CAAC;AAEF,MAAM,UAAU,wBAAwB,CAAC,GAAW;IAClD,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;IAC3D,IAAI,CAAC,OAAO;QAAE,OAAO,IAAI,CAAC;IAC1B,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC;QAAE,OAAO,IAAI,CAAC;IACpC,OAAO,QAAQ,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;AACjD,CAAC"}

package/dist/voice/tts/CartesiaTTS.d.ts

@@ -0,0 +1,31 @@
+/**
+ * CartesiaTTS — WebSocket streaming text-to-speech via Cartesia API.
+ *
+ * Lowest-latency option: sends text, receives raw PCM16 chunks
+ * with no codec decode step. Supports mid-stream cancellation
+ * via context_id.
+ */
+import { TTSAdapter, type TTSConfig } from "./TTSAdapter.js";
+export declare class CartesiaTTS extends TTSAdapter {
+    readonly id = "cartesia";
+    readonly streaming = true;
+    private ws;
+    private config;
+    private voiceId;
+    private modelId;
+    private contextId;
+    private retryCount;
+    private closing;
+    private connected;
+    private textBuffer;
+    private audioFormat;
+    connect(config: TTSConfig): Promise<void>;
+    private open;
+    private handleMessage;
+    sendText(text: string): void;
+    flush(): void;
+    cancel(): void;
+    private maybeReconnect;
+    disconnect(): Promise<void>;
+}
+//# sourceMappingURL=CartesiaTTS.d.ts.map

package/dist/voice/tts/CartesiaTTS.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CartesiaTTS.d.ts","sourceRoot":"","sources":["../../../src/voice/tts/CartesiaTTS.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAKH,OAAO,EACL,UAAU,EAEV,KAAK,SAAS,EACf,MAAM,iBAAiB,CAAC;AAUzB,qBAAa,WAAY,SAAQ,UAAU;IACzC,QAAQ,CAAC,EAAE,cAAc;IACzB,QAAQ,CAAC,SAAS,QAAQ;IAE1B,OAAO,CAAC,EAAE,CAA0B;IACpC,OAAO,CAAC,MAAM,CAA0B;IACxC,OAAO,CAAC,OAAO,CAAiB;IAChC,OAAO,CAAC,OAAO,CAAiB;IAChC,OAAO,CAAC,SAAS,CAAgB;IACjC,OAAO,CAAC,UAAU,CAAK;IACvB,OAAO,CAAC,OAAO,CAAS;IACxB,OAAO,CAAC,SAAS,CAAS;IAC1B,OAAO,CAAC,UAAU,CAAM;IACxB,OAAO,CAAC,WAAW,CAIjB;IAII,OAAO,CAAC,MAAM,EAAE,SAAS,GAAG,OAAO,CAAC,IAAI,CAAC;IAc/C,OAAO,CAAC,IAAI;IA6CZ,OAAO,CAAC,aAAa;IAgCrB,QAAQ,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI;IAM5B,KAAK,IAAI,IAAI;IAkCb,MAAM,IAAI,IAAI;IAgBd,OAAO,CAAC,cAAc;IAqBhB,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;CAuBlC"}