@seawork/server 1.0.0

package/dist/server/server/speech/providers/openai/stt.js

@@ -0,0 +1,206 @@
+import { EventEmitter } from "node:events";
+import OpenAI from "openai";
+import { writeFile, unlink } from "fs/promises";
+import { join } from "path";
+import { tmpdir } from "os";
+import { v4 } from "uuid";
+import { inferAudioExtension } from "../../../agent/audio-utils.js";
+function isObject(value) {
+    return typeof value === "object" && value !== null;
+}
+function isLogprobToken(value) {
+    if (!isObject(value)) {
+        return false;
+    }
+    if (typeof value.token !== "string") {
+        return false;
+    }
+    if (typeof value.logprob !== "number") {
+        return false;
+    }
+    if (value.bytes === undefined) {
+        return true;
+    }
+    return Array.isArray(value.bytes) && value.bytes.every((entry) => typeof entry === "number");
+}
+function isLogprobTokenArray(value) {
+    return Array.isArray(value) && value.every((entry) => isLogprobToken(entry));
+}
+export class OpenAISTT {
+    constructor(sttConfig, parentLogger) {
+        this.id = "openai";
+        this.config = sttConfig;
+        this.logger = parentLogger.child({ module: "agent", provider: "openai", component: "stt" });
+        this.openaiClient = new OpenAI({
+            apiKey: sttConfig.apiKey,
+        });
+        this.logger.info({ model: sttConfig.model || "whisper-1" }, "STT (OpenAI Whisper) initialized");
+    }
+    createSession(params) {
+        const emitter = new EventEmitter();
+        const logger = params.logger.child({ provider: "openai", component: "stt-session" });
+        const requiredSampleRate = 24000;
+        let connected = false;
+        let segmentId = v4();
+        let previousSegmentId = null;
+        let pcm16 = Buffer.alloc(0);
+        const transcribeAudio = this.transcribeAudioInternal.bind(this);
+        const convertPCMToWavBuffer = (pcmBuffer) => {
+            const headerSize = 44;
+            const channels = 1;
+            const bitsPerSample = 16;
+            const sampleRate = requiredSampleRate;
+            const wavBuffer = Buffer.alloc(headerSize + pcmBuffer.length);
+            const byteRate = (sampleRate * channels * bitsPerSample) / 8;
+            const blockAlign = (channels * bitsPerSample) / 8;
+            wavBuffer.write("RIFF", 0);
+            wavBuffer.writeUInt32LE(36 + pcmBuffer.length, 4);
+            wavBuffer.write("WAVE", 8);
+            wavBuffer.write("fmt ", 12);
+            wavBuffer.writeUInt32LE(16, 16);
+            wavBuffer.writeUInt16LE(1, 20);
+            wavBuffer.writeUInt16LE(channels, 22);
+            wavBuffer.writeUInt32LE(sampleRate, 24);
+            wavBuffer.writeUInt32LE(byteRate, 28);
+            wavBuffer.writeUInt16LE(blockAlign, 32);
+            wavBuffer.writeUInt16LE(bitsPerSample, 34);
+            wavBuffer.write("data", 36);
+            wavBuffer.writeUInt32LE(pcmBuffer.length, 40);
+            pcmBuffer.copy(wavBuffer, 44);
+            return wavBuffer;
+        };
+        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 {
+                        if (pcm16.length === 0) {
+                            emitter.emit("transcript", {
+                                segmentId: committedId,
+                                transcript: "",
+                                isFinal: true,
+                                language: params.language,
+                                isLowConfidence: true,
+                            });
+                            return;
+                        }
+                        const wav = convertPCMToWavBuffer(pcm16);
+                        const result = await transcribeAudio(wav, "audio/wav", params.language ?? "en", logger);
+                        emitter.emit("transcript", {
+                            segmentId: committedId,
+                            transcript: result.text,
+                            isFinal: true,
+                            language: result.language,
+                            logprobs: result.logprobs,
+                            avgLogprob: result.avgLogprob,
+                            isLowConfidence: result.isLowConfidence,
+                        });
+                    }
+                    catch (err) {
+                        emitter.emit("error", err);
+                    }
+                    finally {
+                        previousSegmentId = committedId;
+                        segmentId = v4();
+                        pcm16 = Buffer.alloc(0);
+                    }
+                })();
+            },
+            clear() {
+                pcm16 = Buffer.alloc(0);
+                segmentId = v4();
+            },
+            close() {
+                connected = false;
+                pcm16 = Buffer.alloc(0);
+            },
+            on(event, handler) {
+                emitter.on(event, handler);
+                return undefined;
+            },
+        };
+    }
+    async transcribeAudioInternal(audioBuffer, format, language, logger) {
+        const startTime = Date.now();
+        let tempFilePath = null;
+        try {
+            const ext = inferAudioExtension(format);
+            tempFilePath = join(tmpdir(), `audio-${v4()}.${ext}`);
+            await writeFile(tempFilePath, audioBuffer);
+            logger.debug({ tempFilePath, bytes: audioBuffer.length }, "Transcribing audio file");
+            const modelToUse = this.config.model ?? "whisper-1";
+            const supportsLogprobs = modelToUse === "gpt-4o-transcribe" || modelToUse === "gpt-4o-mini-transcribe";
+            const includeLogprobs = ["logprobs"];
+            const response = await this.openaiClient.audio.transcriptions.create({
+                file: await import("fs").then((fs) => fs.createReadStream(tempFilePath)),
+                language,
+                model: modelToUse,
+                ...(supportsLogprobs ? { include: includeLogprobs } : {}),
+                response_format: "json",
+            });
+            const duration = Date.now() - startTime;
+            const confidenceThreshold = this.config.confidenceThreshold ?? -3.0;
+            let avgLogprob;
+            let isLowConfidence = false;
+            const logprobs = supportsLogprobs && isObject(response) && isLogprobTokenArray(response.logprobs)
+                ? response.logprobs
+                : undefined;
+            if (logprobs && logprobs.length > 0) {
+                const totalLogprob = logprobs.reduce((sum, token) => sum + token.logprob, 0);
+                avgLogprob = totalLogprob / logprobs.length;
+                isLowConfidence = avgLogprob < confidenceThreshold;
+                if (isLowConfidence) {
+                    logger.debug({
+                        avgLogprob,
+                        threshold: confidenceThreshold,
+                        text: response.text,
+                        tokenLogprobs: logprobs.map((t) => `${t.token}:${t.logprob.toFixed(2)}`).join(", "),
+                    }, "Low confidence transcription detected");
+                }
+            }
+            logger.debug({ duration, text: response.text, avgLogprob }, "Transcription complete");
+            return {
+                text: response.text,
+                duration: duration,
+                logprobs: logprobs,
+                avgLogprob: avgLogprob,
+                isLowConfidence: isLowConfidence,
+                language: isObject(response) && typeof response.language === "string"
+                    ? response.language
+                    : undefined,
+            };
+        }
+        catch (error) {
+            logger.error({ err: error }, "Transcription error");
+            throw new Error(`STT transcription failed: ${error.message}`);
+        }
+        finally {
+            if (tempFilePath) {
+                try {
+                    await unlink(tempFilePath);
+                }
+                catch (cleanupError) {
+                    logger.warn({ tempFilePath }, "Failed to clean up temp file");
+                }
+            }
+        }
+    }
+}
+//# sourceMappingURL=stt.js.map

package/dist/server/server/speech/providers/openai/stt.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"stt.js","sourceRoot":"","sources":["../../../../../../src/server/speech/providers/openai/stt.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAE3C,OAAO,MAAM,MAAM,QAAQ,CAAC;AAC5B,OAAO,EAAE,SAAS,EAAE,MAAM,EAAE,MAAM,aAAa,CAAC;AAChD,OAAO,EAAE,IAAI,EAAE,MAAM,MAAM,CAAC;AAC5B,OAAO,EAAE,MAAM,EAAE,MAAM,IAAI,CAAC;AAC5B,OAAO,EAAE,EAAE,EAAE,MAAM,MAAM,CAAC;AAC1B,OAAO,EAAE,mBAAmB,EAAE,MAAM,+BAA+B,CAAC;AAgBpE,SAAS,QAAQ,CAAC,KAAc;IAC9B,OAAO,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,KAAK,IAAI,CAAC;AACrD,CAAC;AAED,SAAS,cAAc,CAAC,KAAc;IACpC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACrB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,OAAO,KAAK,CAAC,KAAK,KAAK,QAAQ,EAAE,CAAC;QACpC,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,OAAO,KAAK,CAAC,OAAO,KAAK,QAAQ,EAAE,CAAC;QACtC,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,KAAK,CAAC,KAAK,KAAK,SAAS,EAAE,CAAC;QAC9B,OAAO,IAAI,CAAC;IACd,CAAC;IACD,OAAO,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC;AAC/F,CAAC;AAED,SAAS,mBAAmB,CAAC,KAAc;IACzC,OAAO,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,cAAc,CAAC,KAAK,CAAC,CAAC,CAAC;AAC/E,CAAC;AAED,MAAM,OAAO,SAAS;IAMpB,YAAY,SAAoB,EAAE,YAAyB;QAF3C,OAAE,GAAG,QAAiB,CAAC;QAGrC,IAAI,CAAC,MAAM,GAAG,SAAS,CAAC;QACxB,IAAI,CAAC,MAAM,GAAG,YAAY,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC,CAAC;QAC5F,IAAI,CAAC,YAAY,GAAG,IAAI,MAAM,CAAC;YAC7B,MAAM,EAAE,SAAS,CAAC,MAAM;SACzB,CAAC,CAAC;QACH,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,SAAS,CAAC,KAAK,IAAI,WAAW,EAAE,EAAE,kCAAkC,CAAC,CAAC;IAClG,CAAC;IAEM,aAAa,CAAC,MAIpB;QACC,MAAM,OAAO,GAAG,IAAI,YAAY,EAAE,CAAC;QACnC,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,QAAQ,EAAE,QAAQ,EAAE,SAAS,EAAE,aAAa,EAAE,CAAC,CAAC;QACrF,MAAM,kBAAkB,GAAG,KAAK,CAAC;QAEjC,IAAI,SAAS,GAAG,KAAK,CAAC;QACtB,IAAI,SAAS,GAAG,EAAE,EAAE,CAAC;QACrB,IAAI,iBAAiB,GAAkB,IAAI,CAAC;QAC5C,IAAI,KAAK,GAAW,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QACpC,MAAM,eAAe,GAAG,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAEhE,MAAM,qBAAqB,GAAG,CAAC,SAAiB,EAAU,EAAE;YAC1D,MAAM,UAAU,GAAG,EAAE,CAAC;YACtB,MAAM,QAAQ,GAAG,CAAC,CAAC;YACnB,MAAM,aAAa,GAAG,EAAE,CAAC;YACzB,MAAM,UAAU,GAAG,kBAAkB,CAAC;YACtC,MAAM,SAAS,GAAG,MAAM,CAAC,KAAK,CAAC,UAAU,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;YAC9D,MAAM,QAAQ,GAAG,CAAC,UAAU,GAAG,QAAQ,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;YAC7D,MAAM,UAAU,GAAG,CAAC,QAAQ,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;YAElD,SAAS,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;YAC3B,SAAS,CAAC,aAAa,CAAC,EAAE,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;YAClD,SAAS,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;YAC3B,SAAS,CAAC,KAAK,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;YAC5B,SAAS,CAAC,aAAa,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC;YAChC,SAAS,CAAC,aAAa,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YAC/B,SAAS,CAAC,aAAa,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;YACtC,SAAS,CAAC,aAAa,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;YACxC,SAAS,CAAC,aAAa,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;YACtC,SAAS,CAAC,aAAa,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;YACxC,SAAS,CAAC,aAAa,CAAC,aAAa,EAAE,EAAE,CAAC,CAAC;YAC3C,SAAS,CAAC,KAAK,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;YAC5B,SAAS,CAAC,aAAa,CAAC,SAAS,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;YAC9C,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;YAE9B,OAAO,SAAS,CAAC;QACnB,CAAC,CAAC;QAEF,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;gBACJ,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,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;4BACtB,OAAe,CAAC,IAAI,CAAC,YAAY,EAAE;gCAClC,SAAS,EAAE,WAAW;gCACtB,UAAU,EAAE,EAAE;gCACd,OAAO,EAAE,IAAI;gCACb,QAAQ,EAAE,MAAM,CAAC,QAAQ;gCACzB,eAAe,EAAE,IAAI;6BACtB,CAAC,CAAC;4BACH,OAAO;wBACT,CAAC;wBAED,MAAM,GAAG,GAAG,qBAAqB,CAAC,KAAK,CAAC,CAAC;wBACzC,MAAM,MAAM,GAAG,MAAM,eAAe,CAAC,GAAG,EAAE,WAAW,EAAE,MAAM,CAAC,QAAQ,IAAI,IAAI,EAAE,MAAM,CAAC,CAAC;wBAEvF,OAAe,CAAC,IAAI,CAAC,YAAY,EAAE;4BAClC,SAAS,EAAE,WAAW;4BACtB,UAAU,EAAE,MAAM,CAAC,IAAI;4BACvB,OAAO,EAAE,IAAI;4BACb,QAAQ,EAAE,MAAM,CAAC,QAAQ;4BACzB,QAAQ,EAAE,MAAM,CAAC,QAAQ;4BACzB,UAAU,EAAE,MAAM,CAAC,UAAU;4BAC7B,eAAe,EAAE,MAAM,CAAC,eAAe;yBACxC,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,EAAE,EAAE,CAAC;wBACjB,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,EAAE,EAAE,CAAC;YACnB,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;IAEO,KAAK,CAAC,uBAAuB,CACnC,WAAmB,EACnB,MAAc,EACd,QAAgB,EAChB,MAAmB;QAEnB,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,YAAY,GAAkB,IAAI,CAAC;QAEvC,IAAI,CAAC;YACH,MAAM,GAAG,GAAG,mBAAmB,CAAC,MAAM,CAAC,CAAC;YACxC,YAAY,GAAG,IAAI,CAAC,MAAM,EAAE,EAAE,SAAS,EAAE,EAAE,IAAI,GAAG,EAAE,CAAC,CAAC;YACtD,MAAM,SAAS,CAAC,YAAY,EAAE,WAAW,CAAC,CAAC;YAE3C,MAAM,CAAC,KAAK,CAAC,EAAE,YAAY,EAAE,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,EAAE,yBAAyB,CAAC,CAAC;YAErF,MAAM,UAAU,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,IAAI,WAAW,CAAC;YACpD,MAAM,gBAAgB,GACpB,UAAU,KAAK,mBAAmB,IAAI,UAAU,KAAK,wBAAwB,CAAC;YAChF,MAAM,eAAe,GAAiB,CAAC,UAAU,CAAC,CAAC;YAEnD,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,cAAc,CAAC,MAAM,CAAC;gBACnE,IAAI,EAAE,MAAM,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC,gBAAgB,CAAC,YAAa,CAAC,CAAC;gBACzE,QAAQ;gBACR,KAAK,EAAE,UAAU;gBACjB,GAAG,CAAC,gBAAgB,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,eAAe,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACzD,eAAe,EAAE,MAAM;aACxB,CAAC,CAAC;YAEH,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,CAAC;YACxC,MAAM,mBAAmB,GAAG,IAAI,CAAC,MAAM,CAAC,mBAAmB,IAAI,CAAC,GAAG,CAAC;YAEpE,IAAI,UAA8B,CAAC;YACnC,IAAI,eAAe,GAAG,KAAK,CAAC;YAC5B,MAAM,QAAQ,GACZ,gBAAgB,IAAI,QAAQ,CAAC,QAAQ,CAAC,IAAI,mBAAmB,CAAC,QAAQ,CAAC,QAAQ,CAAC;gBAC9E,CAAC,CAAC,QAAQ,CAAC,QAAQ;gBACnB,CAAC,CAAC,SAAS,CAAC;YAEhB,IAAI,QAAQ,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACpC,MAAM,YAAY,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE,EAAE,CAAC,GAAG,GAAG,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;gBAC7E,UAAU,GAAG,YAAY,GAAG,QAAQ,CAAC,MAAM,CAAC;gBAC5C,eAAe,GAAG,UAAU,GAAG,mBAAmB,CAAC;gBAEnD,IAAI,eAAe,EAAE,CAAC;oBACpB,MAAM,CAAC,KAAK,CACV;wBACE,UAAU;wBACV,SAAS,EAAE,mBAAmB;wBAC9B,IAAI,EAAE,QAAQ,CAAC,IAAI;wBACnB,aAAa,EAAE,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC;qBACpF,EACD,uCAAuC,CACxC,CAAC;gBACJ,CAAC;YACH,CAAC;YAED,MAAM,CAAC,KAAK,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,EAAE,UAAU,EAAE,EAAE,wBAAwB,CAAC,CAAC;YAEtF,OAAO;gBACL,IAAI,EAAE,QAAQ,CAAC,IAAI;gBACnB,QAAQ,EAAE,QAAQ;gBAClB,QAAQ,EAAE,QAAQ;gBAClB,UAAU,EAAE,UAAU;gBACtB,eAAe,EAAE,eAAe;gBAChC,QAAQ,EACN,QAAQ,CAAC,QAAQ,CAAC,IAAI,OAAO,QAAQ,CAAC,QAAQ,KAAK,QAAQ;oBACzD,CAAC,CAAC,QAAQ,CAAC,QAAQ;oBACnB,CAAC,CAAC,SAAS;aAChB,CAAC;QACJ,CAAC;QAAC,OAAO,KAAU,EAAE,CAAC;YACpB,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,qBAAqB,CAAC,CAAC;YACpD,MAAM,IAAI,KAAK,CAAC,6BAA6B,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;QAChE,CAAC;gBAAS,CAAC;YACT,IAAI,YAAY,EAAE,CAAC;gBACjB,IAAI,CAAC;oBACH,MAAM,MAAM,CAAC,YAAY,CAAC,CAAC;gBAC7B,CAAC;gBAAC,OAAO,YAAY,EAAE,CAAC;oBACtB,MAAM,CAAC,IAAI,CAAC,EAAE,YAAY,EAAE,EAAE,8BAA8B,CAAC,CAAC;gBAChE,CAAC;YACH,CAAC;QACH,CAAC;IACH,CAAC;CACF"}

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

@@ -0,0 +1,18 @@
+import type pino from "pino";
+import type { SpeechStreamResult, TextToSpeechProvider } from "../../speech-provider.js";
+export type { SpeechStreamResult };
+export interface TTSConfig {
+    apiKey: string;
+    model?: "tts-1" | "tts-1-hd";
+    voice?: "alloy" | "echo" | "fable" | "onyx" | "nova" | "shimmer";
+    responseFormat?: "mp3" | "opus" | "aac" | "flac" | "wav" | "pcm";
+}
+export declare class OpenAITTS implements TextToSpeechProvider {
+    private readonly openaiClient;
+    private readonly config;
+    private readonly logger;
+    constructor(ttsConfig: TTSConfig, parentLogger: pino.Logger);
+    getConfig(): TTSConfig;
+    synthesizeSpeech(text: string): Promise<SpeechStreamResult>;
+}
+//# sourceMappingURL=tts.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"tts.d.ts","sourceRoot":"","sources":["../../../../../../src/server/speech/providers/openai/tts.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,IAAI,MAAM,MAAM,CAAC;AAG7B,OAAO,KAAK,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,0BAA0B,CAAC;AAEzF,YAAY,EAAE,kBAAkB,EAAE,CAAC;AAEnC,MAAM,WAAW,SAAS;IACxB,MAAM,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,OAAO,GAAG,UAAU,CAAC;IAC7B,KAAK,CAAC,EAAE,OAAO,GAAG,MAAM,GAAG,OAAO,GAAG,MAAM,GAAG,MAAM,GAAG,SAAS,CAAC;IACjE,cAAc,CAAC,EAAE,KAAK,GAAG,MAAM,GAAG,KAAK,GAAG,MAAM,GAAG,KAAK,GAAG,KAAK,CAAC;CAClE;AAED,qBAAa,SAAU,YAAW,oBAAoB;IACpD,OAAO,CAAC,QAAQ,CAAC,YAAY,CAAS;IACtC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAY;IACnC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAc;gBAEzB,SAAS,EAAE,SAAS,EAAE,YAAY,EAAE,IAAI,CAAC,MAAM;IAkBpD,SAAS,IAAI,SAAS;IAIhB,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,kBAAkB,CAAC;CAwCzE"}

package/dist/server/server/speech/providers/openai/tts.js

@@ -0,0 +1,46 @@
+import OpenAI from "openai";
+export class OpenAITTS {
+    constructor(ttsConfig, parentLogger) {
+        this.config = {
+            model: "tts-1",
+            voice: "alloy",
+            responseFormat: "pcm",
+            ...ttsConfig,
+        };
+        this.logger = parentLogger.child({ module: "agent", provider: "openai", component: "tts" });
+        this.openaiClient = new OpenAI({
+            apiKey: ttsConfig.apiKey,
+        });
+        this.logger.info({ voice: this.config.voice, model: this.config.model, format: this.config.responseFormat }, "TTS (OpenAI) initialized");
+    }
+    getConfig() {
+        return this.config;
+    }
+    async synthesizeSpeech(text) {
+        if (!text || text.trim().length === 0) {
+            throw new Error("Cannot synthesize empty text");
+        }
+        const startTime = Date.now();
+        try {
+            this.logger.debug({ textLength: text.length, preview: text.substring(0, 50) }, "Synthesizing speech");
+            const response = await this.openaiClient.audio.speech.create({
+                model: this.config.model,
+                voice: this.config.voice,
+                input: text,
+                response_format: this.config.responseFormat,
+            });
+            const audioStream = response.body;
+            const duration = Date.now() - startTime;
+            this.logger.debug({ duration }, "Speech synthesis stream ready");
+            return {
+                stream: audioStream,
+                format: this.config.responseFormat || "mp3",
+            };
+        }
+        catch (error) {
+            this.logger.error({ err: error }, "Speech synthesis error");
+            throw new Error(`TTS synthesis failed: ${error.message}`);
+        }
+    }
+}
+//# sourceMappingURL=tts.js.map

package/dist/server/server/speech/providers/openai/tts.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"tts.js","sourceRoot":"","sources":["../../../../../../src/server/speech/providers/openai/tts.ts"],"names":[],"mappings":"AACA,OAAO,MAAM,MAAM,QAAQ,CAAC;AAa5B,MAAM,OAAO,SAAS;IAKpB,YAAY,SAAoB,EAAE,YAAyB;QACzD,IAAI,CAAC,MAAM,GAAG;YACZ,KAAK,EAAE,OAAO;YACd,KAAK,EAAE,OAAO;YACd,cAAc,EAAE,KAAK;YACrB,GAAG,SAAS;SACb,CAAC;QACF,IAAI,CAAC,MAAM,GAAG,YAAY,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC,CAAC;QAC5F,IAAI,CAAC,YAAY,GAAG,IAAI,MAAM,CAAC;YAC7B,MAAM,EAAE,SAAS,CAAC,MAAM;SACzB,CAAC,CAAC;QAEH,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,EAAE,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,CAAC,cAAc,EAAE,EAC1F,0BAA0B,CAC3B,CAAC;IACJ,CAAC;IAEM,SAAS;QACd,OAAO,IAAI,CAAC,MAAM,CAAC;IACrB,CAAC;IAEM,KAAK,CAAC,gBAAgB,CAAC,IAAY;QACxC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACtC,MAAM,IAAI,KAAK,CAAC,8BAA8B,CAAC,CAAC;QAClD,CAAC;QAED,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAE7B,IAAI,CAAC;YACH,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,EAAE,UAAU,EAAE,IAAI,CAAC,MAAM,EAAE,OAAO,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE,EAC3D,qBAAqB,CACtB,CAAC;YAEF,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC;gBAC3D,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,KAAM;gBACzB,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,KAAM;gBACzB,KAAK,EAAE,IAAI;gBACX,eAAe,EAAE,IAAI,CAAC,MAAM,CAAC,cAMpB;aACV,CAAC,CAAC;YAEH,MAAM,WAAW,GAAG,QAAQ,CAAC,IAA2B,CAAC;YAEzD,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,CAAC;YACxC,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,QAAQ,EAAE,EAAE,+BAA+B,CAAC,CAAC;YAEjE,OAAO;gBACL,MAAM,EAAE,WAAW;gBACnB,MAAM,EAAE,IAAI,CAAC,MAAM,CAAC,cAAc,IAAI,KAAK;aAC5C,CAAC;QACJ,CAAC;QAAC,OAAO,KAAU,EAAE,CAAC;YACpB,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,wBAAwB,CAAC,CAAC;YAC5D,MAAM,IAAI,KAAK,CAAC,yBAAyB,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;QAC5D,CAAC;IACH,CAAC;CACF"}

package/dist/server/server/speech/speech-config-resolver.d.ts

@@ -0,0 +1,11 @@
+import type { PersistedConfig } from "../persisted-config.js";
+import type { SeaworkOpenAIConfig, SeaworkSpeechConfig } from "../bootstrap.js";
+export declare function resolveSpeechConfig(params: {
+    seaworkHome: string;
+    env: NodeJS.ProcessEnv;
+    persisted: PersistedConfig;
+}): {
+    openai: SeaworkOpenAIConfig | undefined;
+    speech: SeaworkSpeechConfig;
+};
+//# sourceMappingURL=speech-config-resolver.d.ts.map

package/dist/server/server/speech/speech-config-resolver.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"speech-config-resolver.d.ts","sourceRoot":"","sources":["../../../../src/server/speech/speech-config-resolver.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AAC9D,OAAO,KAAK,EAAE,mBAAmB,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAC;AAiIhF,wBAAgB,mBAAmB,CAAC,MAAM,EAAE;IAC1C,WAAW,EAAE,MAAM,CAAC;IACpB,GAAG,EAAE,MAAM,CAAC,UAAU,CAAC;IACvB,SAAS,EAAE,eAAe,CAAC;CAC5B,GAAG;IACF,MAAM,EAAE,mBAAmB,GAAG,SAAS,CAAC;IACxC,MAAM,EAAE,mBAAmB,CAAC;CAC7B,CA0BA"}

package/dist/server/server/speech/speech-config-resolver.js

@@ -0,0 +1,104 @@
+import { z } from "zod";
+import { resolveLocalSpeechConfig } from "./providers/local/config.js";
+import { resolveOpenAiSpeechConfig } from "./providers/openai/config.js";
+import { SpeechProviderIdSchema, } from "./speech-types.js";
+const OptionalSpeechProviderSchema = z
+    .string()
+    .trim()
+    .toLowerCase()
+    .pipe(SpeechProviderIdSchema)
+    .optional();
+const OptionalBooleanFlagSchema = z
+    .union([z.boolean(), z.string().trim().toLowerCase()])
+    .optional()
+    .transform((value) => {
+    if (typeof value === "boolean") {
+        return value;
+    }
+    if (value === undefined) {
+        return undefined;
+    }
+    if (["1", "true", "yes", "y", "on"].includes(value)) {
+        return true;
+    }
+    if (["0", "false", "no", "n", "off"].includes(value)) {
+        return false;
+    }
+    return undefined;
+});
+const RequestedSpeechProvidersSchema = z.object({
+    dictationStt: OptionalSpeechProviderSchema.default("local"),
+    voiceTurnDetection: OptionalSpeechProviderSchema.default("local"),
+    voiceStt: OptionalSpeechProviderSchema.default("local"),
+    voiceTts: OptionalSpeechProviderSchema.default("local"),
+});
+function resolveOptionalBooleanFlag(value) {
+    return OptionalBooleanFlagSchema.parse(value) ?? true;
+}
+function resolveRequestedSpeechProviders(params) {
+    const resolveFeatureProvider = (configuredValue, parsedValue, enabled) => ({
+        provider: parsedValue,
+        explicit: configuredValue !== undefined,
+        enabled,
+    });
+    const voiceModeEnabled = resolveOptionalBooleanFlag(params.env.SEAWORK_VOICE_MODE_ENABLED ?? params.persisted.features?.voiceMode?.enabled);
+    const featureProviders = {
+        dictationStt: {
+            configuredValue: params.env.SEAWORK_DICTATION_STT_PROVIDER ??
+                params.persisted.features?.dictation?.stt?.provider,
+            enabled: resolveOptionalBooleanFlag(params.env.SEAWORK_DICTATION_ENABLED ?? params.persisted.features?.dictation?.enabled),
+        },
+        voiceTurnDetection: {
+            configuredValue: params.env.SEAWORK_VOICE_TURN_DETECTION_PROVIDER ??
+                params.persisted.features?.voiceMode?.turnDetection?.provider,
+            enabled: voiceModeEnabled,
+        },
+        voiceStt: {
+            configuredValue: params.env.SEAWORK_VOICE_STT_PROVIDER ??
+                params.persisted.features?.voiceMode?.stt?.provider,
+            enabled: voiceModeEnabled,
+        },
+        voiceTts: {
+            configuredValue: params.env.SEAWORK_VOICE_TTS_PROVIDER ??
+                params.persisted.features?.voiceMode?.tts?.provider,
+            enabled: voiceModeEnabled,
+        },
+    };
+    const parsed = RequestedSpeechProvidersSchema.parse({
+        dictationStt: featureProviders.dictationStt.configuredValue ?? "local",
+        voiceTurnDetection: featureProviders.voiceTurnDetection.configuredValue ?? "local",
+        voiceStt: featureProviders.voiceStt.configuredValue ?? "local",
+        voiceTts: featureProviders.voiceTts.configuredValue ?? "local",
+    });
+    return {
+        dictationStt: resolveFeatureProvider(featureProviders.dictationStt.configuredValue, parsed.dictationStt, featureProviders.dictationStt.enabled),
+        voiceTurnDetection: resolveFeatureProvider(featureProviders.voiceTurnDetection.configuredValue, parsed.voiceTurnDetection, featureProviders.voiceTurnDetection.enabled),
+        voiceStt: resolveFeatureProvider(featureProviders.voiceStt.configuredValue, parsed.voiceStt, featureProviders.voiceStt.enabled),
+        voiceTts: resolveFeatureProvider(featureProviders.voiceTts.configuredValue, parsed.voiceTts, featureProviders.voiceTts.enabled),
+    };
+}
+export function resolveSpeechConfig(params) {
+    const providers = resolveRequestedSpeechProviders({
+        env: params.env,
+        persisted: params.persisted,
+    });
+    const local = resolveLocalSpeechConfig({
+        seaworkHome: params.seaworkHome,
+        env: params.env,
+        persisted: params.persisted,
+        providers,
+    });
+    const openai = resolveOpenAiSpeechConfig({
+        env: params.env,
+        persisted: params.persisted,
+        providers,
+    });
+    return {
+        openai,
+        speech: {
+            providers,
+            ...(local.local ? { local: local.local } : {}),
+        },
+    };
+}
+//# sourceMappingURL=speech-config-resolver.js.map

package/dist/server/server/speech/speech-config-resolver.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"speech-config-resolver.js","sourceRoot":"","sources":["../../../../src/server/speech/speech-config-resolver.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAIxB,OAAO,EAAE,wBAAwB,EAAE,MAAM,6BAA6B,CAAC;AACvE,OAAO,EAAE,yBAAyB,EAAE,MAAM,8BAA8B,CAAC;AACzE,OAAO,EACL,sBAAsB,GAGvB,MAAM,mBAAmB,CAAC;AAE3B,MAAM,4BAA4B,GAAG,CAAC;KACnC,MAAM,EAAE;KACR,IAAI,EAAE;KACN,WAAW,EAAE;KACb,IAAI,CAAC,sBAAsB,CAAC;KAC5B,QAAQ,EAAE,CAAC;AAEd,MAAM,yBAAyB,GAAG,CAAC;KAChC,KAAK,CAAC,CAAC,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC,CAAC;KACrD,QAAQ,EAAE;KACV,SAAS,CAAC,CAAC,KAAK,EAAE,EAAE;IACnB,IAAI,OAAO,KAAK,KAAK,SAAS,EAAE,CAAC;QAC/B,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QACxB,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,IAAI,CAAC,GAAG,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACpD,OAAO,IAAI,CAAC;IACd,CAAC;IACD,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,GAAG,EAAE,KAAK,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACrD,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC,CAAC,CAAC;AAEL,MAAM,8BAA8B,GAAG,CAAC,CAAC,MAAM,CAAC;IAC9C,YAAY,EAAE,4BAA4B,CAAC,OAAO,CAAC,OAAO,CAAC;IAC3D,kBAAkB,EAAE,4BAA4B,CAAC,OAAO,CAAC,OAAO,CAAC;IACjE,QAAQ,EAAE,4BAA4B,CAAC,OAAO,CAAC,OAAO,CAAC;IACvD,QAAQ,EAAE,4BAA4B,CAAC,OAAO,CAAC,OAAO,CAAC;CACxD,CAAC,CAAC;AAEH,SAAS,0BAA0B,CAAC,KAAc;IAChD,OAAO,yBAAyB,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC;AACxD,CAAC;AAED,SAAS,+BAA+B,CAAC,MAGxC;IACC,MAAM,sBAAsB,GAAG,CAC7B,eAAmC,EACnC,WAAmD,EACnD,OAAgB,EACS,EAAE,CAAC,CAAC;QAC7B,QAAQ,EAAE,WAAW;QACrB,QAAQ,EAAE,eAAe,KAAK,SAAS;QACvC,OAAO;KACR,CAAC,CAAC;IAEH,MAAM,gBAAgB,GAAG,0BAA0B,CACjD,MAAM,CAAC,GAAG,CAAC,0BAA0B,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,SAAS,EAAE,OAAO,CACvF,CAAC;IACF,MAAM,gBAAgB,GAAG;QACvB,YAAY,EAAE;YACZ,eAAe,EACb,MAAM,CAAC,GAAG,CAAC,8BAA8B;gBACzC,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,SAAS,EAAE,GAAG,EAAE,QAAQ;YACrD,OAAO,EAAE,0BAA0B,CACjC,MAAM,CAAC,GAAG,CAAC,yBAAyB,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,SAAS,EAAE,OAAO,CACtF;SACF;QACD,kBAAkB,EAAE;YAClB,eAAe,EACb,MAAM,CAAC,GAAG,CAAC,qCAAqC;gBAChD,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,SAAS,EAAE,aAAa,EAAE,QAAQ;YAC/D,OAAO,EAAE,gBAAgB;SAC1B;QACD,QAAQ,EAAE;YACR,eAAe,EACb,MAAM,CAAC,GAAG,CAAC,0BAA0B;gBACrC,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,SAAS,EAAE,GAAG,EAAE,QAAQ;YACrD,OAAO,EAAE,gBAAgB;SAC1B;QACD,QAAQ,EAAE;YACR,eAAe,EACb,MAAM,CAAC,GAAG,CAAC,0BAA0B;gBACrC,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,SAAS,EAAE,GAAG,EAAE,QAAQ;YACrD,OAAO,EAAE,gBAAgB;SAC1B;KAOF,CAAC;IAEF,MAAM,MAAM,GAAG,8BAA8B,CAAC,KAAK,CAAC;QAClD,YAAY,EAAE,gBAAgB,CAAC,YAAY,CAAC,eAAe,IAAI,OAAO;QACtE,kBAAkB,EAAE,gBAAgB,CAAC,kBAAkB,CAAC,eAAe,IAAI,OAAO;QAClF,QAAQ,EAAE,gBAAgB,CAAC,QAAQ,CAAC,eAAe,IAAI,OAAO;QAC9D,QAAQ,EAAE,gBAAgB,CAAC,QAAQ,CAAC,eAAe,IAAI,OAAO;KAC/D,CAAC,CAAC;IAEH,OAAO;QACL,YAAY,EAAE,sBAAsB,CAClC,gBAAgB,CAAC,YAAY,CAAC,eAAe,EAC7C,MAAM,CAAC,YAAY,EACnB,gBAAgB,CAAC,YAAY,CAAC,OAAO,CACtC;QACD,kBAAkB,EAAE,sBAAsB,CACxC,gBAAgB,CAAC,kBAAkB,CAAC,eAAe,EACnD,MAAM,CAAC,kBAAkB,EACzB,gBAAgB,CAAC,kBAAkB,CAAC,OAAO,CAC5C;QACD,QAAQ,EAAE,sBAAsB,CAC9B,gBAAgB,CAAC,QAAQ,CAAC,eAAe,EACzC,MAAM,CAAC,QAAQ,EACf,gBAAgB,CAAC,QAAQ,CAAC,OAAO,CAClC;QACD,QAAQ,EAAE,sBAAsB,CAC9B,gBAAgB,CAAC,QAAQ,CAAC,eAAe,EACzC,MAAM,CAAC,QAAQ,EACf,gBAAgB,CAAC,QAAQ,CAAC,OAAO,CAClC;KACF,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,MAInC;IAIC,MAAM,SAAS,GAAG,+BAA+B,CAAC;QAChD,GAAG,EAAE,MAAM,CAAC,GAAG;QACf,SAAS,EAAE,MAAM,CAAC,SAAS;KAC5B,CAAC,CAAC;IAEH,MAAM,KAAK,GAAG,wBAAwB,CAAC;QACrC,WAAW,EAAE,MAAM,CAAC,WAAW;QAC/B,GAAG,EAAE,MAAM,CAAC,GAAG;QACf,SAAS,EAAE,MAAM,CAAC,SAAS;QAC3B,SAAS;KACV,CAAC,CAAC;IAEH,MAAM,MAAM,GAAG,yBAAyB,CAAC;QACvC,GAAG,EAAE,MAAM,CAAC,GAAG;QACf,SAAS,EAAE,MAAM,CAAC,SAAS;QAC3B,SAAS;KACV,CAAC,CAAC;IAEH,OAAO;QACL,MAAM;QACN,MAAM,EAAE;YACN,SAAS;YACT,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;SAC/C;KACF,CAAC;AACJ,CAAC"}

package/dist/server/server/speech/speech-provider.d.ts

@@ -0,0 +1,59 @@
+import type pino from "pino";
+import type { Readable } from "node:stream";
+export interface LogprobToken {
+    token: string;
+    logprob: number;
+    bytes?: number[];
+}
+export interface TranscriptionResult {
+    text: string;
+    language?: string;
+    duration?: number;
+    logprobs?: LogprobToken[];
+    avgLogprob?: number;
+    isLowConfidence?: boolean;
+}
+export interface StreamingTranscriptionCommittedEvent {
+    segmentId: string;
+    previousSegmentId: string | null;
+}
+export interface StreamingTranscriptionEvent {
+    segmentId: string;
+    transcript: string;
+    isFinal: boolean;
+    language?: string;
+    logprobs?: LogprobToken[];
+    avgLogprob?: number;
+    isLowConfidence?: boolean;
+}
+export type StreamingTranscriptionSession = {
+    /**
+     * Required PCM16LE sample rate for `appendPcm16()`.
+     * Callers are responsible for resampling before appending.
+     */
+    requiredSampleRate: number;
+    connect(): Promise<void>;
+    appendPcm16(pcm16le: Buffer): void;
+    commit(): void;
+    clear(): void;
+    close(): void;
+    on(event: "committed", handler: (payload: StreamingTranscriptionCommittedEvent) => void): unknown;
+    on(event: "transcript", handler: (payload: StreamingTranscriptionEvent) => void): unknown;
+    on(event: "error", handler: (err: unknown) => void): unknown;
+};
+export interface SpeechToTextProvider {
+    id: "openai" | "local" | (string & {});
+    createSession(params: {
+        logger: pino.Logger;
+        language?: string;
+        prompt?: string;
+    }): StreamingTranscriptionSession;
+}
+export interface SpeechStreamResult {
+    stream: Readable;
+    format: string;
+}
+export interface TextToSpeechProvider {
+    synthesizeSpeech(text: string): Promise<SpeechStreamResult>;
+}
+//# sourceMappingURL=speech-provider.d.ts.map

package/dist/server/server/speech/speech-provider.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"speech-provider.d.ts","sourceRoot":"","sources":["../../../../src/server/speech/speech-provider.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,IAAI,MAAM,MAAM,CAAC;AAC7B,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAE5C,MAAM,WAAW,YAAY;IAC3B,KAAK,EAAE,MAAM,CAAC;IACd,OAAO,EAAE,MAAM,CAAC;IAChB,KAAK,CAAC,EAAE,MAAM,EAAE,CAAC;CAClB;AAED,MAAM,WAAW,mBAAmB;IAClC,IAAI,EAAE,MAAM,CAAC;IACb,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,YAAY,EAAE,CAAC;IAC1B,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,eAAe,CAAC,EAAE,OAAO,CAAC;CAC3B;AAED,MAAM,WAAW,oCAAoC;IACnD,SAAS,EAAE,MAAM,CAAC;IAClB,iBAAiB,EAAE,MAAM,GAAG,IAAI,CAAC;CAClC;AAED,MAAM,WAAW,2BAA2B;IAC1C,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,OAAO,EAAE,OAAO,CAAC;IACjB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,YAAY,EAAE,CAAC;IAC1B,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,eAAe,CAAC,EAAE,OAAO,CAAC;CAC3B;AAED,MAAM,MAAM,6BAA6B,GAAG;IAC1C;;;OAGG;IACH,kBAAkB,EAAE,MAAM,CAAC;IAE3B,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC;IACzB,WAAW,CAAC,OAAO,EAAE,MAAM,GAAG,IAAI,CAAC;IACnC,MAAM,IAAI,IAAI,CAAC;IACf,KAAK,IAAI,IAAI,CAAC;IACd,KAAK,IAAI,IAAI,CAAC;IAEd,EAAE,CAAC,KAAK,EAAE,WAAW,EAAE,OAAO,EAAE,CAAC,OAAO,EAAE,oCAAoC,KAAK,IAAI,GAAG,OAAO,CAAC;IAClG,EAAE,CAAC,KAAK,EAAE,YAAY,EAAE,OAAO,EAAE,CAAC,OAAO,EAAE,2BAA2B,KAAK,IAAI,GAAG,OAAO,CAAC;IAC1F,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,CAAC,GAAG,EAAE,OAAO,KAAK,IAAI,GAAG,OAAO,CAAC;CAC9D,CAAC;AAEF,MAAM,WAAW,oBAAoB;IACnC,EAAE,EAAE,QAAQ,GAAG,OAAO,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IACvC,aAAa,CAAC,MAAM,EAAE;QACpB,MAAM,EAAE,IAAI,CAAC,MAAM,CAAC;QACpB,QAAQ,CAAC,EAAE,MAAM,CAAC;QAClB,MAAM,CAAC,EAAE,MAAM,CAAC;KACjB,GAAG,6BAA6B,CAAC;CACnC;AAED,MAAM,WAAW,kBAAkB;IACjC,MAAM,EAAE,QAAQ,CAAC;IACjB,MAAM,EAAE,MAAM,CAAC;CAChB;AAED,MAAM,WAAW,oBAAoB;IACnC,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,kBAAkB,CAAC,CAAC;CAC7D"}

package/dist/server/server/speech/speech-provider.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=speech-provider.js.map

package/dist/server/server/speech/speech-provider.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"speech-provider.js","sourceRoot":"","sources":["../../../../src/server/speech/speech-provider.ts"],"names":[],"mappings":""}

package/dist/server/server/speech/speech-runtime.d.ts

@@ -0,0 +1,43 @@
+import type { Logger } from "pino";
+import type { SeaworkOpenAIConfig, SeaworkSpeechConfig } from "../bootstrap.js";
+import type { LocalSpeechModelId } from "./providers/local/config.js";
+import type { SpeechToTextProvider, TextToSpeechProvider } from "./speech-provider.js";
+import type { TurnDetectionProvider } from "./turn-detection-provider.js";
+export type SpeechReadinessReasonCode = "ready" | "disabled" | "model_download_in_progress" | "models_missing" | "model_download_failed" | "turn_detection_unavailable" | "stt_unavailable" | "tts_unavailable";
+export type SpeechReadinessState = {
+    enabled: boolean;
+    available: boolean;
+    reasonCode: SpeechReadinessReasonCode;
+    message: string;
+    retryable: boolean;
+    missingModelIds: LocalSpeechModelId[];
+};
+export type SpeechReadinessSnapshot = {
+    generatedAt: string;
+    requiredLocalModelIds: LocalSpeechModelId[];
+    missingLocalModelIds: LocalSpeechModelId[];
+    download: {
+        inProgress: boolean;
+        error: string | null;
+    };
+    realtimeVoice: SpeechReadinessState;
+    dictation: SpeechReadinessState;
+    voiceFeature: SpeechReadinessState;
+};
+export type SpeechService = {
+    resolveStt: () => SpeechToTextProvider | null;
+    resolveTts: () => TextToSpeechProvider | null;
+    resolveTurnDetection: () => TurnDetectionProvider | null;
+    resolveDictationStt: () => SpeechToTextProvider | null;
+    getReadiness: () => SpeechReadinessSnapshot;
+    onReadinessChange: (listener: (snapshot: SpeechReadinessSnapshot) => void) => () => void;
+    start: () => void;
+    stop: () => void;
+    ready: Promise<void>;
+};
+export declare function createSpeechService(params: {
+    logger: Logger;
+    openaiConfig?: SeaworkOpenAIConfig;
+    speechConfig?: SeaworkSpeechConfig;
+}): SpeechService;
+//# sourceMappingURL=speech-runtime.d.ts.map

package/dist/server/server/speech/speech-runtime.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"speech-runtime.d.ts","sourceRoot":"","sources":["../../../../src/server/speech/speech-runtime.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,MAAM,CAAC;AAEnC,OAAO,KAAK,EAAE,mBAAmB,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAC;AAChF,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,6BAA6B,CAAC;AAYtE,OAAO,KAAK,EAAE,oBAAoB,EAAE,oBAAoB,EAAE,MAAM,sBAAsB,CAAC;AAEvF,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,8BAA8B,CAAC;AAI1E,MAAM,MAAM,yBAAyB,GACjC,OAAO,GACP,UAAU,GACV,4BAA4B,GAC5B,gBAAgB,GAChB,uBAAuB,GACvB,4BAA4B,GAC5B,iBAAiB,GACjB,iBAAiB,CAAC;AAEtB,MAAM,MAAM,oBAAoB,GAAG;IACjC,OAAO,EAAE,OAAO,CAAC;IACjB,SAAS,EAAE,OAAO,CAAC;IACnB,UAAU,EAAE,yBAAyB,CAAC;IACtC,OAAO,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,OAAO,CAAC;IACnB,eAAe,EAAE,kBAAkB,EAAE,CAAC;CACvC,CAAC;AAEF,MAAM,MAAM,uBAAuB,GAAG;IACpC,WAAW,EAAE,MAAM,CAAC;IACpB,qBAAqB,EAAE,kBAAkB,EAAE,CAAC;IAC5C,oBAAoB,EAAE,kBAAkB,EAAE,CAAC;IAC3C,QAAQ,EAAE;QACR,UAAU,EAAE,OAAO,CAAC;QACpB,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;KACtB,CAAC;IACF,aAAa,EAAE,oBAAoB,CAAC;IACpC,SAAS,EAAE,oBAAoB,CAAC;IAChC,YAAY,EAAE,oBAAoB,CAAC;CACpC,CAAC;AA6RF,MAAM,MAAM,aAAa,GAAG;IAC1B,UAAU,EAAE,MAAM,oBAAoB,GAAG,IAAI,CAAC;IAC9C,UAAU,EAAE,MAAM,oBAAoB,GAAG,IAAI,CAAC;IAC9C,oBAAoB,EAAE,MAAM,qBAAqB,GAAG,IAAI,CAAC;IACzD,mBAAmB,EAAE,MAAM,oBAAoB,GAAG,IAAI,CAAC;IACvD,YAAY,EAAE,MAAM,uBAAuB,CAAC;IAC5C,iBAAiB,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,EAAE,uBAAuB,KAAK,IAAI,KAAK,MAAM,IAAI,CAAC;IACzF,KAAK,EAAE,MAAM,IAAI,CAAC;IAClB,IAAI,EAAE,MAAM,IAAI,CAAC;IACjB,KAAK,EAAE,OAAO,CAAC,IAAI,CAAC,CAAC;CACtB,CAAC;AAEF,wBAAgB,mBAAmB,CAAC,MAAM,EAAE;IAC1C,MAAM,EAAE,MAAM,CAAC;IACf,YAAY,CAAC,EAAE,mBAAmB,CAAC;IACnC,YAAY,CAAC,EAAE,mBAAmB,CAAC;CACpC,GAAG,aAAa,CAyWhB"}