Add realtime loopback STT prototype

2026-05-02 20:20:54 +09:00
parent 10e0dd75db
commit 5775c4809a
17 changed files with 1034 additions and 0 deletions
--- a/src/index.ts
+++ b/src/index.ts
@@ -0,0 +1,107 @@
+import process from "node:process";
+
+import { loadConfig } from "./config.js";
+import { Logger } from "./logger.js";
+import { printAudioDevices, spawnLoopbackCapture } from "./audio/capture.js";
+import { RealtimeSegmenter } from "./audio/realtime-segmenter.js";
+import { FasterWhisperSttService } from "./services/faster-whisper-stt.js";
+
+const mode = process.argv[2] ?? "loopback";
+
+async function runLoopback(): Promise<void> {
+  const config = loadConfig();
+  const logger = new Logger(config.LOG_LEVEL);
+  const stt = new FasterWhisperSttService(config, logger);
+
+  await stt.warmup();
+
+  const transcriptionQueue: Buffer[] = [];
+  let transcribing = false;
+
+  const runNext = async (): Promise<void> => {
+    if (transcribing) {
+      return;
+    }
+    const next = transcriptionQueue.shift();
+    if (!next) {
+      return;
+    }
+
+    transcribing = true;
+    try {
+      const text = await stt.transcribePcm16(next);
+      if (!text) {
+        logger.info("빈 전사 결과");
+      } else {
+        logger.info("Transcript", text);
+        if (config.DEBUG_TRANSCRIPTS) {
+          console.log(`\n[text] ${text}\n`);
+        }
+      }
+    } catch (error) {
+      logger.warn("STT failed", error);
+    } finally {
+      transcribing = false;
+      void runNext();
+    }
+  };
+
+  const segmenter = new RealtimeSegmenter({
+    onSegment: (pcm16) => {
+      transcriptionQueue.push(pcm16);
+      void runNext();
+    },
+  });
+
+  const capture = spawnLoopbackCapture(config, logger);
+  capture.stdout.on("data", (chunk: Buffer) => {
+    segmenter.pushChunk(chunk);
+  });
+  capture.stderr.on("data", (chunk: Buffer) => {
+    const text = chunk.toString().trim();
+    if (text) {
+      logger.debug("[capture]", text);
+    }
+  });
+  capture.on("exit", (code, signal) => {
+    logger.warn("capture exited", { code, signal });
+  });
+
+  console.log("실시간 출력장치 STT를 시작합니다. Ctrl+C 로 종료합니다.");
+  console.log(`source: ${config.AUDIO_SOURCE ?? "unset"}`);
+  console.log(`model: ${config.WHISPER_MODEL}`);
+  console.log(`language: ${config.WHISPER_LANGUAGE}`);
+
+  const shutdown = async (): Promise<void> => {
+    if (!capture.killed) {
+      capture.kill("SIGTERM");
+    }
+    await stt.destroy();
+    process.exit(0);
+  };
+
+  process.on("SIGINT", () => {
+    void shutdown();
+  });
+  process.on("SIGTERM", () => {
+    void shutdown();
+  });
+}
+
+async function main(): Promise<void> {
+  switch (mode) {
+    case "devices":
+      await printAudioDevices();
+      return;
+    case "loopback":
+      await runLoopback();
+      return;
+    default:
+      throw new Error(`알 수 없는 실행 모드입니다: ${mode}. 사용 가능: loopback, devices`);
+  }
+}
+
+void main().catch((error) => {
+  console.error(error instanceof Error ? error.message : String(error));
+  process.exit(1);
+});