TTs whisper

backend/src/controllers/audio.controller.ts

@@ -0,0 +1,119 @@
+import { Router } from 'express';
+import multer from 'multer';
+import { execFile } from 'child_process';
+import { tmpdir } from 'os';
+import { join } from 'path';
+import { promisify } from 'util';
+import { writeFile, unlink } from 'fs';
+import { raspiService } from '../services/raspi.service.js';
+
+const execFileAsync = promisify(execFile);
+const writeFileAsync = promisify(writeFile);
+const unlinkAsync = promisify(unlink);
+
+const router = Router();
+
+const upload = multer({ storage: multer.memoryStorage() });
+
+router.get('/incoming', async (_req, res) => {
+  try {
+    const result = await raspiService.listIncomingAudio();
+    res.json(result);
+  } catch (err: unknown) {
+    const message = err instanceof Error ? err.message : 'Unknown error';
+    res.status(500).json({ error: `List incoming failed: ${message}` });
+  }
+});
+
+router.post('/lock/:filename', async (req, res) => {
+  try {
+    const { filename } = req.params;
+    const result = await raspiService.lockAudio({ filename });
+    res.json(result);
+  } catch (err: unknown) {
+    const message = err instanceof Error ? err.message : 'Unknown error';
+    res.status(500).json({ error: `Lock audio failed: ${message}` });
+  }
+});
+
+router.post('/unlock/:filename', async (req, res) => {
+  try {
+    const { filename } = req.params;
+    const result = await raspiService.unlockAudio({ filename });
+    res.json(result);
+  } catch (err: unknown) {
+    const message = err instanceof Error ? err.message : 'Unknown error';
+    res.status(500).json({ error: `Unlock audio failed: ${message}` });
+  }
+});
+
+router.post('/cancel/:filename', async (req, res) => {
+  try {
+    const { filename } = req.params;
+    const result = await raspiService.cancelAudio({ filename });
+    res.json(result);
+  } catch (err: unknown) {
+    const message = err instanceof Error ? err.message : 'Unknown error';
+    res.status(500).json({ error: `Cancel audio failed: ${message}` });
+  }
+});
+
+router.post('/process/:filename', async (req, res) => {
+  try {
+    const { filename } = req.params;
+    const result = await raspiService.processAudio({ filename });
+    res.json(result);
+  } catch (err: unknown) {
+    const message = err instanceof Error ? err.message : 'Unknown error';
+    res.status(500).json({ error: `Process audio failed: ${message}` });
+  }
+});
+
+const whisperModel = process.env.WHISPER_MODEL ?? 'base';
+const whisperLanguage = process.env.WHISPER_LANGUAGE ?? 'ca';
+
+router.post('/upload', upload.single('file'), async (req, res) => {
+  let tmpFile: string | undefined;
+  try {
+    if (!req.file) {
+      return res.status(400).json({ error: 'No audio file provided' });
+    }
+
+    const ext = req.file.originalname.split('.').pop()?.toLowerCase() || 'wav';
+    tmpFile = join(tmpdir(), `quibot-audio-${Date.now()}.${ext}`);
+    await writeFileAsync(tmpFile, req.file.buffer);
+
+    console.log(`[whisper] Model: ${whisperModel}, Language: ${whisperLanguage}, File: ${tmpFile}`);
+
+    const { stdout, stderr } = await execFileAsync('whisper', [
+      tmpFile,
+      '--model', whisperModel,
+      '--language', whisperLanguage,
+      '--output_format', 'txt',
+    ], { maxBuffer: 50 * 1024 * 1024 });
+
+    if (stderr) {
+      console.log(`[whisper] stderr: ${stderr}`);
+    }
+
+    const transcription = stdout.trim();
+
+    res.json({
+      transcription,
+      originalFilename: req.file.originalname,
+    });
+  } catch (err: unknown) {
+    const message = err instanceof Error ? err.message : 'Unknown error';
+    res.status(500).json({ error: `Audio transcription failed: ${message}` });
+  } finally {
+    if (tmpFile) {
+      try {
+        await unlinkAsync(tmpFile);
+      } catch {
+        // ignore cleanup errors
+      }
+    }
+  }
+});
+
+export default router;