Fix speech-to-text with Groq Whisper

server.js

@@ -7,6 +7,7 @@ const googleTTS = require('google-tts-api');
 require('dotenv').config();
 
 const HUGGINGFACE_API_TOKEN = process.env.HUGGINGFACEHUB_API_TOKEN || process.env.HF_API_TOKEN;
+const GROQ_API_KEY = process.env.GROQ_API_KEY;
 const nllbLangMap = {
     'en': 'eng_Latn',
     'es': 'spa_Latn',
@@ -77,7 +78,7 @@ function loadSslOptions() {
     // Generate a temporary self-signed certificate for localhost
     console.warn('No SSL cert/key provided. Generating a temporary self-signed certificate for https://localhost:' + SSL_PORT);
     const attrs = [{ name: 'commonName', value: 'localhost' }];
-    const pems = selfsigned.generate(attrs, { days: 365 });
+    const pems = selfsigned.generate(attrs, { days: 365, keySize: 2048 });
     return { key: pems.private, cert: pems.cert };
 }
 
@@ -332,7 +333,25 @@ const translateViaLibreTranslate = async (text, sourceLang, targetLang) => {
     }
 };
 
-// Speech-to-text endpoint using Hugging Face Whisper
+function getTranscriptionFilename(contentType) {
+    const mimeType = (contentType || 'audio/webm').split(';')[0].trim().toLowerCase();
+
+    if (mimeType.includes('mp4') || mimeType.includes('m4a') || mimeType.includes('aac')) {
+        return 'audio.mp4';
+    }
+
+    if (mimeType.includes('webm')) {
+        return 'audio.webm';
+    }
+
+    if (mimeType.includes('wav')) {
+        return 'audio.wav';
+    }
+
+    return 'audio.webm';
+}
+
+// Speech-to-text endpoint using Groq Whisper
 app.post('/api/transcribe', async (req, res) => {
     try {
         const audioBuffer = req.body;
@@ -340,29 +359,40 @@ app.post('/api/transcribe', async (req, res) => {
             return res.status(400).json({ error: 'No audio provided' });
         }
 
-        if (!HUGGINGFACE_API_TOKEN) {
-            return res.status(500).json({ error: 'Hugging Face token not configured' });
+        if (!GROQ_API_KEY) {
+            return res.status(503).json({ error: 'Speech-to-text is temporarily unavailable: GROQ API key is not configured.' });
         }
 
-        const contentType = req.headers['content-type'] || 'application/octet-stream';
-        const response = await axios.post(
-            'https://api-inference.huggingface.co/models/openai/whisper-large-v3',
-            audioBuffer,
-            {
-                headers: {
-                    Authorization: `Bearer ${HUGGINGFACE_API_TOKEN}`,
-                    'Content-Type': contentType
-                },
-                timeout: 60000
-            }
-        );
+        const contentType = req.headers['content-type'] || 'audio/webm';
+        const mimeType = contentType.split(';')[0].trim().toLowerCase();
+        const filename = getTranscriptionFilename(contentType);
+        const formData = new FormData();
 
-        const data = response.data;
-        const transcript = Array.isArray(data) ? data[0]?.text : data?.text || '';
-        res.json({ transcript });
+        formData.append('file', new Blob([audioBuffer], { type: mimeType }), filename);
+        formData.append('model', 'whisper-large-v3-turbo');
+        formData.append('response_format', 'json');
+
+        const response = await fetch('https://api.groq.com/openai/v1/audio/transcriptions', {
+            method: 'POST',
+            headers: {
+                Authorization: `Bearer ${GROQ_API_KEY}`
+            },
+            body: formData
+        });
+
+        const data = await response.json().catch(() => null);
+
+        if (!response.ok) {
+            const errorMessage = data?.error?.message || data?.error || 'Speech-to-text failed';
+            console.error('Groq speech-to-text error:', errorMessage);
+            return res.status(response.status).json({ error: errorMessage });
+        }
+
+        const transcript = data?.text || '';
+        return res.json({ transcript });
     } catch (error) {
         console.error('Speech-to-text error:', error.message || error);
-        res.status(500).json({ error: 'Speech-to-text failed' });
+        return res.status(500).json({ error: 'Speech-to-text failed' });
     }
 });