Add LiteLLM STT and TTS support

- TRANSCRIBE_PROVIDER=litellm routes audio to LiteLLM /audio/transcriptions
- TTS_PROVIDER=litellm routes to LiteLLM /audio/speech
- Both auto-detect when LITELLM_API_BASE is set (no extra config needed)
- LITELLM_STT_MODEL (default: whisper-1), LITELLM_TTS_MODEL (default: tts-1)
- LITELLM_TTS_VOICE (default: alloy) — alloy/echo/fable/onyx/nova/shimmer
- ElevenLabs still works if ELEVENLABS_API_KEY is set and TTS_PROVIDER=elevenlabs
- Health endpoint now reports tts provider

.env.example

@@ -32,6 +32,15 @@ OPENROUTER_API_KEY=sk-or-v1-your-key
 # LITELLM_API_BASE=http://localhost:4000
 # LITELLM_API_KEY=sk-litellm-your-key
 # Admin can discover available models via the admin panel
+#
+# LiteLLM Speech-to-Text (auto-enabled when LITELLM_API_BASE is set)
+# TRANSCRIBE_PROVIDER=litellm
+# LITELLM_STT_MODEL=whisper-1          # or: groq/whisper-large-v3, openai/whisper-1
+#
+# LiteLLM Text-to-Speech (auto-enabled when LITELLM_API_BASE is set)
+# TTS_PROVIDER=litellm
+# LITELLM_TTS_MODEL=tts-1              # or: tts-1-hd, openai/tts-1
+# LITELLM_TTS_VOICE=alloy              # alloy, echo, fable, onyx, nova, shimmer
 
 # ============================================================
 # TRANSCRIPTION (speech-to-text)

server.js

@@ -148,7 +148,8 @@ app.get('/api/health', (req, res) => {
     azure: process.env.AZURE_OPENAI_ENDPOINT ? 'configured' : 'not configured',
     vertex: process.env.GOOGLE_VERTEX_PROJECT ? 'configured' : 'not configured',
     litellm: process.env.LITELLM_API_BASE ? 'configured' : 'not configured',
-    whisper: process.env.OPENAI_API_KEY ? 'configured' : 'missing'
+    whisper: process.env.OPENAI_API_KEY ? 'configured' : 'missing',
+    tts: process.env.LITELLM_API_BASE ? 'litellm' : (process.env.ELEVENLABS_API_KEY ? 'elevenlabs' : 'none')
   });
 });
 

src/routes/transcribe.js

@@ -1,7 +1,7 @@
 const express = require('express');
 const router = express.Router();
 const multer = require('multer');
-const { whisperClient } = require('../utils/ai');
+const { whisperClient, litellmClient } = require('../utils/ai');
 const { transcribeWithAWS, isAWSTranscribeConfigured } = require('../utils/transcribeAWS');
 const { transcribeWithLocal, isLocalWhisperConfigured } = require('../utils/transcribeLocal');
 const { authMiddleware } = require('../middleware/auth');
@@ -9,24 +9,29 @@ const { authMiddleware } = require('../middleware/auth');
 const upload = multer({ storage: multer.memoryStorage(), limits: { fileSize: 25 * 1024 * 1024 } });
 
 // Determine which provider to use:
-// TRANSCRIBE_PROVIDER=aws  → always use Amazon Transcribe
-// TRANSCRIBE_PROVIDER=openai → always use OpenAI Whisper
-// (default) if AWS Bedrock region is set → use AWS Transcribe
-//           otherwise → use OpenAI Whisper
+// TRANSCRIBE_PROVIDER=litellm → LiteLLM proxy (any Whisper-compatible model)
+// TRANSCRIBE_PROVIDER=aws     → Amazon Transcribe
+// TRANSCRIBE_PROVIDER=local   → local whisper.cpp / faster-whisper
+// TRANSCRIBE_PROVIDER=openai  → OpenAI Whisper directly
+// (auto) LITELLM_API_BASE set → litellm
+// (auto) AWS Bedrock region set → aws
+// (default) → openai whisper
 function getTranscribeProvider() {
   var env = process.env.TRANSCRIBE_PROVIDER;
-  if (env === 'local') return 'local';
-  if (env === 'aws') return 'aws';
-  if (env === 'openai') return 'openai';
-  // Auto-detect: prefer AWS when Bedrock is already configured
+  if (env === 'litellm') return 'litellm';
+  if (env === 'local')   return 'local';
+  if (env === 'aws')     return 'aws';
+  if (env === 'openai')  return 'openai';
+  // Auto-detect
+  if (process.env.LITELLM_API_BASE && litellmClient) return 'litellm';
   if (isAWSTranscribeConfigured()) return 'aws';
   return 'openai';
 }
 
-// Check if any transcription provider is actually available
 function isTranscribeAvailable() {
   if (isLocalWhisperConfigured()) return true;
   if (isAWSTranscribeConfigured()) return true;
+  if (litellmClient) return true;
   if (whisperClient) return true;
   return false;
 }
@@ -63,7 +68,23 @@ router.post('/transcribe', authMiddleware, upload.single('audio'), async (req, r
       return res.json({ success: true, text: text, provider: 'aws-transcribe', duration: Date.now() - startTime });
     }
 
-    // OpenAI Whisper
+    if (provider === 'litellm') {
+      if (!litellmClient) return res.status(400).json({ error: 'LiteLLM not configured. Set LITELLM_API_BASE.' });
+      var sttModel = process.env.LITELLM_STT_MODEL || 'whisper-1';
+      var file = new File([req.file.buffer], 'audio.webm', { type: req.file.mimetype || 'audio/webm' });
+      var result = await litellmClient.audio.transcriptions.create({
+        file: file,
+        model: sttModel,
+        language: 'en',
+        response_format: 'text',
+        prompt: 'Medical patient encounter. Pediatric. Clinical terms, diagnoses, medications.'
+      });
+      var text = typeof result === 'string' ? result : result.text;
+      console.log('[Transcribe] LiteLLM (' + sttModel + ') done in ' + (Date.now() - startTime) + 'ms');
+      return res.json({ success: true, text: text, provider: 'litellm-stt', duration: Date.now() - startTime });
+    }
+
+    // OpenAI Whisper (direct)
     if (!whisperClient) return res.status(400).json({ error: 'Whisper not configured. Set OPENAI_API_KEY.' });
     var file = new File([req.file.buffer], 'audio.webm', { type: req.file.mimetype || 'audio/webm' });
     var result = await whisperClient.audio.transcriptions.create({

src/routes/tts.js

@@ -1,22 +1,61 @@
 const express = require('express');
 const router = express.Router();
 const axios = require('axios');
+const { litellmClient } = require('../utils/ai');
 const { authMiddleware } = require('../middleware/auth');
 
+// TTS_PROVIDER=litellm  → LiteLLM proxy (routes to any OpenAI-compatible TTS)
+// TTS_PROVIDER=elevenlabs → ElevenLabs (not HIPAA)
+// (auto) LITELLM_API_BASE set → litellm
+// (auto) ELEVENLABS_API_KEY set → elevenlabs
+function getTTSProvider() {
+  var env = process.env.TTS_PROVIDER;
+  if (env === 'litellm')   return 'litellm';
+  if (env === 'elevenlabs') return 'elevenlabs';
+  if (process.env.LITELLM_API_BASE && litellmClient) return 'litellm';
+  if (process.env.ELEVENLABS_API_KEY) return 'elevenlabs';
+  return 'none';
+}
+
+var ttsProvider = getTTSProvider();
+console.log('🔊 TTS provider:', ttsProvider);
+
 router.post('/text-to-speech', authMiddleware, async (req, res) => {
   try {
-    if (!process.env.ELEVENLABS_API_KEY) return res.status(400).json({ error: 'Not configured' });
-    const response = await axios({
-      method: 'POST',
-      url: 'https://api.elevenlabs.io/v1/text-to-speech/pNInz6obpgDQGcFmaJgB',
-      headers: { 'xi-api-key': process.env.ELEVENLABS_API_KEY, 'Content-Type': 'application/json' },
-      data: { text: req.body.text.substring(0, 5000), model_id: 'eleven_turbo_v2_5', voice_settings: { stability: 0.5, similarity_boost: 0.75 } },
-      responseType: 'arraybuffer'
-    });
-    res.set('Content-Type', 'audio/mpeg');
-    res.send(Buffer.from(response.data));
+    var text = (req.body.text || '').substring(0, 5000);
+    if (!text) return res.status(400).json({ error: 'No text provided' });
+
+    if (ttsProvider === 'litellm') {
+      if (!litellmClient) return res.status(400).json({ error: 'LiteLLM not configured. Set LITELLM_API_BASE.' });
+      var ttsModel = process.env.LITELLM_TTS_MODEL || 'tts-1';
+      var ttsVoice = process.env.LITELLM_TTS_VOICE || 'alloy';
+      var mp3 = await litellmClient.audio.speech.create({
+        model: ttsModel,
+        voice: ttsVoice,
+        input: text,
+        response_format: 'mp3'
+      });
+      var buffer = Buffer.from(await mp3.arrayBuffer());
+      res.set('Content-Type', 'audio/mpeg');
+      return res.send(buffer);
+    }
+
+    if (ttsProvider === 'elevenlabs') {
+      if (!process.env.ELEVENLABS_API_KEY) return res.status(400).json({ error: 'ElevenLabs not configured' });
+      const response = await axios({
+        method: 'POST',
+        url: 'https://api.elevenlabs.io/v1/text-to-speech/pNInz6obpgDQGcFmaJgB',
+        headers: { 'xi-api-key': process.env.ELEVENLABS_API_KEY, 'Content-Type': 'application/json' },
+        data: { text: text, model_id: 'eleven_turbo_v2_5', voice_settings: { stability: 0.5, similarity_boost: 0.75 } },
+        responseType: 'arraybuffer'
+      });
+      res.set('Content-Type', 'audio/mpeg');
+      return res.send(Buffer.from(response.data));
+    }
+
+    res.status(400).json({ error: 'TTS not configured. Set LITELLM_API_BASE or ELEVENLABS_API_KEY.' });
   } catch (err) {
-    res.status(500).json({ error: 'TTS failed' });
+    res.status(500).json({ error: 'TTS failed: ' + err.message });
   }
 });