Add voice chat with real-time speech-to-text and model responses

[devin-ai-integration]

Summary

Adds a new Voice Chat tab that enables voice-driven conversations with AI models. Users tap a microphone button to record speech, which is transcribed in real-time via OpenAI Whisper and sent to the selected AI model (Claude, GPT, or Gemini). Model responses stream back and are read aloud via text-to-speech.

What's included

App (React Native / Expo)

• New VoiceChat screen (app/src/screens/voice.tsx) with:
  • Microphone recording via expo-av
  • Audio upload to server for transcription
  • Streaming AI responses via existing SSE infrastructure
  • Text-to-speech playback of responses via expo-speech
  • Animated pulsing mic button with state indicators (idle → recording → transcribing → responding)
  • Conversation history with chat bubbles and markdown rendering
  • Clear conversation and stop speaking controls
• New "Voice" tab in bottom navigation with mic icon
• Dependencies: expo-av, expo-speech

Server (Node.js / Express)

• New /chat/transcribe endpoint (server/src/chat/transcribe.ts) that:
  • Accepts audio file uploads via multer
  • Forwards to OpenAI Whisper API for transcription
  • Returns transcribed text
  • Cleans up temporary files after processing

Works with all existing chat providers — respects the user's model selection from Settings.

Review & Testing Checklist for Human

• Test voice recording and transcription on iOS and Android (requires microphone permissions)
• Verify OPENAI_API_KEY is set in server .env for Whisper transcription
• Test with different AI models (Claude, GPT, Gemini) to confirm SSE streaming works in voice mode
• Verify text-to-speech reads responses aloud and can be stopped mid-speech

Notes

• The transcription uses OpenAI's Whisper API (whisper-1 model), which requires the OPENAI_API_KEY already used by GPT chat
• Audio is recorded in high-quality m4a format via expo-av
• The voice screen maintains conversation history per session with Claude prompt context tracking
• TTS strips markdown formatting before speaking for cleaner audio output

Link to Devin session: https://app.devin.ai/sessions/e8849f2537fb43108fa7294521fe0ae7
Requested by: @dabit3

---

Devin Review

Status	Commit
⚪ Not started	—

💡 Connect your GitHub account to enable automatic code reviews.

---

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring