Live voice agent — speak into your microphone and hear the agent respond.
Requires ffmpeg and ffplay installed. Uses Deepgram STT, OpenAI LLM, and ElevenLabs TTS.
Environment variables (or .env file)
OPENAI_API_KEY=...
DEEPGRAM_API_KEY=...
ELEVENLABS_API_KEY=...
ELEVENLABS_VOICE_ID=...Usage
mix feline.talk
mix feline.talk --system "You are a pirate. Respond in pirate speak."