Features
Local speech-to-text via faster-whisper (voice never leaves machine)
ElevenLabs TTS integration for natural voice output
WebSocket-based low-latency audio streaming
Self-hosted on your own hardware
No installation required – works in any browser (desktop and mobile)
Connects to OpenAI, Anthropic Claude, or custom AI agents
OpenClaw gateway integration available
Sub-second response times
Privacy-focused – voice data stays on your servers
Open source under MIT license
Part of the OpenClaw ecosystem
Built with Python, FastAPI, and WebSockets
Local TTS option via Chatterbox for fully self-hosted
No subscriptions or vendor lock-in
Multilingual speech transcription (99+ languages)
To-English speech translation
Zero-shot robustness to accents, noise, technical language
Open-source models and inference code
Encoder-decoder Transformer architecture
Trained on 680,000 hours of diverse data
Log-Mel spectrogram input
30-second audio chunk processing
Multiple model sizes (tiny to large)
Whisper.cpp for CPU inference
Fine-tuning via Hugging Face integration
Turbo model on OpenAI API
OpenAI API at $0.006 per minute
