Silent Voice

Silent Voice is an assistive communication system that speaks silently mouthed words in a cloned voice.

Silent Voice gives a voice back to people who lost their voice box. You mouth the words without a sound, and Silent Voice speaks them in your own voice while you are still mouthing. The system combines live lip reading via MediaPipe and Auto-AVSR, facial expression emotion detection, an EMG model reading muscle signals from BioAmp EXG Pill boards and an Arduino Uno, and Claude for context and repairs. Cartesia Sonic speaks each word in your cloned voice as soon as it locks, and Twilio handles phone calls. The project is open source under the MIT license.

Silent Voice at a glance

Best for: Support
Runs on: Web, API, self-hosted, open source
Pricing: Free
Released: November 2026
Category: AI audio tool
Website: github.com/thaimtl/silent-voice

Key features

Real-time silent speech translation
Live lip reading and facial emotion detection
EMG muscle signal reading via Arduino
Claude context prediction and word repairs
Cartesia voice cloning and speech synthesis
Twilio phone call integration
Live dashboard for latency and stats

Use it for

Speaking during phone calls without a voice box
Translating silent mouthing into synthesized speech
Communicating in real time using facial muscle signals

Alternatives to Silent Voice

Bland AI — Runs entirely on your own infrastructure with sub-second latency and enterprise compliance.