Downloads ~200MB model for local processing
WebGPU is faster but requires compatible browser
Voice uses the Parakeet TDT 0.6B model for speech recognition. All audio is processed locally in your browser — nothing is sent to any server.
Processing