RapidSpeech — WebAssembly Demo

Model

Model URL

Threads Use LLM

Enter a model URL and click Load.

VAD (optional — silero-vad or firered-vad)

VAD model URL

Threshold 0.50 Min seg (s)

No VAD loaded — full clip will be transcribed.

Input

Source

Audio file

Transcript

Model

Model URL

Threads Use LLM Two-pass

Enter a model URL and click Load.

VAD (neural — silero-vad or firered-vad; falls back to energy gate)

VAD model URL

Energy gate (default).

Speech threshold 0.500

Silence frames (energy mode only) 15

Microphone

silence

Transcript

Model

Model URL

Threads

Enter a model URL and click Load.

Generation params

Instruct Language

Seed Diffusion steps 32

Voice cloning (optional)

Reference WAV

Reference text

Synthesize

Text