Sherpa-ONNX

Non-Streaming Speech Recognition

Uses VAD to split long audio into speech segments, then recognizes each segment with an offline ASR model.

0%

Recognition Results

Start	End	Text

VAD & Recognizer Settings

VAD Threshold (0.0 – 1.0) Min Silence Duration (s) Min Speech Duration (s) Max Speech Duration (s) Recognizer Threads (1 – 16)