VocalDocs

VadSettings

Voice activity detection.

An object with 4 fields.

Fields

min_silence_msinteger
Silence this long after speech is a speech end (t0); it bounds the endpoint stage from below.
min_speech_msinteger
Speech shorter than this is not a speech start.
prefix_padding_msinteger
Audio kept from before a detected speech start, so short first words survive.
thresholdnumber
The voice activity probability above which a frame counts as speech.