You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Added new Quail Voice Focus STT model (QuailVfSttL16), purpose-built to isolate and elevate the foreground speaker while suppressing both interfering speech and background noise.
Added new variants of the Quail STT model: QuailSttL8, QuailSttS16 and QuailSttS8.
Added new VAD parameter VadParameter.MinimumSpeechDuration used to control for how long speech needs to be present in the audio signal before the VAD considers it speech.
Breaking changes
QuailXS was renamed to QuailXs
QuailXXS was renamed to QuailXxs
QuailSTT was replaced with specific STT model variants: QuailSttL16, QuailSttL8, QuailSttS16, QuailSttS8
Replaced VAD parameter VadParameter.LookbackBufferSize with VadParameter.SpeechHoldDuration, used to control for how long the VAD continues to detect speech after the audio signal no longer contains speech.
Fixes
VAD now works correctly when EnhancementParameter.EnhancementLevel is set to 0 or EnhancementParameter.Bypass is enabled (previously non-functional in these cases)