Starstruk Logo MarkStarstruk

v0.2.2

KoljaB/RealtimeSTTv0.2.2Aug 7, 2024by KoljaB

AI Summary

Added a new parameter to utilize the Silero model for end-of-speech detection, improving accuracy in noisy environments.

Key Highlights

  • Enabled Silero model for end-of-speech detection
  • Improved noise robustness in background environments
  • Utilizes GPU resources for enhanced accuracy

New Features

  • silero_deactivity_detection

Full Release Notes

- new parameter silero_deactivity_detection (bool, default=False)
  Enables the Silero model for end-of-speech detection. More robust against background noise. Utilizes additional GPU resources but improves accuracy in noisy environments. When False, uses the default WebRTC VAD, which is more sensitive and may continue recording longer due to background sounds.