Sobes.tech

Download Sobes Copilot

Choose the version for your operating system

Changelog

All updates and improvements to Sobes Copilot in chronological order.

v2.9.0

Context and audio transcription settings:

  • Speech recognition settings have been removed from the context. They are now separate parameters.
  • Added speech recognition model selection (in the client and on the website), available options are Whisper Large V3 Turbo (0.3-0.5s per transcription), GPT-4o Mini Transcribe (0.5-0.8s per transcription), and Deepgram Nova 3 (0.5-1s per transcription).

Audio recording:

  • Noise reduction has been removed from the client, as it caused more problems than it solved.
  • Changed the audio recording algorithm in VAD mode. When audio separation is enabled, the recording is divided after the end of a word/phrase, rather than strictly by timer - breaks in the middle of words will occur much less frequently.
v2.8.15

Quick Actions:

  • You can now specify which model will be used when performing quick actions.

Window:

  • The position of the Overlay window is now saved when you exit the application. Quick Actions:
  • You can now specify which model will be used when performing quick actions.

Window:

  • The position of the Overlay window is now saved when you exit the application.