v1.6.9

ictnlp/LLaMA-Omniv1.6.9Aug 18, 2026by altic-dev

Upgrading?

AI Summary

This update focuses on improving transcription accuracy and microphone handling. It resolves issues where AI responses lose text, speaker labels drop, and microphones are misidentified.

Key Highlights

  • Spoken Formatting in Custom Dictionary now supports newlines, paragraphs, and symbols via configurable start words
  • Microphone alerts are no longer shown on startup and wired headsets work correctly in clamshell mode
  • Fixed AI response text loss when models emit extra closing tags
  • Bluetooth microphones remain stable during AirPods route changes

New Features

  • Spoken Formatting in Custom Dictionary can insert new lines, paragraphs, tabs, spaces, punctuation, and symbols using a configurable start word

Full Release Notes

# FluidVoice v1.6.9

## Changes

- Spoken Formatting in Custom Dictionary can insert new lines, paragraphs, tabs, spaces, punctuation, and symbols using a configurable start word
- Microphone alerts no longer appear on routine startup and can be disabled from the alert or Notifications settings.
- AI responses no longer lose answer text when a model emits an extra closing thinking tag.
- Wired headset microphones remain available in clamshell mode instead of being mistaken for the unavailable built-in microphone.
- Fixed External microphones not working when connected through 3.5mm 
- Bluetooth microphones now ride through brief AirPods route and format changes without being marked unavailable
- Speaker-labeled file transcription no longer drops trailing audio when same-speaker segments overlap.
- Speaker labels are preserved when only brief transcription gaps occur, with omitted audio clearly noted.

## Contributors

Thanks to @YuriNachos and @nsutcliffe for fixes and improvements

Related Repositories