github epicenter-md/epicenter v7.5.3
Whispering v7.5.3: Local Transcription Compatibility Improvements

latest release: v7.5.4
3 days ago

Whispering now converts audio using pure Rust before falling back to FFmpeg, so most CPAL and VAD recordings work without FFmpeg installed.

Previously, any audio format conversion for Whisper C++ or Parakeet required FFmpeg. Now the Rust layer handles common conversions (sample rate, channels, bit depth) automatically. FFmpeg is still needed for Browser API recordings since they produce compressed formats.

Also fixed resampling bugs that caused blank transcriptions and optimized audio processing for better performance.

What's Changed

  • refactor(copyable): rename to TextPreviewDialog and move TranscribedTextDialog by @braden-w in #864
  • fix(CompressionBody): make ffmpeg install link inline by @braden-w in #874
  • feat(transcription): add specific FFmpeg not found error handling by @braden-w in #875
  • fix(dev): use separate identifier for dev builds to allow multiple instances by @braden-w in #880
  • feat(transcription): add pure Rust audio conversion fallback by @braden-w in #881

Full Changelog: v7.5.2...v7.5.3

Don't miss a new epicenter release

NewReleases is sending notifications on new releases.