Skip to content

Transcripted 1.1.72

Choose a tag to compare

@r3dbars r3dbars released this 10 Oct 05:37

What changed

  • transcripted-cli now uses the same Nemotron speaker engine and windowing as the app for import-audio, diarize and batch. Use --diarization-engine pyannote to opt back in.
  • --name-likely-speakers labels close voice matches as "Name (likely)" without marking them confirmed.
  • Undoing a speaker merge keeps recordings you moved to or from that person after the merge.
  • Failed-queue cleanup keeps offline rows from a relocated library while their deletion is still pending.
  • Dictation Undo keeps hand-edited whitespace after a newer save and no longer adds blank lines to day files.

Requires macOS 26 or later and Apple Silicon.

Full changelog