Choose or drop an M4A recording, transcribe it locally, title it in the detected language, and save a timestamped text document beside the source.
- Select or drop an audio file and detect its spoken language.
- Transcribe it locally with sentence timestamps.
- Generate a title from the transcript and save a named text document.
- Native SwiftUI macOS interface.
- Desert Ant Ear, Voz, and Title models through the local
desertantruntime. - Model weights download once and stay in Desert Ant's managed cache.
- Audio and transcript stay on the Mac.
- No recording, batch queue, speaker diarization, or cloud fallback.
- Align is intentionally unused: Voz already supplies word timestamps.
- The Desert Ant command-line runtime is installed separately with Homebrew.