Home › Speaker diarization on Mac
Speaker diarization on Mac — who said what, offline
Lesskeys automatically labels each speaker in your calls and interviews, entirely on your Mac. No cloud diarization, no uploads — just a transcript that tells you who said what.
·
What is speaker diarization?
Speaker diarization is the process of working out who is speaking and when, then splitting a transcript into labeled turns — Speaker 1, Speaker 2, or real names. It's what turns a wall of text into a readable interview or meeting record. Lesskeys does it automatically, and does it locally.
Can you do it on a Mac without the cloud?
Yes — and that's the point. Most transcription tools that label speakers send your audio to a server to do it. Lesskeys runs diarization on-device using local models on Apple Silicon, so your recordings never leave your Mac and it all works offline. Confidential interviews stay confidential by design.
How Lesskeys labels speakers
- Automatic detection: it finds speaker changes and assigns each segment to a speaker.
- Persistent voiceprints: recurring speakers can be recognized and named across recordings, on-device.
- Fast or Precise: pick a lightweight model for speed or a higher-accuracy model for hard, overlapping audio.
- Name them: assign real names to speakers, kept locally with the transcript.
Who it's for
Journalists transcribing multi-person interviews, researchers coding focus groups, lawyers reviewing depositions, podcasters building show notes — anyone who needs to know who said what and can't (or won't) upload the audio to a cloud service.
Pay once. No subscription. · macOS 13+
FAQ
What is speaker diarization?
Automatically detecting who is speaking and when, then labeling each part of a transcript by speaker — Speaker 1, Speaker 2, or named people.
Can you do speaker diarization on a Mac without the cloud?
Yes. Lesskeys performs diarization entirely on-device, so recordings are never uploaded and it works fully offline.
Does Lesskeys remember speakers across recordings?
Yes. It can build persistent on-device voiceprints, so a recurring speaker can be recognized and named across different recordings.
How accurate is on-device diarization?
Lesskeys offers Fast and Precise models plus overlap detection, so you can trade speed for accuracy on tricky multi-speaker audio — all locally.