Which Setup Fits You?

    Most "which model do I pick" advice compares hardware specs. Useful, but it skips what actually matters: what you do in meetings. This section starts from the scenario instead.

    Five minutes here will save you an hour of trial and error.

    Quick picks

    You…Start withWhy
    Take calls with remote people (Zoom, Meet, Teams)Large-v3-turbo + system audio onYou need both voices, not just yours
    Interview people and quote them laterLarge-v3 or large-v3-turboNames and terms matter; accuracy wins over speed
    Run ten short standups a weekParakeetFastest results by a wide margin
    Record lectures or long solo talksParakeet or large-v3-turboLong recordings punish slow models
    Have a quiet-voiced colleagueVAD Sensitivity (Pro)Tune what the speech detector counts as speech
    Speak in a meeting room with open speakersEcho Suppression (Pro)Stops the same words being transcribed twice
    Recorded something outside ClearminutesAudio Import (Pro)Feed the file in, retranscribe, done

    Each row has a worked example in Setups by Scenario.

    The short version of engine choice

    Three transcription engines cover everything:

    • Parakeet — the speed option on Apple Silicon. English meetings. A 15-minute standup comes back in about two minutes. It's a ~670 MB download in the app's model manager, same one-click flow as the Whisper models.
    • Large-v3-turbo — the best all-rounder if you want Whisper accuracy without the wait. Good on Apple Silicon and on Windows machines with an NVIDIA GPU.
    • Large-v3 — for when every word matters: legal, medical, board meetings. Slowest, biggest download, best accuracy.

    On a Windows laptop without an NVIDIA GPU, local models get slow and small fast. If that's your fleet, Pro's cloud transcription (Deepgram, Groq) is the honest answer: no GPU, no download, fast results.

    If you can't decide: Parakeet for everyday Apple Silicon meetings, large-v3-turbo for the ones where every word ends up in a report. Switch any time in Settings → Transcription → Model; nothing you already recorded is affected.

    Free vs Pro, in one paragraph

    Transcription is unlimited on free, and the recording always saves. Pro removes the 5-summaries-a-month cap, adds auto-summarize, cloud AI models, custom prompts and templates, speaker labels (diarization), VAD Sensitivity, Echo Suppression, and Audio Import with retranscription. If your meetings are single-voice in a quiet room and you rarely summarize, free covers you. If you record calls, quote people, or summarize often, Pro earns its keep. See Pro features.

    Where to go next