Meeting notes,by speaker, summarised.All on your Mac.

Only if your microphone can hear it. Memo records the Mac's default microphone and nothing else — not system audio, not a particular app. For a Zoom, Meet or Teams call, use the speakers rather than headphones so the microphone picks up both sides. With headphones on, Memo records only you.

That's also why no bot joins the call: there's no integration with the meeting service at all.

A few minutes on an M-series Mac, most of it transcription; speaker separation takes about 45 seconds of it. It runs in the background, and each step is saved before the next starts, so quitting halfway loses nothing — Memo carries on next time you open it.

Not from inside the app: there's no export, share or copy button. Instead every recording is a folder of plain files — transcript.md and summary.md are Markdown with the timecodes kept in. Choose Show in Finder on a recording and open, copy, email or paste them wherever you like.

By title: a search field appears once you have more than four recordings, and titles are written from the summary automatically (you can edit them). Memo doesn't search inside transcripts; because they're ordinary Markdown files, your own tools and Spotlight can.

Treat it as a draft — the app labels it one. A model small enough to run on a laptop summarises less well than a large one in a data centre. What makes it checkable: each point links to the transcript line it came from, so checking is a click, and the same transcript always gives the same summary. There's no regenerate button or custom prompt.

Summaries are written in the recording's language or always in English, as you choose; English is usually the better summary.

Usually. Clearly different voices separate reliably; two very similar ones can be merged into a single speaker. Renaming fixes a label, not a merge — so in a group where several people sound alike, read the transcript rather than trusting the count.

Fifteen, plus Automatic (the default): English, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Serbian, Ukrainian, Russian, Chinese, Japanese, Korean and Hindi. The language is fixed when a recording starts. The Small model is English only; Medium and Large handle all fifteen. The app's own interface comes in the same fifteen languages; it follows your Mac's language, or pick one in Settings.

Four things, none of them your audio or your text:

• model downloads from Hugging Face, on first use;
• licence activation and validation — only once you've entered a key;
• one anonymous install count, sent once: app, version, macOS version and chip type, with no identifier;
• automatic update checks, for Memo and for Altypist's feed.

There's no analytics SDK, no usage tracking and no account.

Recording and importing stop until you enter a licence. Everything you recorded stays readable — transcripts, summaries and playback are never locked, with or without a licence.

Apple Silicon (M1 or later) with macOS 15 or later; Intel Macs aren't supported. The app is about 25 MB and the speech model you choose is 600 MB to 3 GB, downloaded on first use. It asks for the microphone and nothing else — no Accessibility permission, no calendar.

Only if you want to keep conversations. Altypist is for dictation: you speak, the words appear at your cursor, and nothing is kept. Memo records the whole conversation and keeps the transcript, the speakers and a timecoded draft summary in files you own. Owning Altypist doesn't include Memo; if you already have one, the other is $30 $10.00.

Start with the free 7 days — every feature, no card, no account. If you've already bought, you have 14 days to ask for a full refund, no reason required. Refund policy.