A screen recording with a transcript, kept as files
Apple's Mac can record the screen (Shift-Command-5), and Notes can transcribe audio. They don't meet. Spake records both at once and keeps them together, as a note.1
What you do
Double-tap right Option. Talk about whatever is on screen. Double-tap again.
What you get
A folder in a place you choose:
note.md- The transcript, in timestamped paragraphs; editable.
metadata.json- Title, labels, paths, status.
audio.wav- What was said.
recording.mp4- What was on screen.
captions.vtt- The transcript as captions.
frames/- Stills sampled from the recording.
context.json- The text recognized in those stills.
Play the note back and each word lights up as it's said. Search finds a note by what you said or by text that was only on screen: the error message you never read aloud, the tram stop, the file name. The recognized text may contain errors; so may the screen.
Where the work happens
On the Mac. Transcription runs locally: Parakeet for 25 languages, Whisper for 75 more, with the language detected from the first 30 seconds. There's no account, and no cloud speech service. Every request Spake makes over the network is listed at spake.app/privacy.
What it isn't
- Not a share link. Spake doesn't host anything. If the recording is for someone else to watch in a browser, a hosted recorder like Loom does that, and Spake doesn't.2
- Not always on. It records from the moment you start a note to the moment you stop it.
- Not small while it works. The speech models use about 1.5 GB of memory while loaded and are released after three idle minutes. The first launch downloads about 714 MB of speech model, or about 1.3 GB when it needs both.
The files outlast the app
Every file opens without Spake: Markdown, JSON, WAV, MP4, WebVTT, JPEG. Delete Spake and the notes are still there in Finder. The format is documented at spake.app/format.
Also close to this
Try it
$29 once, up to 3 Macs. 14-day trial with everything on. 30-day refunds. Apple silicon, macOS 15 or later.
What a note is on disk: the note format. What Spake sends over the network, and when: privacy.
Sources
Facts about other products, and the day each was read.
macOS. Shift-Command-5 records the screen and makes no transcript; Notes transcribes audio and records no screen.
Loom. Plans are per user: Starter $0 (25 recordings, 5 minutes each), Business $18 and Business + AI $24 per user per month. Recordings are hosted in a Loom account. The page does not describe searching on-screen text.
Lesskeys. Records the microphone, system audio and the screen or a window as one video, with an on-device transcript and speaker labels. Sold through the Mac App Store.
Summit. Records screen video alongside audio with local transcription (versions 1.10 and 1.11). Sold through the Mac App Store.