Skip to main content

Takes

A writing workspace built around speaking.

Say a thought out loud and Takes keeps it: the words, and the recording they came from. Gather the takes you want into a manuscript, arrange them, and come back tomorrow to where you left off.

Not available to download yet.

The first Mac preview is being prepared. It will run on macOS 13 Ventura or later, on a Mac with Apple silicon (M1 or later).

The Takes window. On the left, a column of spoken takes with one selected; on the right, the manuscript with a chapter heading, a section heading and paragraphs arranged from the takes.
Your takes on the left, as you said them; the manuscript on the right, as you arrange it. (Example writing.)

How it works

  1. Speak

    Choose Record, or press Space, and say the thought as it comes. Takes shows “Starting microphone…” until it is really recording, then a running clock and a level meter. The words appear a moment after each pause, and the recording is saved as you go, about once a second.

  2. Keep your takes

    Every take is kept: the words as they were recognised, each correction you type as a new version, and the original recording. When recognition gets a word wrong, play the take back and hear what you said. Nothing is thrown away for being quiet, short or unrecognised.

  3. Arrange a manuscript

    Send the takes you want into the manuscript beside them, put them in order, and add chapter and section headings. Change the wording in the manuscript without touching the take it came from. Export the manuscript as plain Markdown.

  4. Return to your work

    Takes opens where you left off: the last project, ready to write. Each book, essay or article is its own project in the Library, with a history of earlier versions you can go back to.

Takes while recording: a red Stop button, “Recording 0:03” with a level meter in the top bar, and the new take appearing at the bottom of the takes column.
While you speak: Stop, the capture clock and the input level — taken from the recording itself. (Example writing.)

Private by design

Your recordings and your writing stay on your Mac. The speech recognition runs on your Mac too, so what you say is never sent anywhere to be transcribed. There is no account and no analytics in the app.

Takes connects to the internet only to check for and download updates, and it asks before checking automatically. What goes where, in detail.

The Takes Library: two projects, “Tide Tables” (open, edited today, ten takes, five paragraphs in the manuscript) and “Letters from the Island”, each with Open, Rename and Archive.
The Library: each book, essay or article is its own project. (Example writing.)

What to expect

  • Transcription is recognition, not a guarantee. Names, quiet speech and noisy rooms produce mistakes; that is why the recording is always kept.
  • It is not hands-free. You speak to write, and use the keyboard and pointer to correct, arrange and export.
  • English only, for now.
  • Nothing said before recording has started can be captured. After Record, wait for the clock.
  • Takes saves as it goes, but no software can promise zero loss: a sudden power cut can lose the last few seconds. Keep your own backups, such as Time Machine.

Supported systems

macOS 13 Ventura or later, on a Mac with Apple silicon (M1 or later). Intel Macs, Windows and Linux are not supported. About 600 MB of disk space for the app and its speech model, plus room for your recordings (about 4 MB per minute of speaking: the original recording, and the copy used for transcription).

See the release notes, get support, or write to ian@vaultscaler.com.