Open Whisper Docs

Appendix A5 What Open Whisper doesn't do

Reading time · 3 min

The things Open Whisper leaves out on purpose, and the concrete reason behind each one.

Open Whisper does one thing: it writes down what you say, where your cursor is. Everything below was considered and left out on purpose. The reason matters more than the item.

Transcription stays on your computer

  • No cloud transcription services. Sending your voice to an online service would break the promise the app is built on: your audio never leaves the device.
  • No AI cleanup, rewriting, chat or read-aloud. Open Whisper writes down what you said. Deciding what you meant is a different product.
  • No translation or summaries. Same reason: the text is what you said, in the language you said it.

Other people's voices

  • No meeting recording. Recording other people raises consent questions that a dictation tool shouldn't answer for you.
  • No speaker identification. Telling who said what belongs to meeting transcription, which is out for the same reason.

Files and videos

Transcribing one video link or one file is a secondary feature, kept small on purpose.

  • One at a time. No folders, playlists, channels or watched folders. The app is about the moment you're typing, not about processing a library. From the command line, you can loop over files in your own script.
  • YouTube only. Links from other sites are refused before anything runs.
  • No copy of the media. A file is read where it is and never copied. A video's audio is held in memory and discarded.
  • No pasting a video transcript at your cursor. It shows in its own window, for you to copy or save.
  • Some formats aren't read. Opus, AC-3, E-AC-3, DTS, AVI, WMV and ASF would need a second outside tool. The app names the format it can't read and tells you how to convert it.
  • No subtitle tracks from inside a file, no choice of audio track, no live output while it works.
  • No dictation from the command line. Dictation is the shortcut.

Your data

  • No history. The last dictation is kept in memory until you quit, so you can copy or paste it again. Nothing else is stored.
  • No account and no cloud sync. There is nothing to sync, because nothing leaves your computer.

Platforms and stores

  • No Mac App Store or Microsoft Store. To paste into other apps, Open Whisper sends the paste keystroke for you. The stores' sandboxes don't allow that, so a store version couldn't do the one thing the app exists to do.
  • On Windows, a download only. The installer comes from the Releases page, with no package-manager listing for now.
  • No Linux or phone version yet. They are on the roadmap. No date is promised.
  • On Windows, no native ARM build and no graphics-card acceleration yet. Windows on ARM runs the app emulated, and transcription runs on the processor.

A simple interface

One icon (in the menu bar on the Mac, in the notification area on Windows) and a small Settings window. There is no main window, and the only thing you need to set up is the shortcut.

An edge case is weighed, not refused on sight

If your idea sits close to one of these lines, suggest it anyway. An idea that needs a decision is marked for review and answered with a reason, not closed on sight. See Reporting a problem.