Mac
Universal DMG · Intel and Apple Silicon · macOS 12 or later. Ad-hoc signed; not notarized.
INSTANTTRANSCRIBER DESKTOP · PUBLIC BETA
Say it. Put it where you work.
Hold a shortcut, speak, then release. Explain a task to your coding agent, draft a long terminal prompt or write an email, with a copy mode for computers that restrict automatic typing.
An InstantTranscriber account and internet connection are required. Audio is transcribed by our online service after you release the shortcut; this is not offline or live word-by-word recognition.
Ready to copy···
Please add a short explanation of this change and the steps to verify it.
Your text stays here until you copy or close it.
Choose your desktop
Version 0.1.3 · Direct downloads · No separate desktop purchase
Windows
64-bit EXE installer. Not publisher-signed; SmartScreen or company policy may block it.
Linux
Built on Ubuntu 22.04. DEB needs the listed system libraries; AppImage compatibility depends on your distribution. Wayland uses manual paste and a supported shortcut portal.
For the context that takes longer to type
Coding agents and terminal prompts
Speak the goal, constraints and background for your agent. Review the result, then submit it yourself. The app never presses Enter for you.
Code reviews and issue trackers
Draft a review comment, explain a bug or describe reproduction steps while the details are fresh. Edit exact identifiers and code syntax before sending.
Emails, notes and documents
Turn a thought into a paragraph in your existing tools. Choose typing, automatic clipboard output or manual copying.
Inspect the failing test and explain its cause. Propose a small fix, keep the public interface unchanged, and run the relevant checks before summarizing the result.
Three ways to use your words
Type into the active app
Insert text into the original text field. If the destination changes or blocks typing, the result stays available in the app. Return is never pressed automatically.
Copy automatically
Put each result on the clipboard, then paste it yourself. This intentionally replaces the clipboard when a result arrives.
Show result · copy manually
Keep the result window visible without typing or changing the clipboard. Click Copy text to copy and close it, or Close to dismiss it. A failed copy keeps your text visible.
“Insert and copy” is also available when you want both automatic actions.
Setup and permissions
Install and connect
Open the app and choose Connect & get started. Finish sign-in or signup in your browser, then return to the app. Before updating, quit the old background app and replace it with the new version.
Allow the microphone
Use device setup before your first dictation. On Mac, explicitly allow microphone access. Windows and Linux offer a local microphone test that discards the audio without uploading it.
Choose how text is delivered
Automatic typing on Mac additionally needs Accessibility access. Clipboard and manual copy modes do not. Permissions are requested from setup rather than repeatedly after each dictation.
Hold, speak, release
The default shortcut is Command+Shift+Space on Mac and Ctrl+Alt+Space on Windows/Linux. Customize it or choose toggle mode in preferences. Start/stop sounds and login startup are optional.
Included with your account
Free accounts include 1 audio hour per UTC calendar month. Premium includes 15 hours per billing month, shared across desktop, API and MCP. The API Plan includes 100 hours. These are separate from web-transcription limits.
Dictations can be up to 2 minutes. Successful results use actual audio duration rounded up to a second; failed, cancelled and no-speech jobs use no audio time. Desktop does not use paid API overage. Beta capacity can be busy; try again if admission is temporarily unavailable.
Server-side transcription. No model setup on your laptop.
Internet is required because recognition runs on our servers. You do not need to download model weights, configure a local AI runtime or own a GPU. Recognition can be updated centrally without making you manage models. Accuracy still depends on language, microphone and recording conditions; review names, identifiers and commands before using the text.
A microphone that waits for you
The microphone opens only when you start dictation or explicitly test it. There is no local recording archive. The last result stays in app memory until cleared or the app quits. Service audio is removed after processing or expiry; temporary result recovery lasts up to 15 minutes.
Account-bound technical diagnostics include OS, architecture, app version, permission outcomes, errors and timings. Diagnostic events are retained for up to 90 days and contain no speech text, audio, clipboard content or other app names.
Desktop data and privacyBefore you install
Is this the online file transcriber?
Desktop is for short dictations into your other apps. For longer audio/video recordings, speaker labels, timestamps and file exports, keep using the online transcriber. Guest file transcription still works without signing up.
Does it need computer-control permission?
Only automatic typing on Mac needs Accessibility. Manual copy and clipboard modes avoid text injection. Microphone access is still required. The app cannot override an employer’s device policy.
Does it work offline?
No. The desktop app sends your recording to the InstantTranscriber service and returns a final result after release. It requires an account and an internet connection.
Can I install it from an app store?
This beta uses direct downloads. No store purchase is required. Verified Mac signing/notarization and Windows publisher signing are still pending.
What happens after copying?
The small window closes after a successful manual copy. The last text remains available in the main app. Close dismisses it without copying; Clear removes it from the app. Copy or close a pending manual result before the next dictation.