Skip to content
Docs
Download

Install and configure Whisperstream

Install Whisperstream on Windows or the Mac beta, complete first-run setup, and make your first dictation.

Applies to
Whisperstream 2.0.0
Last reviewed

Install Whisperstream

Whisperstream supports Windows 10 and Windows 11 on x64 computers. A Mac version is available as a beta for Macs with Apple silicon running macOS 14 Sonoma or later. Download the free official installer from Whisperstream.

Windows

  1. Download and open the Whisperstream installer.
  2. Review the license agreement and select I Agree to continue.
  3. Choose whether to install for your Windows account or make the app available across Windows accounts.
  4. Approve elevation if you selected an all-users installation.
  5. Launch Whisperstream when setup finishes.

Choose Install just for me for your Windows account. Choose Install for anyone using this computer to make the app available to other Windows accounts when you can approve the administrator prompt. Each user still accepts the EULA.

If you are switching from an earlier installation, back up your data and keep its installer before installing the open-source edition. The switch requires running the new installer yourself.

Mac (beta)

  1. Download and open the Whisperstream disk image (DMG).
  2. Drag Whisperstream to your Applications folder.
  3. Open Whisperstream from Applications.
  4. Review the license agreement and select I agree to continue.

Intel Macs are not supported.

Build and run from source

Start with the source README for requirements, exact dependencies and the contributor launcher. Run the native Tauri shell for dictation into other applications. Browser mode is for UI development and does not prove native text insertion or window behavior. The source uses the public sherpa-onnx and sherpa-onnx-core 1.13.8 packages. No private package archive or official signing credentials are required.

The launcher keeps source development data separate from existing installations. Follow the selected revision's setup and update instructions.

Complete first-run setup

On first launch, Whisperstream asks you to accept the EULA before the main setup opens. The guided setup then helps you:

  1. Choose a dictation language.
  2. On a Mac, allow Microphone and Accessibility access. See Allow access on a Mac.
  3. Confirm or change the Push to Talk shortcut while the matching speech model downloads in the background.
  4. Try dictation in the setup text box after the model is ready.
  5. Optionally, turn on AI enhancement with Stream-1 Mini, Whisperstream's on-device cleanup model.

Follow the progress shown in setup. Dictation becomes available when the model is ready.

The enhancement step is optional. Choose Skip & finish to leave enhancement off and download nothing. Choose Turn on enhancement to download Stream-1 Mini (about 460 MB) and check how fast it runs on this computer:

  • If it runs well, local enhancement turns on and you can try two examples before finishing.
  • If it runs slowly, choose Use it anyway or Leave enhancement off.
  • If it is too slow, enhancement stays off.

You can change this later in Enhancement; see Enhancement providers. To prepare an offline installation or restore model files manually, see Install and recover models offline.

Allow access on a Mac

Whisperstream needs two macOS permissions:

PermissionWhy it is needed
MicrophoneRecords your voice for dictation.
AccessibilityLets your shortcut work in other apps and types text into them.

In setup, select Allow microphone and Allow accessibility, then respond to the macOS prompts. If access was denied earlier, select Open System Settings, go to Privacy & Security, open Microphone or Accessibility, turn on Whisperstream, then return and select Check again. If Accessibility is on but Whisperstream still asks for it, select Restart app.

Without Accessibility, your shortcuts and text insertion do not work. To check access later, open Home and find Permissions.

Make your first dictation

The default Push to Talk shortcut is Right Shift.

  1. Use the test step in setup, or finish setup and click a text field in another app.
  2. Hold your Push to Talk shortcut (Right Shift if you did not change it).
  3. Speak one short sentence in the language you selected.
  4. Release the shortcut to stop recording.
  5. Wait for Whisperstream to transcribe the sentence. The text appears in the setup box or the selected app.

If the setup test works, installation is complete.

Choose your next guide

Use the documentation navigation for every other feature.