Projects / Amical

Amical

Desktop & Productivity

Dictate into the focused desktop application using push-to-talk or a floating widget, with local Whisper or optional cloud transcription.

Use with conditions
Last updatedRelease 2026-09-25
Open source licenseMIT
PlatformsmacOS (local Whisper: 15+) · Windows 10+ (official download page)
DeploymentDesktop

About this project

Amical turns speech into text in the desktop application that currently has focus. A push-to-talk key or floating widget starts dictation; a hands-free shortcut supports longer passages. Its desktop documentation covers macOS and Windows, and the source is published under MIT.

Local mode downloads a Whisper speech model onto the computer and can work offline after setup without an account. Model sizes and processing speed vary; the settings show download size and let users change the default model. Optional local language processing uses a separate provider such as Ollama.

Audio handling depends on the selected mode. Cloud transcription streams audio to Amical and requires sign-in. Anonymous telemetry starts enabled; users can switch it off in Advanced settings. The telemetry documentation distinguishes ordinary usage events from audio and transcripts intentionally submitted through the Report action.

Decision guide
01

Why it is worth a look

  • Ready-made macOS and Windows desktop downloads
  • Local speech models support offline dictation after setup
  • Places dictated words in the application being used
02

Before you use it

  • Grant microphone access; macOS also requires accessibility permission to paste into other apps
  • Local speech needs a model download and suitable hardware; the repository states macOS 15+ for local Whisper
  • Anonymous telemetry is enabled by default and can be disabled in Advanced settings
  • Cloud mode sends audio to a hosted endpoint and needs sign-in; do not describe all modes as fully local
  • The current desktop repository does not support Linux; meeting transcription and MCP features remain planned
Manual setupModerate
With an AI agentModerate
DeploymentDesktop
Additional requirementsNot verified

A supported macOS or Windows computer; Microphone access and macOS accessibility permission; Disk space and internet access for the initial local model download; Some requirements have not been verified. Check the official setup instructions.

03

What it can do

  • Push-to-talk dictation into the focused application
  • Hands-free dictation through a shortcut or floating widget
  • Downloadable local Whisper speech models
  • Optional hosted transcription after account sign-in

Who it fits

Good for

Writing emails and messages by voice; People who want a local speech-transcription option

Not a fit

Linux desktop dictation; Assuming all advertised roadmap features are already available

Getting started

  1. Download the official desktop stable installer for the computer architecture
  2. Install and launch Amical, then grant microphone and required OS permissions
  3. Choose Local during onboarding and download a Whisper model that fits the computer
  4. Review Anonymous Telemetry in Advanced settings and choose whether to enable it
  5. Set a dictation hotkey, focus a text field and start a short test dictation