Get Amical
Amical installs like any desktop app, so no terminal is needed. Download the installer for your system from the official site or the GitHub release page; the Whisper model is downloaded inside the app during first-run setup.
Platform | Get It |
|---|---|
| Mac | Download for Mac — .dmg for Apple Silicon or Intel; also brew install --cask amical |
| Windows | Download for Windows — .exe for x64 or arm64 |
| iPhone and Android | Mobile apps on the Amical site — App Store and Google Play; sign-in with an Amical account |
| Release notes | GitHub Releases and the Amical documentation |
| Website | amical.ai |
| Privacy policy | Amical privacy policy |
| Source code | github.com/amicalhq/amical — MIT license |
The privacy policy page is labeled for the mobile app; PromptQuorum found no separate desktop privacy policy.
Amical at a Glance
Attribute | Amical |
|---|---|
| Platform | macOS and Windows desktop; Linux not supported |
| Price | Free (local); paid Premium plan for cloud |
| License | MIT, open source |
| Runs | On your computer with Whisper; optional cloud when signed in |
| Languages | 100+ per the maker (not verified) |
| Speaker labels | Not stated |
| Real-time transcription | Planned for meetings; dictation is push-to-talk |
| Usable on CPU only | Not stated; model size from 78 MB to 3.1 GB |
| Local API server | Not stated |
Attributes follow the speech-to-text comparison used in the Local LLM Software Directory. A missing statement on the site means unknown, not unsupported.
What Amical Is
Amical is a push-to-talk dictation app, not a transcription editor. You hold a key or click the floating widget, speak, and the text lands in the app that has focus; the app detects that app and adjusts the formatting, for example a greeting on its own line in an email.
The code lives under the GitHub organization amicalhq. The project also lists voice notes (in progress) and real-time meeting transcription and voice commands (planned); this review covers the dictation app only.
This review draws on the website, documentation, privacy policy and the GitHub repository. It involved no hands-on testing, so accuracy, speed and stability are not rated here.
How to Get Started
Setup is an install, a model choice and a shortcut. The steps follow the desktop quick-start in the documentation.
- 1Install the app
Why it matters: Download the .dmg for Mac or the .exe for Windows from the [GitHub release page](https://github.com/amicalhq/amical/releases/latest), or run `brew install --cask amical` on a Mac. - 2Grant permissions
Why it matters: Allow microphone access; on a Mac also allow accessibility access so Amical can paste into other apps. - 3Choose Local or Cloud
Why it matters: Onboarding offers Cloud (sign-in, no setup) or Local (Whisper on your machine, offline) and recommends one based on your hardware. - 4Hold, speak, release
Why it matters: Set the push-to-talk shortcut, click into any text field, hold it, speak and let go; a press shorter than 500 ms is treated as a slip and discarded.
Features
The feature set centres on dictation into any app. Everything below comes from the README, documentation and release notes.
- Dictation anywhere. Text is pasted into the focused app via a push-to-talk shortcut or the floating widget.
- Context-aware formatting. An optional AI pass, marked Alpha, shapes output to the active app and needs a local language model or the Amical Cloud connection.
- Languages. 100+ languages per the maker, with automatic language detection; version 1.13.0 added regional variants and Cantonese.
- Personalization. Custom vocabulary, snippets, custom prompts, a transcription history and, since version 1.13.0, vocabulary learned from corrections after a paste.
- Notes. Voice notes are listed as in progress.
Models and Hardware
Amical runs Whisper locally in six variants. The documentation lists Tiny (about 78 MB), Base (about 148 MB), Small (about 488 MB), Medium (about 1.5 GB), Large v3 (about 3.1 GB) and Large v3 Turbo (about 1.5 GB), each shown with speed and accuracy ratings in Settings.
The README says local Whisper requires macOS 15 or later on a Mac. No RAM or GPU requirement is stated, so pick a model that fits your machine; for the optional formatting pass, the documentation names Ollama or any OpenAI-compatible endpoint. No speed or accuracy figures were reproduced here.
Privacy, License and Cost
The documentation says Amical can run entirely on your machine, with no data leaving the computer and no account required. Transcripts, notes, vocabulary and audio files are stored in the app-data folder. Cloud speech, cloud formatting and sign-in send the audio for that request and nothing else, per the FAQ.
Telemetry is a separate matter. The documentation says anonymous telemetry covers app launches, errors and performance, is on by default and can be switched off under Settings, Advanced; pressing the Report flag on a history entry also sends that audio and transcript. The version 1.13.0 release notes say the app now asks for telemetry consent at sign-in, and the two statements are not reconciled on the maker's pages. The mobile privacy policy lists product analytics and ad measurement. The code is open and PromptQuorum did not examine network traffic, so these statements are the maker's own.
The license is MIT. Local dictation is free and unlimited; the free plan adds 2,000 words a week of cloud dictation, and Premium is $7 a month billed yearly ($84) or $10 a month billed monthly. Enterprise is custom.
Trade-Offs: Benefits vs. Limitations
Whisper on your computer
- What it means in real use:
- Dictation can run offline once a model is downloaded, with no account.
- Limitation / caveat:
- Cloud options and telemetry exist; PromptQuorum did not inspect network traffic.
Free and open source
- What it means in real use:
- MIT code and unlimited local dictation at no cost.
- Limitation / caveat:
- Unlimited cloud dictation needs the paid Premium plan.
Dictate into any app
- What it means in real use:
- One shortcut works in email, chat, editors and documents.
- Limitation / caveat:
- Mac and Windows only; local Whisper needs macOS 15 or newer.
Six model sizes
- What it means in real use:
- Small models for older machines, larger ones for accuracy.
- Limitation / caveat:
- Large v3 needs about 3.1 GB of disk; no speed figures are published here.
Context-aware formatting
- What it means in real use:
- Output is shaped to the active app, such as email or chat.
- Limitation / caveat:
- Marked Alpha and needs a local language model or the cloud connection.
Active development
- What it means in real use:
- Version 1.13.0 shipped on 7 November 2026.
- Limitation / caveat:
- Voice notes are in progress; meeting transcription is only planned.
Amical vs. Alternatives
These are the closest dictation tools in the directory: a free open-source app for three desktop systems, a free Mac-only app, a hybrid open-source app and a paid closed-source app.
Amical
- Platforms:
- Mac, Windows
- Price and license:
- Free local, Premium cloud, MIT
- Role:
- Push-to-talk dictation
- Key difference:
- Open source; local Whisper plus optional cloud
Also mentioned in Amical (2)
Also mentioned in:
- Local Voice & Speech Tools Compared (2026): Text-to-Speech, Speech-to-Text and Voice AgentsUpdated November 13, 2026
- Best Local AI App for Non-Technical Users (2026)Updated October 28, 2026
Handy
- Platforms:
- Mac, Windows, Linux
- Price and license:
- Free, MIT
- Role:
- Push-to-talk dictation
- Key difference:
- Free and open source; adds Linux
Articles about Handy (2)
- Handy Review: Free Offline Speech-to-Text Dictation AppUpdated October 24, 2026
- Parrot Review: Free Offline Text-to-Speech That Reads Selected Text AloudUpdated October 18, 2026
Also mentioned in:
- Argos Translate Review: Free Offline Translation for Python and the Command LineUpdated November 13, 2026
- Local AI on Android in 2026: What Is Possible, What Is Not, and the Apps We CoverUpdated November 13, 2026
- Weesper Neon Flow Review: Offline Dictation for Mac and WindowsUpdated November 9, 2026
- Loci AI Review (2026): Offline AI for iPhone, Android, iPad, Mac and WindowsUpdated November 8, 2026
- LobeChat vs Big-AGI vs NextChat vs Page Assist vs Chatbox: 2026 ComparisonUpdated November 7, 2026
- Facet Review: Local Photo Scoring, Culling and GalleryUpdated November 6, 2026
- Swerve Review: AI Story and Character Chat for iPhoneUpdated November 6, 2026
+12 more not shown
OpenWhispr
- Platforms:
- Mac, Windows, Linux, iOS
- Price and license:
- Freemium, MIT
- Role:
- Dictation with local or cloud modes
- Key difference:
- You choose local models, your own API key or its cloud
Articles about OpenWhispr (1)
- OpenWhispr Review: Open-Source Dictation and Meeting Notes AppUpdated October 24, 2026
Also mentioned in:
- Local AI on iPhone in 2026: What Is Possible, What Is Not, and the iPhone Apps We CoverUpdated November 13, 2026
FluidVoice
- Platforms:
- Mac
- Price and license:
- Free, GPL-3.0
- Role:
- Mac dictation app
- Key difference:
- Free and open source; Mac only
Articles about FluidVoice (1)
- FluidVoice Review: Free Open-Source Mac Dictation AppUpdated October 24, 2026
Weesper Neon Flow
- Platforms:
- Mac, Windows
- Price and license:
- Trial, then paid, proprietary
- Role:
- Push-to-talk dictation
- Key difference:
- Closed source and paid; Whisper on the device
Articles about Weesper Neon Flow (1)
- Weesper Neon Flow Review: Offline Dictation for Mac and WindowsUpdated November 9, 2026
Also mentioned in:
- Local Voice & Speech Tools Compared (2026): Text-to-Speech, Speech-to-Text and Voice AgentsUpdated November 13, 2026
Platform, price, and feature details for third-party apps change frequently. Verify current specifics on each project's own page before deciding.
Who Should Use Amical
- Writers and office workers on Mac or Windows. One shortcut dictates into email, chat and documents.
- People who avoid cloud dictation. Local Whisper keeps speech recognition on the computer, per the documentation.
- Open-source users. The MIT code is public and local dictation is free.
- Multilingual users. 100+ languages per the maker, with automatic detection.
Who Should Not Use Amical
- Linux users. The README says Linux is not supported; Handy covers it.
- Mac users on macOS 14 or older. Local Whisper needs macOS 15 or newer.
- Anyone who wants no telemetry by default. The documentation says it is on by default and must be switched off.
- Anyone who needs speaker labels or live meeting transcripts today. Neither is documented as available.
Frequently Asked Questions
Does Amical work offline?
The FAQ says yes, once you have downloaded a local Whisper model; the first-run setup needs a connection for that download. Cloud models need a connection. PromptQuorum did not test it offline.
Is my voice sent to the cloud?
Only if you choose the cloud model or cloud formatting, per the FAQ, which says those send the audio for that request. The code is open and PromptQuorum did not inspect network traffic.
Is Amical really free?
Per the pricing page, local dictation is free and unlimited, and the free plan adds 2,000 words a week of cloud dictation. Only unlimited cloud dictation needs the paid Premium plan.
Can I use my own language model?
For the Alpha formatting pass, yes: Ollama or any OpenAI-compatible local endpoint. For speech, the documentation says you pick a shipped Whisper variant or the cloud model; bringing your own speech provider is not supported.
Who makes Amical?
The code is published by the GitHub organization amicalhq, and the privacy policy names Exa Labs LLC as the service provider.
Verdict
Amical is a good fit if you want free, open-source push-to-talk dictation on a Mac or Windows PC where, per the documentation, Whisper runs on your own computer. It has an MIT license, a free local tier and a release from this week.
The caveats are scope and verifiability: no Linux version, macOS 15 for local Whisper, telemetry that is on by default per the docs, an Alpha formatting pass, and privacy claims that PromptQuorum could not check. Try the free local tier before deciding on Premium.
Sources
- Amical website — features, platforms and pricing.
- Pricing page — Free, Premium and Enterprise plans.
- Documentation — local models, AI formatting, FAQ, telemetry and quick-start guides.
- Privacy policy — data handling, labeled for the mobile app.
- GitHub repository — README, MIT license and source code.
- GitHub Releases — installer assets and release date of version 1.13.0.
