An open-source voice input tool (MIT license) developed by agata. It runs OpenAI’s Whisper speech recognition model directly on your device, turning what you say into text in real time. Because all processing happens locally, there is no API key to register and no audio sent to the cloud. Transcribed text is copied to the clipboard automatically and, depending on your settings, pasted straight into the app in front of you. It supports macOS, Windows, and Linux, and it is free to use.
Key Features
- Fully local processing: Audio never leaves your device — recognition is completed on the machine itself. No API key is required, and there are no usage-based charges. It works offline as well
- Global hotkey to start recording: Ctrl+Shift+R by default (configurable). Whatever app you are in, a single key starts and stops recording. On Linux and macOS, recording can also be triggered via a SIGUSR1 signal
- Automatic copy and paste: Results are copied to the clipboard automatically and, optionally, pasted into the active window — so you can write directly into an email or chat input field
- Model management: Choose from Tiny / Base / Small / Medium / Large-v3 presets to balance accuracy against speed. The default is Large-v3 (about 3.1 GB), downloaded automatically on first launch
- User dictionary: Register terms that are often misrecognized — proper nouns, internal jargon — as sets of canonical form, alternate name, and apply condition, and have them applied as replacement rules
- Floating toolbar and device controls: A small always-on-top window shows the current state (idle, recording, processing). Input and output device selection, a level meter, gain adjustment, and test recording are all available from the settings screen
- GPU acceleration: Vulkan on Windows and Linux, Metal / Core ML on macOS
Pricing
| Plan | Price | Highlights |
|---|---|---|
| Open source | $0 | All features available. MIT license, no API key, no usage limits |
Pricing is current as of August 2026. Please check the official site for the latest information.
There are no paid plans or subscriptions. Because processing runs on your own device, there are no usage-based charges of the kind cloud transcription services apply.
Pros & Cons
✅ Pros
- Audio never leaves your machine, which makes it easier to handle confidential conversations or content with personal information
- Free with no usage-based billing, so cost does not grow with long or heavy use
- Works even without a network connection
- Available from a single hotkey, so you can dictate without switching apps
- The source code is public, so you can inspect and modify its behavior yourself
⚠️ Cons
- The default model (Large-v3) is about 3.1 GB, requiring an initial download and disk space
- Recognition speed depends on your hardware; waits can be long where GPU acceleration is unavailable
- 4 GB of RAM is required (8 GB recommended), which is demanding for lightweight machines
- There is no auto-update, so new versions must be installed manually
- Automatic pasting needs extra setup (accessibility and automation permissions on macOS; wtype or xdotool on Linux)
- It does not offer advanced post-recording features such as speaker separation or summarization
Comparison with Similar Services
| Criteria | HootVoice | superwhisper | MacWhisper | Cloud transcription services |
|---|---|---|---|---|
| Where processing runs | On device (local) | On device (mainly local) | On device (mainly local) | Cloud |
| Supported OS | macOS / Windows / Linux | Mainly macOS | macOS | Browser and various apps |
| Price | Free (open source) | Free tier + paid plans | Free tier + one-time purchase | Free tier + usage or monthly billing |
| API key | Not required | Not required (local use) | Not required (local use) | Often required |
| Source published | Yes (MIT) | No | No | No |
| Primary use | Real-time voice input | Real-time voice input | Mainly transcribing audio files | Automated meeting minutes |
Pricing and features change from service to service, so check each official site for the latest details when comparing.
Who Is It For
- People who handle work conversations or personal information and do not want audio sent to the cloud
- People who write a lot and want to reduce typing effort with voice input
- People concerned about the cost of usage-based transcription services
- People who want the same voice input environment on Windows and Linux as well
- Developers who want to inspect and customize an open-source tool as they use it
Summary
HootVoice combines accurate voice input with the reassurance of keeping data on your own machine by running Whisper locally. Being free, requiring no API key, and working the same way on macOS, Windows, and Linux matters a great deal for anyone who wants to make voice input part of daily work. On the other hand, local execution brings its own overhead — the size of the default model, memory requirements, and the absence of auto-updates. Start with a lighter model to confirm how it runs, then move up to a larger one as needed.