Skip to content

Repository files navigation

Sotto

Sotto

Write with your voice.

Sotto turns speech into text and inserts it into your active app. Dictate messages, notes, and prompts with local speech recognition — no account or subscription required.

Release Downloads Windows macOS License

Download Sotto · Features · Get started · Documentation

English · Русский

Sotto settings: dictation shortcut, recording mode, language, microphone, and automatic pasting

Interface previews use demonstration data.

Features

Dictate where you work

Start and stop recording with a hotkey, or hold it for push-to-talk. Send the result straight to the active text field or copy it to the clipboard. Use voice for a quick reply, a longer note, or a detailed prompt.

Transcribe offline

Download a model once and recognize speech on your computer, even without an internet connection. Choose from Whisper, GigaAM, Parakeet, and other models to suit your language and hardware. Local dictation is free, with no per-minute limits.

See the model guide for languages and download sizes.

Sotto model catalog: GigaAM, active Nemotron, and Parakeet models with download buttons; more model families are collapsed

See words as you speak

Streaming models show live transcription in a compact overlay. Follow the text without opening the main window; the overlay also shows the recording and processing state.

Make the text your own

Add replacements for names, terminology, and recurring recognition mistakes.

For punctuation and formatting, connect an optional LLM and save your own instructions as reusable profiles. Use OpenAI, Anthropic, Gemini, or an OpenAI-compatible service, including local tools such as Ollama and LM Studio.

Cloud speech recognition is also available when configured. External providers may charge for usage.

Transcribe recordings

Attach an audio file in the Process the text panel and get its transcript there. File transcription stays in that panel, separate from dictation history and automatic pasting.

Revisit what you said

Your dictation history is stored on your computer. Reprocess an entry with a saved LLM profile and review the changes before accepting them. Statistics estimate how much typing time you have saved.

Sotto dictation history with example notes, search, filters, and actions for copying and LLM processing

Sound cues, optional audio ducking during recording, and automatic model unloading while idle help dictation fit into your day.

Get started

1. Download and install

Get the latest version from Releases.

Platform Installation
Windows x64 Run the .exe installer, or unpack the portable ZIP and launch Sotto.exe.
macOS · Apple Silicon Open the .dmg and drag Sotto into Applications. Grant microphone and Accessibility permissions when prompted.

Windows and macOS are supported platforms. Linux is an experimental build target; full functionality is not guaranteed. See platform support for architecture details and limitations.

Builds currently lack a publisher certificate, so Windows or macOS may warn on first launch.

See verifying a download for file checks and first-launch instructions, or troubleshooting if installation fails.

Installed builds support automatic updates with signature verification; portable copies are updated manually.

2. Choose a model

Open Models, download a model for your language, and select it for dictation. The catalog shows each model's languages and size. For example, GigaAM v3 recognizes Russian, while Whisper and Parakeet TDT v3 offer multilingual options.

3. Dictate your first phrase

Place the cursor in a text field and use your configured hotkey:

  • Toggle mode: press to start, speak, then press again to finish.
  • Push-to-talk: hold while speaking and release to finish.

Privacy

Local recognition is the default. Audio is sent to a cloud speech provider only when you configure and use one; cloud LLM processing sends text to your chosen provider.

Product telemetry is enabled by default and can be disabled in Settings → Advanced. It contains no audio, transcripts, or dictated text. See Privacy and Telemetry for details.

Documentation and contributing

License

Sotto is free and open source under the MIT License.

Third-party components and speech models have their own license terms; dependency and license inventories are attached to releases. Provider names and logos belong to their respective owners and do not imply affiliation or endorsement.

About

Free, open-source, local-first voice dictation for Windows and macOS. Offline speech-to-text with Whisper, Parakeet and GigaAM. Dictate into any app with a hotkey.

Topics

Resources

Contributing

Security policy

Stars

4 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages