A Linux notebook that speaks, hears, and never phones home

What it does

Speech Note is a Linux and Sailfish OS note-taking app that handles speech-to-text, text-to-speech, and machine translation entirely on your local machine. It wraps a broad collection of open-source engines—from whisper.cpp and Vosk for transcription to Piper and Bergamot for translation—so you can dictate, listen to, or translate notes without sending data to the internet. The app ships without AI model checkpoints, relying on a built-in graphical browser to fetch them.

The interesting bit

Rather than betting on a single backend, the project acts as a neutral aggregator, bundling many distinct STT, TTS, and MT engines under one roof. That creates a sprawling mix-and-match language matrix where you might use Whisper to dictate in one language and Piper to read it back in another—an unusual level of engine choice for a desktop offline tool.

Key highlights

• Fully offline processing: no network connection is used for text or voice tasks.
• Broad engine roster includes Whisper, Vosk, Faster Whisper, Piper, RHVoice, Coqui TTS, Mimic 3, Kokoro, F5-TTS, Parler-TTS, and Bergamot Translator.
• Supports dozens of languages, though coverage varies significantly by engine.
• Runs on Linux desktop and Sailfish OS, distributed via Flathub.
• Built-in graphical model browser handles downloading checkpoints after install.

Caveats

• AI model checkpoints are not bundled with the app; you must download them separately via the built-in browser.
• Language support is uneven—many languages only work with a subset of the available engines, so check the matrix before committing.

Verdict

Grab it if you need a private, offline voice workflow on Linux and do not mind trading cloud convenience for local control. Look elsewhere if you want a single pre-bundled model or guaranteed high-accuracy results for every language out of the box.