Vosk

Free lightweight offline speech recognition — runs on a Raspberry Pi.
Vosk is a free, open-source offline speech recognition toolkit that runs on everything from Raspberry Pi to servers — 20+ languages, tiny models (50MB), real-time streaming, and zero cloud calls. Apache-2.0 licensed and used by home automation projects, accessibility apps, and privacy-first voice assistants. The perfect free STT engine when Whisper is too heavy.
Tags
Vosk · vosk · open source speech recognition · offline STT free · free speech to text lightweight · Raspberry Pi AI · Audio & Music · free AI tool · voice recognition offline · Apache 2.0 STT · real time transcription free · Simulation