All projects
Play a YouTube video, a class or a talk in English and get two live panels: what is being said, and its Spanish translation. Speech recognition runs offline on your own PC — your audio never leaves your computer — and nothing is recorded or saved.
Key advantages
- Transcribes the audio playing on your PC, in real time
- Spanish translation next to the English text
- Speech recognition 100% offline: your audio stays on your machine
- Ephemeral session: nothing is recorded
- Pick which audio output to listen to
- About 1–2 seconds of delay, typical of live subtitles
Links
Downloads
Linux
Under the hood
A Linux desktop Flutter app that captures the system-audio monitor (PipeWire/PulseAudio), runs streaming speech-to-text offline with sherpa-onnx (zipformer) and translates the text EN→ES. Only text travels to the translation service; the voice model downloads once on first use.
Fluttersherpa-onnxLinuxSTT
A project like this, for your business
Apps and systems like this are designed and built end to end. Describing the need is enough to receive a proposal.