Why VoxisLive exists
Text-based translation is too slow for the moments that matter: a livestream in a language you don't speak, a multilingual call, an undubbed video. Reading subtitles pulls your eyes off the screen — so VoxisLive speaks the translation instead, in a studio-grade voice at 24 kHz, about two seconds behind the speaker.
Four commitments
- Speak, don't subtitle. Reading subtitles pulls your eyes off the screen. The translation should arrive through your ears.
- No drivers, no friction. Direct system-audio capture through native Windows WASAPI — no virtual cables, no setup ritual.
- Open by default. The desktop engine is open source. Review the code, or bring your own API key and run it free.
- Your audio stays yours. Speech detection runs locally, nothing is retained after a session, no data is sold — and the open-source build sends zero usage data.
Who builds it
VoxisLive is built and maintained by Davut Akça, a solo independent developer from Türkiye. The project is self-funded — not venture-backed — and the desktop engine source is public on GitHub. Questions, ideas, or a bug to report? Get in touch — support is answered in English and Turkish.