For the past few days, we have been working on an open-source, self-hosted real-time speech-to-speech translation tool called PolyTalk. The goal was that there are people and organisations who need privacy around the tool they are using, and for the speech-to-speech translation, we haven’t had many options.

We built the tool with Ollama, Faster Whisper, and Piper.

The tool is not limited to speech-to-speech translation only, but you can also share any of your tabs, whether you’re watching a YouTube video in another language, the tool will give you audio output in your target language.

We are aware of how often context and tone get lost in translation, so we ensured translation quality by processing complete sentences instead of individual words.

Now we are focused on context support and tone adaptation.

If this interests you, here is the GitHub repo: https://github.com/PolyTalkIO/polytalk

Most AI translation tools rely on cloud services.

Audio leaves your device, gets processed elsewhere, and comes back translated.

As open speech recognition, translation, and TTS models continue to improve, it feels increasingly possible to build communication tools that run on infrastructure users actually control.

That’s one of the ideas behind PolyTalk, an open-source translation platform we’re building.

Privacy, ownership, and transparency may soon matter as much as model quality.

Do you think communication tools like translation, transcription, and speech interfaces will eventually move back toward local and self-hosted deployments?

GitHub: https://github.com/PolyTalkIO/polytalk