Free Open-Source Artificial Intelligence

4714 readers

1 users here now

Welcome to Free Open-Source Artificial Intelligence!

We are a community dedicated to forwarding the availability and access to:

Free Open Source Artificial Intelligence (F.O.S.A.I.)

More AI Communities

LLM Leaderboards

Developer Resources

GitHub Projects

GitHub Stars

FOSAI Time Capsule

founded 3 years ago

MODERATORS

Blaed@lemmy.world

fosai@lemmy.world

What Does a Privacy-First AI Translation Stack Look Like? (github.com)

submitted 1 week ago by PolyTalk_BizzAppDev@lemmy.world to c/fosai@lemmy.world

7 comments fedilink hide all child comments

Most AI translation tools rely on cloud services.

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

We wanted to explore a different approach.

PolyTalk is an open-source translation platform built around the idea that speech recognition, translation, and speech synthesis can be powered by open models and deployed on infrastructure you control.

The project combines open-source components for transcription, translation, and TTS into a privacy-first workflow.

Curious how others in the open-source AI community think about privacy and ownership when it comes to AI-powered communication tools.

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

top 7 comments

sorted by: hot top controversial new old

[–] hendrik@palaver.p3x.de 4 points 1 week ago* (last edited 1 week ago) (1 children)

I recently learned about the Offline Translator app. That's awesome. Allows to translate text, documents and what's in front of the phone camera. Completely on device and no external services needed.

I'm also a regular user of Mozilla Firefox Translate. Allows me to read news articles from other European countries, occasionally visit some Japanese websites...

They're all massively helpful. I like talking to people. Listen to perspectives beyond the standard American one (or German in my case). Or go shopping in an Asian supermarket. Sometimes I'll read a datasheet of some obscure electronics and it's in Chinese. And I live in one of the more multicultural regions, so it wouldn't hurt to be able to give directions in other languages. People get lost here all the times because the Deutsche Bahn sucks. And all I can do is speak German, English and 50 words of French. Which sometimes isn't enough. So I'm all for more translation helpers.

[–] PolyTalk_BizzAppDev@lemmy.world 2 points 1 week ago (1 children)

That's exactly the kind of use case that makes translation technology so interesting to me. It's not always about business meetings or travel, sometimes it's reading a news article from another country, understanding a product manual, or simply helping someone find their way.

It's great to see more translation tools moving toward on-device and privacy-friendly approaches. A few years ago, many of these workflows would have required sending everything to external services.

[–] hendrik@palaver.p3x.de 1 points 1 week ago* (last edited 1 week ago)

Yeah, a few years ago, most of that was Google Translate. To be fair it has some limited on-device features. All of this used to be proprietary technology, though.

Not sure if I need tools for business meetings, at least on a regular basis. We kind of all agreed to use either the local language or English as the universal language in software development. And people are expected to be somewhat fluent. And if you clients are abroad, you better hire a real translator at some point. Or you'll end up like Microsoft with all the messed up translations in Windows 11. It'll be handy at times, though. Or if you work on a construction site. And some other jobs.

[–] Sergio@piefed.social 3 points 1 week ago (1 children)

lel I worked on a couple speech interface projects back in the 00s before all these corporate spyware platforms emerged. Naturally, it was all on-device (or a local server we controlled). This was more R&D/prototype stuff so it wasn't as robust as systems nowadays, but the software is still out there:

Speech Recognition: https://cmusphinx.github.io/
we weren't doing translation so idk about that
Text-To-Speech: https://github.com/festvox/festival

[–] PolyTalk_BizzAppDev@lemmy.world 2 points 4 days ago (1 children)

That's really interesting. Sometimes it feels like local AI is a new idea, but a lot of the foundations were already there years ago.

The difference now is that the models have become good enough that these kinds of workflows are practical for everyday users, not just research projects.

[–] Sergio@piefed.social 1 points 4 days ago (1 children)

All AI was local until recently. (late 2010s maybe?) It's important not to let the cloud providers gaslight us.

these kinds of workflows are practical for everyday users

Kind of. A good system will still have a lot of design to it. If you just take an off-the-shelf LLM and do the minimal tuning for it to do the job, then you'll get just another crappy system.

[–] PolyTalk_BizzAppDev@lemmy.world 2 points 3 days ago

That's a fair point. A good user experience usually comes from the engineering around the model, not just the model itself.

The AI gets most of the attention, but things like latency, workflow design, context handling, and reliability often make the difference between something people try once and something they actually use.