#GrapheneOS Speech Services, a high quality on-device text to speech engine built with open source models and training data is now available in the GrapheneOS App Store. image

Replies (27)

Yes. The data is the public domain LJ Speech dataset. LJ Speech uses user-submitted public domain recordings of audiobooks. LibriVox (audiobook platform) requires all recordings made for them to be in public domain.
plasticlove's avatar
plasticlove 2 weeks ago
damn we got People for the Ethical Treatment of AI models on nostr 😅
plasticlove's avatar
plasticlove 2 weeks ago
looking forward to it, been using whisper+, but it's been sorta painful.
Let's gooo!! Just installed to test it. I was using RHVoice with Organic Maps went well... But would not read text out loud for me.
Will this be available on F-Droid? I am a combination magnification and talkback user who uses RH voice and so would be very interested in this.
We don't distribute on F-Droid, but this can be used in other operating systems. Best bet is to get the GrapheneOS App Store app so you can get automatic, verified updates for it.
KernelKind's avatar
KernelKind 2 weeks ago
amazing, thanks to the team for all the hard work!
RIP FUTO Keyboard, which works great, but is not fully open source. FUTO open source in the way that you can view the source, but not branch it and do your own thing with it.
I found the button deep in settings. Which map apps support this? Would you/GrapheneOS be willing to contact some of the major map apps to provide better integration?