That’s why you ask permission first, every one has their cloned voice stored on local machines and if permission is granted, we send the voice sample to the other user’s device and train the model localy there too, so it stays in his own sanbox.

That ask permission touch is wonderful 🤌 well done!

I’ve thought about it more and it could be interesting to learn a new language like that. By hearing your own voice speaking another language with the right accent and intonation, I believe you could learn a language faster and better.

Hey @egonvanpraet, I totally get that. However I feel that even if it might be legally okay to just ask for a permission, users are often not aware of all the data they share with that. We should be super sensible and onboard users carefully so they understand the trade offs.

This reminds me of the Waze feature for using a friends voice for navigation - love this design, clean, simple. 👏

How about using your own voice? I wonder how you’d feel and what you’d experience if your queries are being answered by… yourself.

A bit creepy to me, thinking about how do you get the data? What about data privacy?

Yeah I thought about that, could be a great experience.

Posts