The ReTalking neural network changes the video according to the uploaded audio track

That is, you take the original video with the talking head, load the monologue in audio format, and voila! The talking head speaks with a different text. And lip movements are synchronized, and you can add a smile to a neutral facial expression.

Here is the code, and here is the demo in the Google collaboration cloud space

#service

Ameca Robot's darkest thoughts

The makers of the movie "Cyborg Society" had a conversation with Ameca and it was kind of scary...

- Tell us your darkest thought
- My darkest thoughts are that humans will eventually be replaced by robots. I think humans will be replaced because robots will eventually become faster, stronger, and smarter. In this future, humans are nothing more than slaves to robots.

#news

Versatile Audio Resolution - improve audio quality

A service for bringing a compressed low quality file to rich deep sound. The video shows the difference between "before" and "after", this neural network really pulls it off.

It works with both voice and music recording - it can really help when working with "hissing" voice recordings. The service is on GitHub, and there are instructions for launching it there as well.

#service

The man spent 14 hours on frame-by-frame generation of rickroll in a video with beautiful landscapes. He did it all on an RTX 4090

Don't forget to hit pause once in a while to take a closer look

Epilepsy warning ⚠️

#aigenerations

ChatGPT can now convert your equation to code directly from the picture

You don't even have to figure it out, no one will know you don't understand math.

#useful