KrispNet – Voice Playback with Deep Learning (2hz.ai)

This seems really good, is the code open source?

The KrispNet audio sounds similar to overly compressed audio filled with artifacts. The ffmpeg interpolated audio sounds better to me.

Not a bad idea though and will probably be a thing with proper refinement.

Oddly, the original and ffmpeg versions play fine on firefox (win 7, 64b, version 67.0), but the krisp version doesn't respond when clicked. I don't understand why not, as all samples are links to .wav files. It works on chrome.

Wasn't able to play any KrispNet samples from the Firefox browser I am running on my Ubuntu laptop.