Nvidia's microphone AI filtering tech might revolutionize recording voicebanks at home

[Exemplar]

I heard about this tech that Nvidia had recently released for their graphic cards & I thought this would be interesting news for folks who record voicebanks at home. What it does it uses the GPU to power an AI to help remove significant portions of background noises picked up by a microphone.

You can get Nvidia’s “RTX Voice” noise filtering without a pricey RTX card [Updated]

Ars testing (and install hack) shows noise-cancellation functioning on GTX 1060.

arstechnica.com

 

[uncreepy]

I downloaded this a few days ago and tried it out with Voidol, but it made Voidol sound worse/basically unintelligible more than it's usual bad quality. I think that this would be good for UTAU/DeepVocal voice bank creators, it really does get rid of noises like typing. But for vocoders, it seems like mechanical mistakes in output are not due to external noises and more like its due to its own patterns or maybe mic quality.

This is the tutorial I followed for my GTX:

 

[Prism]

I would love to see this with a noisy synth like ai kiritan or cevio

 

[Exemplar]

The upgraded version is out