Skip to main content

Nvidia RTX Voice App Eliminates Keyboard Noise From Video Calls

(Image credit: Shutterstock)

Now that most of us are working from home, it's becoming abundantly clear that many people not only lack the best microphones, they also lack a quiet environment that's conducive for voice calls. Nvidia aims to fix that with its latest app: RTX Voice.

No, Nvidia isn't ray tracing your voice. RTX Voice is an AI-based noise-cancellation tool that uses the Tensor cores present in GeForce RTX graphics cards. With these, it cancels out background noise, ambient noise, keyboard typing and more, giving whoever is listening to you a nice clean signal.

(Image credit: Nvidia)

However, it's no good if only your voice gets cleaned up for others if you still have to listen to your colleague munch away at a bag of chips while their partner is vaccuming the house in the background. For that, Nvidia's RTX Voice doesn't only clean up your voice signal, but also incoming signals before passing the AI-processed audio on to your headphones or speakers.

We fired up the app on our own system for a quick test, and the result was remarkable. Not only could the person listening to me hardly hear my typing anymore, I could eliminate the noise of their background chatter and typing on command.

Nvidia's RTX Voice is compatible with the following apps:

  • OBS Studio
  • XSplit Broadcaster
  • XSplit Gamecaster
  • Twitch Studio
  • Discord
  • Google Chrome
  • WebEx
  • Skype
  • Zoom
  • Slack

However, Nvidia did note that RTX Voice output may be troublesome on WebEx, Skype, Zoom and Slack. 

Click here to download RTX Voice, and head here for information to help with the setup if you get stuck.

  • bit_user
    This is really cool, but I'm sure it could also run on quite a few GTX cards, if not also CPU-only.

    Let's hope some opensource clones emerge soon, if they aren't already out there.
    Reply
  • lilkwarrior
    bit_user said:
    This is really cool, but I'm sure it could also run on quite a few GTX cards, if not also CPU-only.

    Let's hope some opensource clones emerge soon, if they aren't already out there.
    It would be drastically insufficient. Deep-learning-capable GPUs are the future. AMD RDNA2 will finally enable AMD GPUs to do the same and a solution that works for both.
    Reply
  • bit_user
    lilkwarrior said:
    It would be drastically insufficient.
    For audio processing? This audio processing, in particular? I find that hard to believe.

    lilkwarrior said:
    Deep-learning-capable GPUs are the future.
    ...and present, and past. Yeah, I get that. I'm just talking about this one "trick".
    Reply
  • bit_user
    If the author or any of the editorial staff is reading this: what is the GPU % utilization (and on what HW), when running it just to filter the microphone input?

    Thanks.
    Reply