AMD is apparently working on outfitting its Radeon GPUs with noise cancellation, in a system that could theoretically rival Nvidia’s RTX Voice technology (now renamed ‘Noise Removal’ by Team Green – more on that later).
Team Red plans to name AMD Noise Suppression, and we got a glimpse of the feature thanks to a video on the company’s official YouTube channel – although that clip was quickly removed (or rather, made private) when people got wind of it.
Fortunately, as Tom’s Hardware (opens in new tab) reports, an on-the-ball Redditor (opens in new tab) (“crazydaveyboy”) captured and shared the pertinent details, and there was even a teaser video about Noise Suppression posted elsewhere on Reddit (opens in new tab) (and although the post has been removed by mods, the short clip is still visible).
AMD Noise Suppression, according to these leaked details, is designed to “communicate without distraction” and offers “two-way noise cancellation for input and output” that works in “various apps and games”.
Essentially, it will clean up the audio using AI (deep learning) in much the same vein as RTX Voice, as mentioned, improving the audio quality of your video chats or streams by, well, suppressing background noise, as the name suggests.
According to the aforementioned Redditor, the Audio and Video tab feature is in the Radeon Adrenalin driver, and enabling it installs a new virtual audio device (sounds like the way RTX Voice works) that you can use in whatever any apps you use.
Since this leak comes from an official source, it seems like a pretty sure bet that we’ll get a full disclosure about AMD’s technology soon. The teaser clip notes that Noise Canceling is “available now” at the end, so with the launch video prepared as such, we can imagine something coming in soon enough.
Analysis: everything you can do, we can do better…
What we don’t know, of course, is whether this AI-powered audio cleanup requires specific hardware, and one of AMD’s more contemporary GPUs. Nvidia pushed RTX Voice for Turing GPUs (the last generation before Ampere RTX 3000), and at launch it was only for RTX graphics cards as the name indicated – but since then it works for GTX cards too (and has been rebranded from RTX Voice to become the ‘Noise Removal’ feature in the Nvidia Broadcast app).
We’d expect AMD to adopt this technology widely, as Nvidia has made it available for GTX models (and obviously, in that case, the AI won’t need the specialized Tensor Cores found on RTX GPUs). AMD’s philosophy with FSR, the upscaling technology, was to make sure it works with older graphics cards compared to DLSS, so we expect this to apply on the audio front as well.
Anyway, this is good news for AMD graphics card owners, as we’re big fans of Nvidia’s RTX Voice (or rather, Noise Removal). It’s very effective at dealing with background noise and distractions, so we hope AMD could achieve similar results with what sounds like a similar AI-powered method (although the first incarnation may need constant tweaking, as always).
It’s also interesting to see that AI can increasingly become part of AMD’s arsenal of tools. It doesn’t seem to be introduced just for this audio trickery, but rumor has it that the plan will be to eventually equip FSR with AI Chops (perhaps with version 3.0).
That’s despite some pretty hefty arguments from AMD that machine learning isn’t necessary to make a good scaling solution (because, as you probably know, Team Red’s current FSR 2.0 doesn’t use AI – and to be fair , the results of the second generation framerate booster are still great).