NVIDIA RTX (Voice Plugin)

[Bad 💀 Motha]

How to use your Nvidia graphics card to improve the quality of your calls.

RTX Voice is available in beta for Nvidia’s RTX GPUs


https://www.theverge.com/platform/amp/2020/4/21/21229508/nvidia-rtx-voice-background-noise-reduce-graphics-card-improve-call-quality

https://www.kitguru.net/components/graphic-cards/christopher-nohall/nvidias-rtx-voice-helps-suppress-background-noise

https://www.nvidia.com/en-us/geforce/guides/nvidia-rtx-voice-setup-guide

[mimizukari]

so how well will my TITAN RTX deal with noise suppression?


cuz TITAN RTX isn't Quadro RTX or Nvidia GeForce (GTX)

[Bad 💀 Motha]

All RTX works the same, it uses the Turing AI within the tensor cores to filter out the noise in real-time. Doesn't matter what RTX gpu you have, all should produce the same great results.

GTX is not anything RTX so let's just nail that in the bud. No GTX gpu is able to use this plugin.

[mimizukari]

might check it out next time I have an operating microphone. does it have any performance degradation when you use it since some cores would be dedicated to that process?

[Bad 💀 Motha]

I don't see why it would. It doesn't take much to handle audio. Also since it only uses the Turing AI, how could it ever really affect performance unless you running something that's using Ray Tracing, DLSS, etc... stuff that actually use those cores, non RT games don't use those cores.

Much like how ShadowPlay does its thing without really affecting overall game performance.

[Autumn_]

Kurumi Tokisaki님이 먼저 게시:

so how well will my TITAN RTX deal with noise suppression?


cuz TITAN RTX isn't Quadro RTX or Nvidia GeForce (GTX)

https://en.m.wikipedia.org/wiki/GeForce_20_series
Wiki says the RTX Titan is a Geforce card.

All older Titans were Geforce cards too. I don't see why the RTX Titan would be any different.

Kurumi Tokisaki님이 먼저 게시:

might check it out next time I have an operating microphone. does it have any performance degradation when you use it since some cores would be dedicated to that process?

I'm pretty sure GPUs handle some audio processing anyway.
Not sure on how much though.
I don't see how cleaning the audio up would cause that much more usage.
And, depends on what cores it uses.

Bad 💀 Motha님이 먼저 게시:

Much like how ShadowPlay does its thing without really affecting overall game performance.

That's because graphics cards have ASICs for video encoding/decoding.
Completely different cores for a specific task.

[mimizukari]

Bad 💀 Motha님이 먼저 게시:

I don't see why it would. It doesn't take much to handle audio. Also since it only uses the Turing AI, how could it ever really affect performance unless you running something that's using Ray Tracing, DLSS, etc... stuff that actually use those cores, non RT games don't use those cores.

Much like how ShadowPlay does its thing without really affecting overall game performance.

shadowplay makes the average person lose about 5% of their FPS, it's not magically no-impact, and 5% is pretty big in some games.

[Autumn_]

Kurumi Tokisaki님이 먼저 게시:

Bad 💀 Motha님이 먼저 게시:

I don't see why it would. It doesn't take much to handle audio. Also since it only uses the Turing AI, how could it ever really affect performance unless you running something that's using Ray Tracing, DLSS, etc... stuff that actually use those cores, non RT games don't use those cores.

Much like how ShadowPlay does its thing without really affecting overall game performance.

shadowplay makes the average person lose about 5% of their FPS, it's not magically no-impact, and 5% is pretty big in some games.

Do you have any source for that info?
In my personal testing, I haven't seen any impact on performance.

[mimizukari]

Autumn님이 먼저 게시:

Kurumi Tokisaki님이 먼저 게시:

shadowplay makes the average person lose about 5% of their FPS, it's not magically no-impact, and 5% is pretty big in some games.

Do you have any source for that info?
In my personal testing, I haven't seen any impact on performance.

I don't notice any drops with my current setup... for reasons, but on my previous setups that used like gtx 960, 1050, etc i definitely lost 3~ frames on average from games I was getting 60 FPS on otherwise, sometimes 5 frames.

[iceman1980]

Zzzzzzz.

[Masque]

This is fascinating. Does this mean the RTX hardware could also be used for better software synths and effects in the future?

[iceman1980]

Boofalo Soja님이 먼저 게시:

This is fascinating. Does this mean the RTX hardware could also be used for better software synths and effects in the future?

https://www.tomshardware.com/news/nvidia-rtx-voice-works-fine-on-non-rtx-gpus

Nothing to do with RTX. Marketing gimmicks used to make people fawn all over the "latest Nvidia advancement". People really have no idea what these Tensor cores do? do they?
FP32 and FP16

This means the following considering FP32 is just the ones-complement representation using the first bit as the sign and the next bits as the mantissa / significand and the last bits as the exponent.

Mixed matrice multiplication, meaning multi-dimensional matrices with mixed 32bit precision / 16bit precision

32bit floating point: 1 bit for the sign : 8 bits for the exponent and 23bits for the mantissa / significand part.

16bitt floating point: 1 bit for the sign 5 bits for the exponent and 10 bits for the fractional component.

This is ALL that is happening inside a Tensor core they're just really fast matrice crunchers. Of half-precision and single-precision numerical representation. Tensor cores sacrifice precision for performance to reduce memory impact and gain performance. as a 32bit number is ones-complement 4bytes per number and FP16 is 2 bytes per number.

[Snow]

Discord already filters out all the noise quite well, and for recording you'd want to configure noise suppression manually, so like whatever. No noise suppression beats having a good microphone.

[iceman1980]

Snow님이 먼저 게시:

Discord already filters out all the noise quite well, and for recording you'd want to configure noise suppression manually, so like whatever. No noise suppression beats having a good microphone.

Good luck with that in regards to "most gamers gear" only sampling about 8kHz - 12kHz. So "good microphone combined with a superior DAC is the way to go" (PCI-E DAC's etc)

[Snow]

Product ∏님이 먼저 게시:

Snow님이 먼저 게시:

Discord already filters out all the noise quite well, and for recording you'd want to configure noise suppression manually, so like whatever. No noise suppression beats having a good microphone.

Good luck with that in regards to "most gamers gear" only sampling about 8kHz - 12kHz. So "good microphone combined with a superior DAC is the way to go" (PCI-E DAC's etc)

Only gaming gear I have is a couple of controllers, and they don't sample ♥♥♥♥. But I don'get get why you quoted my post and then said the same thing.

[Bad 💀 Motha]

Kurumi Tokisaki님이 먼저 게시:

Autumn님이 먼저 게시:

Do you have any source for that info?
In my personal testing, I haven't seen any impact on performance.

I don't notice any drops with my current setup... for reasons, but on my previous setups that used like gtx 960, 1050, etc i definitely lost 3~ frames on average from games I was getting 60 FPS on otherwise, sometimes 5 frames.

Thats because those were crappy gpu to begin with, not even good enough for many demanding games, let alone added tasks on top of that. Something like a 970, 1060 6gb, or rtx 2060 wouldn't see such impacts.