Technical investigation into PS4 and XB1 audio solutions *spawn

[Davros]

1. Has that been confirmed ?
2. I don't think the function of TrueAudio is to decompress audio streams

 

[bgroovy]

1. Yes.
2. One of the oldest known capabilities of the PS4 audio chip was its ability to decompress "hundreds" of audio streams (see: VGLeaks, Cerny interviews)

 

[Rurouni]

From what I understand, there is a programmable audio dsp chip which might be the same chip that powers AMD TrueAudio. The difference is that devs doesn't have direct access to the chip. Basically to the devs, the chip probably acted like a fixed function stuff (audio decompression and maybe something else?), thus less flexible in terms of utilization by devs vs X1 shape and AMD TrueAudio.

 

[Cyan]

As far as I understood, Sony has something similar to AMD TrueAudio in PS4. Why they'd need a processor reserved for audio?

Afaik, the PS4 has a decompressor chip for audio, although for example the Xbox 360 has one too, but X360's audio was processed on the CPU.

PS4 is going to use the GPU -like with AMD TRueaudio technology- for that technology, again...., afaik, thus until that happens the audio is processed on the CPU, according to Killzone: Shadowfall screengrabs from developers.

 

[Davros]

PS4 is going to use the GPU

What do you mean "going to use" why isnt it using it now?

 

[Allandor]

What do you mean "going to use" why isnt it using it now?

because it needs time to develop something like this. Right now, nobody has done this in a game, so the sound engines must first get developed before using them. Also, if you use the GPU for sound processing, you loose some processing time for the graphics. Right now it is not needed to do that on the gpu. Also, sound processing is somewhat a time-critical task, so you shouldn't use async compute because you definitely must have the resources to do the calculations in time.

For the xbox one I hope that shape will be used to offload some cpu-ressources. Even if the CPU could do something faster, it is better to use shape than the cpu. Shape can only be used for some kind of tasks, the cpu however can be used for many other things. if it is not needed to use the cpu for at least some audio tasks, you have more cpu reserves to spare at other parts of your game.

 

[Davros]

because it needs time to develop something like this. Right now, nobody has done this in a game, so the sound engines must first get developed before using them

I dont buy that "its new its going to take time before the devs learn how to use it, the gpu was new nobodies saying "the gpu is new, non of the advanced features it has over the previous gen gpu will be used for a year or so untill the devs learn how to use it" afaik the audio middleware supported the consoles audio hardware from day 1.
This quote is from may 2013, 6 months before xb1 was released.

The high performance processor and dedicated Scalable Hardware Audio Processing Engine (SHAPE) is fully utilized by FMOD Studio to deliver next generation audio for the next generation of games. Authorized Xbox One developers and publishers can download an evaluation today.

look at something very similar trueaudio on the pc, games supporting it came out very quickly after it was released. and unlike shape which is a default part only a small percentage of users have a trueaudio capable system.

For the xbox one I hope that shape will be used to offload some cpu-ressources.

It is, it allows xb1 games to ship with compressed audio thats the reason for its existance, wether it does other stuff apart from the convolution reverb there's practically no info that is presented in a format a non audio programmer can understand

 

[Cyan]

What do you mean "going to use" why isnt it using it now?

Because early SDKs on the PS4 used the CPU cores for audio logic instead of TrueAudio, which is GPU based -gpgpu I think-. Allandor explained it.

As for your previous post, Shape is part of the audio chip on the Xbox One, it has some extra cores that could be unlocked -meaning not used by Kinect- and create some very cool audio effects, but bkilian himself said that it depends on the console designers to let developers use those extra features to create some very nice extra audio effects like those you mention.

 

[Davros]

Do you have a quote for that did bilikan just say extra cores could be freed up for use in games and you've took it to mean "cores could be freed up to create cool audio effects"
Does he give the name of any of the effects?
looking at the block diagram the bits built for connect dont seem to be at all usefull for use in games
as for the shape itself i agree with Betanumerical who said
"From this exact diagram all shape does is Decode streams, mix with clip detection run a EQ and state variable filters on them and also so sample rate conversion."

 

[pjbliverpool]

Because early SDKs on the PS4 used the CPU cores for audio logic instead of TrueAudio, which is GPU based -gpgpu I think-. Allandor explained it.

TrueAudio used 3 dedicated programmable DSP's AFAIK rather than the GPU's compute resources. So in that sense it's similar to SHAPE, although I believe SHAPE has even more dedicated resources, albeit some of those being dedicated to Kinect. I assume PS4 has a similar capability to TrueAudio although I don't think that's been confirmed. It may well be more limited and/or locked down. In either case though I don't think GPGPU comes into it.

 

[Davros]

They are all using Tensilica cores

 

[Cyan]

Sony and MS have had some conferences on several issues yesterday, one of them being an audio face-off.

Interestingly, Sony and Microsoft will also face-off on the audio department.

Microsoft’s Advanced Technology Group Senior Development Lead Scott Selfon will start with the seminar “The Future is Now – But Now What?” at 11:15 AM.

“Next gen” has arrived, with ever increasing technical capabilities in both hardware and software processing. But are these new consoles really just offering the same thing, only more of it? What are the audio frontiers and barriers when so many of the restrictions of the past have been eliminated? And with maturing and increasingly sophisticated audio engine solutions, is game sound programming really now a “solved” problem? Scott reflects on the current state-of-the-art for areas ranging from spatial simulation and acoustic modeling to evolving and dynamic mixing, audio as a feedback mechanism, and highly personalized audio experiences. He’ll use examples from both past and present to highlight the technical achievements that implementers are striving for in making audio not only compelling and realistic, but an equal-footing contributor to immersive, engaging, and rewarding gameplay experiences.

At the end of the afternoon, at 5 PM, Sony Worldwide Studios’ Creative Services Group Audio Director Garry Taylor will host the Audio Keynote.

Garry Taylor has a unique vantage
point over today’s game audio landscape. Heading up audio for Sony Computer Entertainment Europe, he has oversight of content provision and strategy for multiple studios and Playstation titles across PS4, PS3 and PS Vita, as part of Sony’s Worldwide Studios Group and Audio Standards Working Group.

In a wide-ranging keynote address, he will explore his experience of the past, present and future of sound, music and dialogue for videogames, celebrating and in some cases questioning what has gone before, whilst examining the key ingredients that will take the medium forward. How have audio roles, best practice and game audio tech developed, as game development itself has evolved and changed? How has people’s consumption of audio changed? What do mobile platforms and the rise of headphone listening mean for game audio? What about 3D audio over headphones and what of VR? What does the advent of Dolby Atmos mean for game development, if anything? What’s next for toolsets?

Referring to processes, team integration, tools, hierarchy, equipment, training, communication and game design involvement, he will outline the factors he believes necessary for the future success of game audio.

http://www.dualshockers.com/2014/07...kinect-tomorrow-at-develop-face-off-on-audio/

 

[chris1515]

Afaik, the PS4 has a decompressor chip for audio, although for example the Xbox 360 has one too, but X360's audio was processed on the CPU.

PS4 is going to use the GPU -like with AMD TRueaudio technology- for that technology, again...., afaik, thus until that happens the audio is processed on the CPU, according to Killzone: Shadowfall screengrabs from developers.

We have a presentation about audio some part like compression/decompression are done on the DWP true audio part, some part with CPU and some part can be done on GPU.

 

[Cyan]

Remedy presents Umbra. This is similar to what @Davros was asking for, I think. Quantum Break is going to use Umbra graphics technology software for sound.

https://translate.google.com.br/translate?sl=fr&tl=en&js=y&prev=_t&hl=pt-BR&ie=UTF-8&u=http://www.gameblog.fr/news/51841-quantum-break-le-logiciel-umbra-se-montre-et-s-entend-en-vid&edit-text=&act=url

Do you have a quote for that did bilikan just say extra cores could be freed up for use in games and you've took it to mean "cores could be freed up to create cool audio effects"
Does he give the name of any of the effects?
looking at the block diagram the bits built for connect dont seem to be at all usefull for use in games
as for the shape itself i agree with Betanumerical who said
"From this exact diagram all shape does is Decode streams, mix with clip detection run a EQ and state variable filters on them and also so sample rate conversion."

If you were hoping for some cool audio techs in this generation, maybe this one sounds good to you.

 

[Deleted member 11852]

I'm pretty sure I saw the Umbra logo when I started Batman Arkham Knight. It seems to be everywhere.

 

[AlNom]

Aye car-umbra

 

[Jay]

I'm pretty sure I saw the Umbra logo when I started Batman Arkham Knight. It seems to be everywhere.

I'm pretty sure its usually used for graphics (maybe AI also) and not the audio side.
I think that's a section from this years build (or gdc), if you want I can go and find the actual session a bit later if you want.

 

[Jay]

GDC 2015 Using Umbra Spatial Data for Visibility and Audio Propagation in Quantum Break
http://www.gdcvault.com/play/1022402/
26min in.

 

[Davros]

Had a look on umbra's web site
Found this quote

Umbra runs on virtually any modern CPU and does not depend on any specific hardware or external software

Again no mention of shape.
Betanumerical's info seems to be spot on from what i can gather

 

[Cyan]

Had a look on umbra's web site
Found this quote

Again no mention of shape.
Betanumerical's info seems to be spot on from what i can gather

And what if you can use those precious CPU cycles on Shape instead on the CPU?