Latest Steam Audio SDK beta adds AMD TrueAudio Next support

About a year ago, we brought news of Valve's Steam Audio SDK, a set of software tools designed to aid game and VR developers in adding positional audio effects to their titles. The game developer and publisher just released Steam Audio 2.0 beta 13, a release that adds support for AMD's TrueAudio Next technology (TAN). The new version supports TAN in the Unity and FMOD Studio development environments. TrueAudio Next uses standard GPU compute resources instead of the specialized audio processing hardware that the original TrueAudio required.

Latency—less is better

True Audio Next is a software library that implements audio convolution on supported GPUs. Convolution is a math operation that applies a specific type of filtering to an audio signal. Common convolutions implement delay (echo), frequency equalization, head-related transfer functions, and reverb. Those effects, especially reverb, are compute-intensive operations and usually run on the CPU. According to Valve, most game audio engines allow only two to four simultaneous convolution reverb filters in order to keep hardware load within acceptable limits.

The new Steam Audio release also supports Resource Reservation, a technology that allows developers to set aside up to 25% of a GPU's compute units for audio processing. Sound processing tends to be extremely sensitive to latency. Reserving computational resources for sound processing could potentially lead to decreased pops, clicks, and other audio glitches. According to Valve, CPU resources cannot be reserved for sound processing in the same way. The charts on Valve's announcement page say that a system with as few as four CUs set aside for audio processing could handle as many as 80 convolution sources while remaining within acceptable latency limits.

The big caveat for the majority of PC gamers is that the new features are limited to "compatible GPUs," which means Radeon R9 Fury, Fury X, Pro Duo, RX 470, RX 480, RX 570, and RX 580 cards. Those who want to read more can follow this link. Those looking to get their hands dirty right away can download Steam Audio here, and the Unreal Engine, Unity, and FMOD Studio plugins here. We were unable to find a list of games using Valve's Steam Audio, but given the relatively young age of the SDK and the importance of positional audio in VR, we suspect the list is heavy with Steam VR titles.

Comments closed
    • moose17145
    • 2 years ago

    So what ever happened to the first incarnation of AMD’s TrueAudio that I believe was baked into the R9 2xx and 3xx series using dedicated hardware?

    Did that just die off and become largely forgotten about? Because I remember when AMD showed it off… but that it seems like everyone just kind of forgot about it after that… or at the very least I certainly was not hearing anything more about it.

      • LostCat
      • 2 years ago

      Only a few devs jumped onboard. Beyond that I can only assume they didn’t want to pay for third party tech, it wasn’t able to do functionality they wanted to offer, and they talked with devs about why it wasn’t being used.

    • ronch
    • 2 years ago

    To be honest, while having better audio is great, does anybody still really care about TrueAudio? I myself actually totally forgot all about it already.

      • LostCat
      • 2 years ago

      Technically, not even AMD does. TrueAudio Next is a different standard and with Valve implementing it I’d say someone does.

    • brucethemoose
    • 2 years ago

    Funny how VR is the big driver behind game audio improvement. “Regular” games need it too… But apparently no one really cared unless it was VR 🙁

    • brucethemoose
    • 2 years ago

    OK… But this is dead in the water if it only works on new AMD cards, just like the original TrueAudio.

    Superfluous physx effects are one thing, and Nvidia has a hard enough time pushing that as is. But unlike particle fountains, audio is a core feature of a game. Most devs aren’t going to dump man-hours into tuning and testing a secondary audio implementation for their game.

      • LostCat
      • 2 years ago

      As part of Steam Audio, it’ll most likely be used by any developer using Steam Audio.

      …that’s kind of the point, devs not needing to do the implementation because it’s already done.

        • brucethemoose
        • 2 years ago

        Is it really that plug and play?

        Even so, that’s still a separate feature that would require its own round of testing, as “basic” Steam Audio and the amped-up AMD version are going to sound and behave different in-game.

          • Klimax
          • 2 years ago

          Not to mention it needs fine tuning of GPU usage.

          • LostCat
          • 2 years ago

          I’m not an audio developer so I’m not about to download the tools and find out.

      • EndlessWaves
      • 2 years ago

      Except they’re doing just that for other technologies with much smaller potential audiences such as HDR.

    • DragonDaddyBear
    • 2 years ago

    Sound is one area gaming has regressed significantly. Here’s to hoping it’s widely adopted and gets us back the EAX/Direct Sound features that were popular over a decade ago.

      • LostCat
      • 2 years ago

      There’s been a lot of progression since Vista that hasn’t been in peoples faces…huge advances in quality and latency in Windows APIs that don’t usually make the front page. New APIs and newer versions of existing ones as well.

      Here’s a bit more info, there’s more to be had but I should probably be doing other things at the moment.
      [url<]https://docs.microsoft.com/en-us/windows-hardware/drivers/audio/low-latency-audio[/url<]

        • DragonDaddyBear
        • 2 years ago

        Totally agree, computer audio has improved. Motherboards even have passable solutions built in. However, the EAX dynamic surround audio is dead and no standard has arisen to replace it. GAMING audio is not where it once was.

          • LostCat
          • 2 years ago

          I suspect the developers of Steam Audio would disagree. 🙂 But I’m not in the field so probably not the best person to argue the point.

          Meanwhile, VR and Atmos seem to have driven a push towards more realistic audio – [url<]https://www.reddit.com/r/GameAudio/comments/7p81xf/why_is_the_game_dev_community_going_crazy_about/[/url<] As for the market in general? I don't know enough to say. I've certainly been disappointed by a lot of it, but before now FMOD and Wwise seemed to be doing a pretty good job.

    • DoomGuy64
    • 2 years ago

    Great, now get nvidia and creative support and we’ll finally be back to the game audio quality of the 90’s, and Microsoft won’t be able to pull any more Vista jobs on us to sell xbox.

      • Takeshi7
      • 2 years ago

      Creative’s crappy drivers are the reason Microsoft had to pull that Vista job. If they were good drivers it never would have been an issue.

        • DoomGuy64
        • 2 years ago

        That’s not accurate, and I know that because I followed the problems in detail, being a x-fi user, and the vast majority of the public’s opinion on it is based on a mishmash of disinfo.

        Creative’s rep is mostly due from poor business tactics and bad PR. The original X-fi had a lot of issues due to bad motherboard PCI implementations, poor initial hardware design, and lack of driver compensation of the cards themselves. People using good motherboards or audigy cards didn’t have issues, and it took Creative too long to address the issues in the driver, and they weren’t public about the issues or possible fixes. Not to mention the whole x64 change over was also an issue for them. The cloud of hate is due to Creative not being public about the problems, so people made up their own theories about what actually happened. Creative then screwed audigy users over with features being exclusive to x-fi users in Vista, and shtf from that too. There are multiple levels of bad practices and PR fumbles, and people don’t separate these and jumble it together into a mishmash of blind creative hate.

        Nevermind that Creative eventually did fix most of those issues, and you can get working drivers for the x-fi in windows 10 today. Compare that to the Xonar / Cmedia cards, and it’s a vastly different situation, as those cmedia cards have major issues with windows 10.

        Creative just released a new 32-bit soundcard, which seems to be pretty good from any review that goes into depth about how it works. Whether or not it can do any advanced 3d processing is unknown, since soundcards have moved completely into virtual surround functionality and headphones due to the lack of OS sound functionality. If any of creatives recent cards can do some level of 3d, albeit currently restricted to OpenAL, it would make more sense to port that to steam audio at this time.

          • LostCat
          • 2 years ago

          The one dev I spoke to said OpenAL was ridiculously unstable. I wouldn’t get your hopes up on that ever coming back.

            • DoomGuy64
            • 2 years ago

            Early OpenAL with early x-fi cards with early drivers. Once you started getting games like Bioshock and working Vista drivers, OpenAL was fine. UE engine games worked great with OpenAL, no issues.

            You also have the issue of OpenAL soft, which didn’t care to keep decent backwards compatibility with OpenAL, let alone older versions of OpenAL soft. Amnesia Dark Descent is probably the best example of this, as it uses an older version of OpenAL soft that doesn’t work with OpenAL or newer versions of OpenAL soft. That’s not a problem with OpenAL, that’s on the OpenAL soft developers, and also the game developer for not updating the game. Rapture3d is a much better software implementation, but it costs money. There’s also another free port called AeonWave-HD 64, which claims to be more efficient, but I have no personal experience with it.

            If you’re going to use a software version of OpenAL, then it needs to be stable and support the features of hardware OpenAL. It needs to be free, open source, and well supported by the entire community. Software OpenAL is mostly iffy because it doesn’t have solid community support or a definitive edition that everyone can get behind.

            Steam Audio is the going to be the future standard, because it does everything that OpenAL soft failed to accomplish. Valve supports it, and is constantly adding improvements and working with partners like AMD. If Steam audio adds OpenAL support, that will seal OpenAL soft’s fate, not that they haven’t already done most of that by themselves.

          • Klimax
          • 2 years ago

          I don’t think Creative made their own chips for a long time. They’re C-Media since about PCI-E.

            • DoomGuy64
            • 2 years ago

            Nope. It started with EMU20K2 with the normal cards, and CA20K2 for the specialty cards. That moved to DDR, included a RISC chip to buffer pcie latencies, and had native UAA. Then SoundCore3d for the Recon3D, which integrated everything, and the Z series returned to dedicated high quality hardware.

            I have no idea about the specifics of the AE-5, other than it is 32-bit and has dedicated amps for each channel on headphones, but it certainly is something unique and improved over the previous chips. Really doubt they’re using C-Media, especially since the soundcore based cards work in windows10. Neither the soundcore or AE-5 series have any sort of decent hardware documentation, which seems to be typical Creative secrecy, but they are undoubtedly Creative chips.

          • hansmuff
          • 2 years ago

          I’ve had SoundBlaster PRO, 16, AWE32, Live, ZxR.

          ALL of them ran just fine. Hell, Creative was one of the few to deliver stable XP 64-bit drivers for their cards.

          I had an ASUS card in-between the Live and the ZxR, the Xonar DSX. That card sounded fine, but the drivers were the most hideous, non-option things ever and the control panel was buggy as shit.

          Back to Creative on the ZxR, I’m perfectly happy. The drivers are good, the sound is pristine and way above even the good onboard S1220 of the ASUS Maximus Hero IX.

          The Creative bashing has been fashionable since the 90s and will never wane. Then again most people are now fine with onboard so it’s becoming less of a battle anymore.

            • caconym
            • 2 years ago

            I started learning to make music with a monophonic SoundBlaster 8, and my first EP on vinyl was done with an SB16 and Impulse Tracker, so I have a lot of nostalgia for the brand. I was still using an X-Fi Go USB adapter with my laptop up until Windows 10 came out and it never got a compatible driver (it worked as a generic sound device, but you couldn’t activate the graphic EQ anymore, which I was depending on to simulate different speaker setups when mixing).

            Never had any issues with any of them.

        • Klimax
        • 2 years ago

        One of them likely. (Also it wasn’t just Creative who had atrocious drivers. Maybe 50-70% of crashes was due to drivers. (I had first integrated audio in SiS chipset and later had cheap Genius sound card. BTW: Quite an improvement including better performance…)

        Also IIRC it was partially about democratization of availability of advanced sound features without needing more expensive audio HW .

        • DeadOfKnight
        • 2 years ago

        For several years everyone was singing praises about Asus audio cards, but their drivers were more atrocious than Creative’s. Sure, Creative liked to bundle a load of bloatware in with their driver “suite”, but at least they were a lot more user friendly. Today you can get standalone drivers that work just fine through Windows Update that don’t come will all the bloat.

        If you get the drivers from Creative’s site, all of the unnecessary bloat (except for the Creative update client, which can easily be uninstalled) are now optional. Creative has offered a lot more continued support with driver updates than Asus as well. My X-Fi Titanium HD runs better than new on Windows 10. I don’t know about other audio cards, but Creative is better than ever.

        The only real problem with Creative audio today is the same problem for the entire sound card market, which is that they offer very little value over premium motherboard audio, especially if you’re using cheap PC speakers. For anyone who can tell the difference (true audiophiles), they’re almost certainly better off getting external audio solutions instead.

          • swaaye
          • 2 years ago

          I use an XFi Ti in my desktop for its multichannel headphone downmix. And another in my HTPC for DTS Connect over TOSLINK. They are handy if you play old games too, for their EAX support.

          They are cheap and the XFi Support Pack by Daniel K works great for all Windows releases from XP up.

          I would probably prefer HDMI on the HTPC but my receiver cant passthrough 4k and I prefer to not run the video stream thru the receiver anyway. When you connect two HDMI connections (one for audio one for video) I’ve found Windows likes to reset the audio connection to stereo at every boot. Very annoying.

          One thing that the nostalgia nuts tend to forget though is that most PCI and PCIe hardware audio accelerators suffer bus latency problems and you get crackling noises in audio. I’ve seen this in basically every popular sound card since the Ensoniq AudioPCI. Some cards are worse than others and some motherboards have hopeless bus implementations. And sometimes another device causes the problems, like your video card.

            • DoomGuy64
            • 2 years ago

            The CA20K2 includes an embedded RISC processor to combat PCI latency, but is incredibly difficult to find models that use it anymore. The Auzentech Forte and Titanium HD are it, and maybe a few motherboards.

            No idea if the soundcore+ chips include any of that, mostly because creative hasn’t seriously documented any of the newer chips capabilities.

            That said, the OEM PCIe XFi is cheap and works near flawlessly under most circumstances. There does seem to be a small issue with Ryzen motherboards that didn’t exist with older AMD or Intel boards, in which the sound can sometimes start glitching out after long OS uptimes with an open webbrowser and repeated waking up from sleep modes. Can be fixed by using the mode switcher, which resets the driver. Not a particularly bad problem, but AMD is clearly doing something non-standard with Ryzen and powersaving modes which the XFi doesn’t like.

          • caconym
          • 2 years ago

          “Creative liked to bundle a load of bloatware”

          Don’t you dare say anything unkind about Dr. Sbaitso!

      • Krogoth
      • 2 years ago

      It isn’t coming back though. Hardware accelerated audio is dead. Let the poor corpse rest.

      Microsoft do anything expect getting rid of old legacy code and doing a much need overhaul for audio portion of the OS.

      The masses don’t care for 3D audio either. They are more than happy with simple stereo over some cheap headphones/speakers. 3D audio is pretty much in the domain of AV world.

        • DoomGuy64
        • 2 years ago

        That attitude is why we have been stuck where we are, and if not for that the market would have adapted years ago.

        That said, the “masses” and game developers DO care for 3d audio, and are finally overcoming the nihilism espoused from the PC is dead and pay $3000 for a video card shills. It’s taken a long time to recover from Microsoft’s artificial restraints, but that time has finally come, and it won’t be hindered any longer from progress by the detractors. If you wanna be the guy who only uses SCHIIT, go right ahead, but now we are getting back the option of using 3d audio. You can keep using stereo, but gamers can now start getting 3d audio, which is a game changer in FPS. Anyone who used A3D with Counterstrike knows how much of a benefit it was to clearly hear where your opponent was located, and now that feature is back with CS:GO and steam audio.

        One particular example where good sound is necessary is a game I like to play called Depth, in which high level play can only be achieved with good 3d sound. Why? You often can’t see the opponent, but you can hear them. If you can tell the location of your enemy through sound, you can counter the element of surprise and attack. Most of the high level players are using headphones or soundcards that have good 3d virtualization, while the stereo players simply get slaughtered.

        “Hardware accelerated” audio is not dead. Software audio is dead, especially since Microsoft has not been improving it in favor of xbox. Steam audio is CLEARLY bringing it back to the PC, and keeping a level feature set for software and hardware, with hardware only getting features that can’t be done efficiently in software, like raytracing, and most likely will be expanded past AMD later so everyone can benefit. Microsoft may have been “correct” in that problems existed with their old driver model, but they were entirely in the wrong for not compensating for the loss of 3d, and also for not considering a new driver model capable of acceleration. Now that Steam audio is competing with xaudio, Microsoft no longer has control of the market, and will simply be replaced if they don’t adapt. There is no amount of shilling that will ever put the genie back in the bottle now that it escaped, nor will Microsoft be able to stop a decent 3rd party API. Xaudio is dead for lack of improvement, and so is the driver model argument, since steam audio doesn’t compromise it.

          • Krogoth
          • 2 years ago

          Hardware accelerated audio died because mainstream CPUs got powerful enough to effortlessly handle DSP processing. We aren’t using single-core Pentium IIs and K6s anymore. A lowly Kaby Lake-based “Pentium” and Ryzen 3 can breeze through audio processing. Modern audio cards are really just overglorifed DACs.

          Blame the stupidity of Creative’s management for the “death of 3D audio gaming”. They had a stranglehold on most of “3D gaming audio IP” and did nothing with it for years. Microsoft had absolutely nothing to do with it. They just got rid of old-ass legacy coding (ISA-era stuff) which EAX relied on in order to work correctly. Creative saw this an opportunity to force its Audigy/Live! users to upgrade to X-Fi family which used “ALchemy” wrapper to get EAX working under NT 6.x. Creative originally locked down ALchemy to the X-Fi family but some fellow “reverse” engineer it get ALChemy work for Audigy family. This end-up creating a massive clusterstorm that eventually forced Creative to offer ALchemy support for their older hardware platforms.

            • DoomGuy64
            • 2 years ago

            Look at that first picture in the article and repeat that first statement. That’s patently false, and you are a victim of Microsoft’s propaganda to screw PC users into buying xbox, which has had superior audio support ever since Vista.

            The second statement is also false for multiple reasons:
            1: Creative MASSIVELY upgraded EAX with the X-Fi. “Did nothing” is a blatant lie.
            2: EAX is a direct extension of direct sound 3d. Microsoft killed it, not Creative.
            3: Waah! my tnt2 doesn’t support dx12! Yeah that’s right, your old soundcard doesn’t support new features because those features are reliant on hardware support. EG: EAX4 cards do not support EAX 5.
            4: All PCI soundcards including the original XFI had massive problems with motherboards/pci latency and needed to use PCIe to have any level of decent stability across platforms.
            5: Yes, Creative did deliberately not give ALchemy access to older cards, which had multiple issues that Creative did not want to support because they were EOL. EG: The Audigy chips had terrible hardware resampling and other limitations, which were completely unfixable in drivers. I don’t hear you complaining about AMD not supporting VLIW4, hypocrite.
            6: Creative did eventually give Audigy users access to ALchemy for $10, so that statement is still inaccurate. However, the point still stands in that those cards were too old to be decent soundcards in the Vista / x64 era. The chips themselves did not support x64 / 4GB+ memory (required memory hole support), had horrible hardware resampling, were PCI in the age of PCIe, only supported EAX4 in hardware, and worked FAR better in software mode, because software mode bypassed all the hardware limitations.

            This is yet another one of your hypocritical statements, because out of one side of your mouth you argue pro-software modes, yet complain about the Audigy, which you know or admit NOTHING about it’s actual issues on newer hardware and OSes. Audigy users were actually better off skipping all hardware modes, and instead using straight software mode with OpenAL soft, or Creative’s MB3 software which also does these features in software. It may have been bad PR, but Creative had every right to drop hardware support of the Audigy, which they only dropped features that were broken and outdated. The Audigy was simply too old to properly support everything on newer hardware / OSes, so using hardware modes created more problems than it solved. Ignoring those issues is basically a logical fallacy type of argument. Lying by omission. It’s also a red herring, because it has nothing to do with the general discussion.

            Creative may not even be able to do raytracing with their current hardware, cue the audigy type complaints, but they should be able to accelerate the base level of effects here, and any level of support is better than none. That said, I think TAN is more flexible and will most likely become the audio processing standard in the new API, but it would also be hilarious if Creative gets in there if only to piss off the irrational haters. The tears of unfathomable sadness are delicious.

        • LostCat
        • 2 years ago

        Windows 8 and 10 have a hardware audio path, though I don’t know much about it. The details available last I checked are a bit light.

    • madmanmarz
    • 2 years ago

    Excuse my ignorance, but at the moment does this do anything for the end user? What would someone like me get out of this (I have a Vega card and a VR headset) and what exactly do I download and install?

      • enixenigma
      • 2 years ago

      It should give you a more immersive audio experience in compatible games.

      Off topic, but which headset do you have, and how do you feel about Vega’s performance in VR? I’ve been mulling over getting into VR, but I know that Vega doesn’t play as well in VR due to lacking the simultaneous multi-view tech that Nvidia cards have.

      • MOSFET
      • 2 years ago

      At the moment, it does about as much as the original TrueAudio on the Bonaire GPU.

      • madmanmarz
      • 2 years ago

      I’m on a Vega 56 and on some games it’s about like a 1080 and on other games, less. I have the Samsung Odyssey HMD.

      As far as the audio do I need to install anything other than the latest drivers or do I need the Steam Audio and all that? There’s a few links.

        • DoomGuy64
        • 2 years ago

        Just drivers, game devs need to implement the API / updated engines for it to work. CS:GO supposedly uses it, available in options, other games unknown, but new UE4 / FMOD / Unity games should start supporting it soon.

        • enixenigma
        • 2 years ago

        Nice. I have the 64 and I was looking at the HMD Odyssey as well. Now to explain this to the wife…

    • enixenigma
    • 2 years ago

    I think you missed the Vega cards in the list of compatible GPUs.

      • Shobai
      • 2 years ago

      So they did.

    • DPete27
    • 2 years ago

    Great, another 20GB of audio file size on each game.

      • Vaughn
      • 2 years ago

      Time to get off that AOL 56k connection 🙂

      • caconym
      • 2 years ago

      These are real-time filters. They do not increase audio file size.

      If it were an open standard it could decrease game audio size, because storing e.g. a sharp, dry gun shot sound effect is much cheaper than storing a gunshot with a seconds-long reverb tail baked into it. You could just generate the reverb at run time, and tailor it to the current environment and player position.

      Nobody’s going to do that if it’s AMD-only though.

        • DragonDaddyBear
        • 2 years ago

        The good news is it’s open source.
        [url<]https://techreport.com/news/30538/amd-open-sources-vr-audio-and-video-acceleration-sdks[/url<] That said, NVidia probably won't make it happen if they can't milk tons of money out of it, even if VR, and games in general, could benefit from a new widely-used audio spec.

          • derFunkenstein
          • 2 years ago

          I fully expect Nvidia to add audio to GameWorks, but I’ve been expecting that for a while and it hasn’t happened for whatever reason. So maybe I’m wrong.

            • DragonDaddyBear
            • 2 years ago

            IF they do, they will charge developers money to use it and will run like poop on AMD.

            • derFunkenstein
            • 2 years ago

            Indeed that is the fear.

            • stefem
            • 2 years ago

            It’s already available and it’s free like the rest of their GameWorks library…
            I’t called VRWorks Audio and it use path tracing to determine sound reverbs and by tracing which surfaced (of which developer can specify the material proprieties) are hit it then apply a deformation to the reverb.

            • LostCat
            • 2 years ago

            [url<]https://developer.nvidia.com/vrworks/vrworks-audio[/url<] ?

            • DeadOfKnight
            • 2 years ago

            It will probably be a thing as they run out of ideas for new features to tack on. Proprietary features has been a big part their marketing strategy ever since they bought PhysX.

    • the
    • 2 years ago

    Hmmm…. I wonder what professional level features are available for True Audio Next, in particular advanced 3D spatial audio. These could be an interesting substitution for DSPs or audio focused FGPAs in the professional space.

Pin It on Pinterest

Share This