AMD open-sources VR audio and video-acceleration SDKs

AMD is adding source code for two of its SDKs—TrueAudio Next and version 1.3 of the Advanced Media Framework—to its GPUOpen developer resource today. Both SDKs are available free of charge to interested developers under permissive MIT-style licenses.

Unlike the original TrueAudio block in some Radeons, TrueAudio Next runs in software, and it's built atop AMD's Radeon Rays ray-casting engine. Together, these programs could provide more accurate simulation of the way sound waves interact with physical spaces to deliver a better sense of immersion in VR applications. TrueAudio Next also takes advantage of the Compute Unit Reservation feature in AMD's Polaris GPUs to ensure that audio-processing tasks can be completed with the low latency that VR demands. If all this sounds familiar, Nvidia has a similar tool for developers in its VRWorks Audio toolkit, which uses the Optix ray-tracer to promise similar fidelity.

The second SDK that AMD is adding to GPUOpen today, version 1.3 of the Advanced Media Framework (AMF), opens up the Video Coding Engine and Unified Video Decoder hardware in AMD GPUs to developers. It wouldn't surprise us if the developers of Open Broadcaster Software (OBS) take advantage of this SDK so that the popular application can accelerate streaming tasks on AMD GPUs. That's not the only potential application for this code, though. AMD suggests that developers of wireless display, remote desktop, video editing, transcoding, and playback applications will find AMF 1.3 useful, as well.

The source for both TrueAudio Next and AMF 1.3 is available today on GitHub.

Comments closed
    • wingless
    • 3 years ago

    AMD GPUs are best sound cards that I’ve never used because I don’t have an HDMI A/V receiver and a surround sound system hooked up to my damn computer. AMD should have a breakout card with 3.5MM jacks and optical….

    • Kretschmer
    • 3 years ago

    As much as I love open standards, I have the unfortunate suspicion that these are things that AMD simply no longer has the resources to pursue.

      • tipoo
      • 3 years ago

      I get the feeling as well, though I also think it’s good to prune off anything not making them money and funnel everything into GPUs and CPUs.

    • ronch
    • 3 years ago

    More free open sauce stuff from AMD!! Yey!!!

    Profits must really be pouring in!!! 😀

    • ozzuneoj
    • 3 years ago

    Realistic audio effects died with Aureal in 2000.

    A sad time that was…

    Been waiting 16 years and no one has managed to create an audio standard that could do to any game what A3D did back then.

      • Acidicheartburn
      • 3 years ago

      I never used Aureal, but I did use Creative’s X-Fi back in the day when there were a few games that actually managed to support it. I remember vividly how much better it made Battlefield 2 sound. You could hear the artillery strikes coming before the hits landed, allowing you a little time to vacate the area.

      I realize the problem with these proprietary standards that require separate hardware or a specific vendor in order to use, but I’m sure much of what Aureal and Creative were doing is plenty capable of being done in software or something else with today’s modern computing power.

      • ronch
      • 3 years ago

      If you ask me, audio effects from a ~$2 Realtek audio codec these days are far more realistic than the audio effects produced by ~$200 Sound Blaster cards from the 80’s and 90’s. Maybe I’m not as fussy as audiophiles are but I’m amazed how cheap technology is today. And really, Realtek audio isn’t bad at all. Just use decent speakers.

        • ozzuneoj
        • 3 years ago

        There’s a difference between well made sound effects (.wav files) and environmental\spatial effects. Modern games rarely make any effort to do anything more than have a sound file play at a given volume based on its distance to you, with the sound coming from the speakers in such a way that you can generally tell what direction it is coming from. Walls, materials, room size and the “size” of the thing making the noise have no bearing on anything. Someone running down a hallway on the other side of a wall from you generally sounds the same as someone running past you in the same room.

        Aureal’s A3D made some great progress in implementing some of the things above… at a time when most mortals were still rocking gaming systems with 64MB of RAM and CPUs in the 400-800Mhz range. There are demos from Aureal that certainly have fairly low quality sound effects(low sample rate wav files) by today’s standards, but the fact that they change pitch depending on their speed (doppler effect) and a sound can (for example) be heard through an open doorway rather than through a wall… that’s what made it impressive.

        Aside from game developers manually implementing these features per-game or only using them in very specific situations (or even just having pre-baked effects in the audio files, to be used in one scene\cutscene), you just don’t hear these types of effects in any widespread way. With the massive processing power we have available now, you’d think that games would be modeling actual real reverb, reflections and other things to almost create sound effects in real time… but we’re hardly any farther along than we were in the early 90s.

          • ronch
          • 3 years ago

          As far as sound effects go, Realtek is realistic. But they’re no good for environmental effects and sound positioning, which is what I understand you’re saying now.

            • brucethemoose
            • 3 years ago

            Realtek doesn’t have anything to do with that.

            It takes the multichannel output from the game engine and pipes it to a DAC. Realtek codecs have no information about anything within the game itself, all they do is tack on simple effects like resampling, an equalizer or downmixing the audio… Realistic, maybe, but that’s because it isn’t doing anything to the audio.

            Fancier audio drivers (Creative, Asus, Razer) take the raw multichannel audio and put it through an HRTF algorithm for headphones, but even that is kind of iffy. It sounds MUCH better when the game engine does it.

      • meerkt
      • 3 years ago

      Anyplace to find info on the current state of things, what games support what, what hardware is needed? I recall reading something about Ambisonics in the DiRT series of racing games. What does it mean in practice? How does one take advantage of it?

      I’ve never had more than 2 speakers. And for stereo at least, I could never decide if EAX was better, the same, or worse than whatever other sound options the game offered (DS3D, Miles whatever, etc.).

      • DoomGuy64
      • 3 years ago

      A3d used wavetracing on top of HRTF. All anyone has to do is implement wavetracing again, and we’d have our good 3d audio back.

      “Raycasting” just sounds like AMD is doing the same thing, only under a different name. Of course, the games will then have to support it to work.

    • brucethemoose
    • 3 years ago

    [quote<] It wouldn't surprise us if the developers of Open Broadcaster Software (OBS) take advantage of this SDK so that the popular application can accelerate streaming tasks on AMD GPUs. [/quote<] Theres already an OBS branch that uses AMD's AMF API for GPU encoding. I think the only reason it wasn't merged into the main branch already is a licensing issue. In fact, I wouldn't be surprised if a chat with the OBS devs is behind the decision to open source the API. I know they established a channel of communication with AMD a few weeks ago.

    • Acidicheartburn
    • 3 years ago

    I would really love to see stuff like this take off. I’m consistently disappointed with the audio implementation in most games. A few notable exceptions are Battlefield: Bad Company 1 and Half Life 2. Directional sound in most games usually leaves much to desire, and echo/reverb is typically half-baked or nonexistent. For someone who values great sound just as much as great graphics, I’m eager to see game developers (or game engine developers) spend a lot more effort on more realistic audio implementation.

      • brucethemoose
      • 3 years ago

      An HRTF option in-game would be nice too.

      I’m not the only one who games on stereo headphones, right?

        • Acidicheartburn
        • 3 years ago

        I have a Logitech 5.1 surround system I use for gaming, but when I need use voice chat I use stereo headphones with a separate desktop mic, so nope, you definitely aren’t the only one.

      • LostCat
      • 3 years ago

      EA at least seems to be fond of GPUOpen so maybe, but they’ve shown little interest in audio tech recently.

      I’d expect it to need decent marketshare or MS support before they jump on it.

      • DarkUltra
      • 3 years ago

      I found Battlefield 3 and Crysis 2 to also have excellent software audio. Crysis 3 was not so good.

    • chuckula
    • 3 years ago

    [quote<]Unlike the original TrueAudio block in some Radeons, TrueAudio Next runs in software, [/quote<] Guess what hardware block ain't going to be in Vega? [Edit: I did a little research and it looks like Polaris has already dropped TrueAudio and is implementing similar functionality on the shaders just like this TrueAudio Next]

      • Wildchild
      • 3 years ago

      Makes sense. The only time you ever really heard about TrueAudio was when it first came out and I vaguely recall a few niche applications that used it, but beyond that nothing was ever really done with it.

        • chuckula
        • 3 years ago

        It’s actually a more flexible idea to implement it more generically with shaders anyway, especially on a shader-heavy GPU design.

      • EndlessWaves
      • 3 years ago

      Blame the developers, if they don’t use it then of course AMD is going to drop it in favour of something more useful. Die space ain’t free.

Pin It on Pinterest

Share This