Valve Announces Steam Audio SDK


Low-latency 3D audio for hundreds of sources, even on mobile. Experience physics-based sound propagation on a wide range of hardware.

Pretty excited about this. Audio requires a lot (A LOT) of time and effort when making any kind of application, but in VR it takes on a whole new level of importance and consideration, where you’ve moved from reality approximation to reality synthesis. Making things like physics based reverb and occlusion a drop-in endevor is going to seriously speed up development and increase title quality, especially for single player narrative experiences. Baked audio is pretty intruiging as well.

Those of us who have participated in the recent stress tests have experienced first hand how quickly audio from multiple sources in a 3D environment eats up your bandwidth and processing budget, which compounds significantly when you add An Internet™ in the middle. Steam Audio is promising low latency for hundreds, even “thousands” of sources. I’d love to see this thing run through it’s paces and see what kind of performance comparison metrics it gets against other popular audio solutions.


Umm, High Fidelity already has HRTF spatialized audio, and first order ambisonics. Its extensible and distributed audio mixer technology ensures a capped bandwidth audio no matter how many people are in a scene. There is reverb, not sure if they got around to implementing virtual ‘wall’ physics based reverb or sound wall occlusion, but I’d guess it is on the list.

Audio in HF is one of those premier and well done features.


The Funny thing is, High Fidelity has most of this already implemented. Its got everything from reverberation settings (via zones and stuff you can do with reverbTest script that allows you to tweak all sorts of values) and to bandwidth use. It actually quite crazy how little bandwidth audio uses within high fidelity due to the compartmentalization done by the audio mixer: Most of the bandwidth went to loading avatars and streaming avatar data than audio, which is currently being worked on.

Only thing Hifi doesnt have is the raytraced audio and higher order ambisonics.
Personally I dont think Steam Audio may actually out perform what High Fidelity has now in terms of bandwidth, unless they have much less quality in the audio stream or does the same compartmentalization, but it could give hifi a run of its money, as Audio has been to date High Fidelity’s strength compared to its competitors.

Otherwise, its a fantastic thing that they are pushing for such and I am prolly going to play around with it in unity.


Calls bulls*&- t. On this unless headphones suddenly get more than 2 speakers then what u have is stereo + marketing spin


I was under the impression that this subforum was for general VR discussion?

Audio is tackled well in HiFi, but I work on additional VR projects outside of HiFi. The SteamVR SDK (and subsequently the fantstic, community-made VRTK that builds off it) have made creating things in VR a much more rapid affair. I’m happy to see Valve take a stab at audio and see what it can do. If they can make more advanced audio concepts like higher order ambisonics, raytraced audio, and audio baking something that’s more readliy accecssible, then the industry benefits as a whole.


Yes, this category is about VR developments in general, which is why I liked your post. But when you implied all that was missing in HF, I just corrected that misconception in my response. BTW, the audio bandwidth to your interface client does not climb based on the number of people. The audio mixers bake the audio from all the sources, and normalizes the bandwidth.

Any increasing bandwidth you experienced was either from entity downloads or the presence of more avatar movements. And, even that is getting normalized which is what the stress tests are all about. They have made changes to optimize the delivery of avatar movement data when large numbers of avatars are present to avoid bandwidth pileup.


It is I wasn’t getting at your post more the concept in general.that u can create surround sound from 2 speakers using clever processing that it is more than stereo.