Audio roadmap and APIs

It would be great to know what plans, if any, are in store for the audio system in UE4. The lack of low level audio APIs (unless I’ve missed something so far) is extremely limiting as there is no easy way to implement custom DSP. With so much possible on the visual front, the audio system hugely limits creative possibilities of linking audio + visuals. It would be great to know if there are plans to open up and modularise the the audio engine. Until then, there is no choice but to use other game engines that offer such flexibility — which is disappointing. I’m sure I speak for other audio programmers and designers!

Thanks

Right now we don’t have any other choice than to use middleware, like fmod and wwise - it would have been great if the audio system was more of a first class rather than current implementation. If we had a simple low level audio API with sample buffer access, then Epic wouldn’t have to worry about audio since we could implement our own DSP functions and make them available on the future marketplace.

The short answer is: We don’t currently have any near term plans to significantly change the audio API.

The long answer/explanation is that for the most part we rely on the hardware to do the heavy lifting in our audio system and at the moment we have something like 7 different audio implementations for the different platforms. We definitely have some inconsistencies as it stands (which we are of course addressing) and adding/exposing new features in a way that is consistent across platforms is a challenge.

Our long term wish is to move to our own software mixer to get consistent behaviors across all platforms (or at least much closer) and new features require one implementation instead of 7+.

The trouble is that to do our own software mixer requires us to first thread audio handling and both of those tasks are big projects that it is a bit challenging to justify when wWise or fmod are both good options available to people who need the functionality.

I really do wish I had a better answer than (as do our in house audio engineers) but unfortunately is where things stand for the moment.

I think is reasonable, but some time might be spent on docs and such for integrating those third party solutions, so that people don’t stumble on audio due to it. Maybe there are some? I didn’t find them but I wasn’t scouring the docs for them, either.

Not exactly what I was hoping for :slight_smile: Thank you for the information - looking forward to the long term solution. The current implementation will always have some inconsistencies because of the different implementation of ex. reverberation in the XAudio2, Coreaudio, OpenAl etc. libraries.

We just need to implement our DSP code (convolution, environment simulation and binaural processing) in wwise/fmod.

You should bring back hardware audio support, now that Windows 8 supports .
Not that Windows 8 is an extremely popular system, but it’s more popular than it was, say, a year and a half ago.

also makes me sad. As a game consumer (and not developer) I feel audio has taken a huge back seat to graphics.
I’d love to see sounds occluded/reflected based on the actual level geometry/materials and player position vs. the generic approximations that have been in use for the last 10-15 years. You don’t need to be an audiophile to appreciate ‘real’ sounding sound or have some crazy sound system, either. Adding a flag or two to the materials for audio properties when defining materials in the material editor and letting the engine handle the rest just like it does for graphics would be a beautiful solution!
Unfortunately, that means some awesome programmer somewhere has to make audio engine a reality.
I hope becomes a long term goal at least…it’d be a shame if it didn’t!

Good to know that FMOD is free for Indies under 100k USD: http://www.fmod.org/sales/

If I’m right, stuff was supported back in Unreal 1 and UT99 with A3D hardware.

To an extent some of it was, and unfortunately it regressed a lot when Vista came out. Solutions like A3D simulated room environments but were still just DSPs, they didn’t truly take geometry into effect. Check video and video out for some awesome examples of how audio could be handled in the engine. is the ‘next gen’ kind of stuff I’d like to see implemented at some point (not sure if that was what the original thread creator was envisioning or not, though).

EDIT:
I stand corrected. EAX used DSPs but A3D actually did take geometry into effect.

Which is why I think kind of is so desperately needed to be implemented. Since A3D support went deprecated with UE2, Unreal Engine, and nearly (if not all) game engines out there, ended up being stuck behind in audio technology.

BTW guys, the FMOD developers mentioned contacting support to get access to their (apparently ready soon) early UE4 integration, here: UE4 Integration? - FMOD Studio - FMOD Forums

Just a reminder that Wwise has been available in UE4 since the very beginning, and we have releases for all versions (we are currently working on a UE4.4 integration, which will be out soon).

The integration is available here: https://github.com/audiokinetic/WwiseUE4Integration

Also to add another example here, FMOD is releasing a plugin in early September for UE4, and will have a pre-release version available soon if you sign up for it.

Here is the quote from my forum question about availability last week:

Keep an eye on page in early September for the plugin:
http://www.fmod.org/download/

Edit: Sorry n00854180t, noticed right after posting you had replied… whoops! :o

Both of those tools are great, no doubt. But I think what we were referring to are less of the developer facing tools and more of how the actual audio processing stack works in the engine. Between those 2, FMod comes closest because it supports Astound Sound. Unfortunately, Astound Sounds is still just a DSP processor, and it’s main purpose appears to be to locate an audio source in 3D space using 2 speakers. doesn’t address occlusion/reflection of objects in 3D space (watch the 2 videos in my post above to see what I’m talking about). I don’t think FMOD or WWISE consider geometry when handling the audio (correct me if I’m wrong, though).

What does Oculus VR’s announcement about licensing RealSpace3D’s audio stack mean for topic? Will it be integrated with UE4 audio somehow, or work as a plugin?

If isn’t planned, IMO we should petition Epic and/or RealSpace3D to add UE4 support for RS3D.

We at VisiSonics/RealSpace3D are working on a Wwise integration to Unreal.

1 Like