Hi all. This isn't so much of a specific question, but more of a general discussion about a topic...

I've recently started to read a little about Procedural Audio in game design, after seeing an article about AudioGaming tools on Designing Sound. I intially mis-interpreted the term and thought it meant procedural as in 'proceeding' or 'evolving', but it's actually as in 'procedure'; from my understanding it's a way of randomly generating/synthesising 'realistic' sounds (rather than random mish-mash of sounds etc.) using a collection of low-budget (in processing/memory terms) algorithms.

Examples I've heard of so far are for the creation of things like weather or energy weapons where the sounds heard either evolve uniquely over time for a continuous ambient effect such as the sound of rain, or vary slightly each time they're created, such as when firing a pulse rifle, and so on.

When implemented well using such techniques and tools can result in a much more natural and diverse 'sound' to a game, where many effects will never be quite the same, which in many cases can be far better than a sound engine firing off a randomly selected sample from a (finite) bank and slightly altering it's pitch and gain.

Of course the use of such tools seems to be restricted by how well certain types of sound can be (realistically) synthesised, and memory and processing power consumption.

Does anyone here use procedural tools such as those by AudioGaming in game sound design, or indeed other applications? What kind of things have you used them for, and how do you go about implementing them into the game/sound engine etc? And please feel free to correct me if I've misunderstood anything!

4 Answers 4


It's a growing field of research and there's clearly much interest into procedural audio, because of its interactive/dynamic and synthesized nature. P.A. is mainly proposed to be a great solution for anything that's repetitive and for creating common "boilerplate" sounds such as footsteps and ambient noises that consume time, which the busy audio designer could use for more important sounds. But as Andy Farnell demonstrates in his book "Designing Sound" and on his site, and as numerous papers describe, sound synthesis can be quite versatile.

I wouldn't speak out about the use of computational resources, because P.A. hasn't seen much real-world use yet. In some contexts it's seen efficient compared to audio samples in terms of computational resources, because P.A. doesn't consume as much memory and code doesn't need much storage space. The sound model's accuracy and usage of computational resources can also be tweaked. And there's an increasing amount of computational power available in multi-core processors.

In terms of integration, there are barely other tools than APIs. AudioGaming is working on the front line in integrating pre-packaged P.A. solutions to existing and commonly used audio middleware so they can be utilized by audio content creators with ease. If you're familiar with Pure Data, there's also Libpd (http://libpd.cc/), which enables the use of Pd as an "audio engine" and real-time synthesizer in small games/apps.

There's a "Procedural Audio Interest Group" here: http://www.procedural-audio.com/ and forums http://www.procedural-audio.com/forum/


Hi, AudioGaming developer here (which by the way is French, so please forgive my poor english skills) !

PA has been an ignored tool for a very long time, and because of that we sometimes have to struggle a bit against long-established habits, creation pipelines and existing content. But it really just is one more tool, not really designed to replace everything at once but to expand possibilities.

As mentioned earlier, Andy Farnell is currently one of the key people for this topic, as is Nicolas Fournel who maintains procedural-audio.com

As you wrote it, one of the most proeminent problem with PA is computational resources ; not really because of the required horsepower : modern hardware could handle most cases perfectly if there was no display...but mostly because of what the audio part gets usually allocated - for most common PC/consoles games it is below 5% CPU. Displaying shiny things just eat too much, and audio is always the first domain being crippled.

That being said, this is currently changing because available CPU power increases way faster than available memory size - memory is expensive, low latence memory even more. For instance Microsoft demonstrated a few interesting methods in their commercial game Crackdown 2, and last year Wwise released SoundSeed, a "whooosh" sounds tool - AudioWind being OF COURSE way better, as you could expect ;)

As for PureData, this is a great prototyping tool - Andy Farnell's works rely mostly on it and his patches are freely available. However I don't really believe it to be suitable for real, in-game usage - especially on console. As Amaury stated in the Designingsound.com interview, we use it a lot for quick sketches ; but we had to develop our own optimised tool suite to easily target different platforms/architectures/needs...


If you are interested in procedural audio, take some time to peruse Andy Farnell's website:


One of the best resources on the net. His book "Designing Sound" is also fantastic.


Dylan Menzies has done pretty interesting work in the domain as well:



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.