Has anyone tried to fully generate sfx for a game? I’m talking about something that does not try to emulate retro feeling.
Basically, I have impression there would be a node-based audio generator just for this kind of thing, but I’m not seeing it anywhere on the web. There are experimental projects, one fairly old program for Linux, and mentions some ipad-based program. And that’s it.
Am I missing something?
I’ve made some stuff. Soundcloud
I’d love to spend more time on it but too much going on right now.
Which software were you using? Because I’m looking for generators.
I made my own. I basically fiddled in Audacity to get an idea of what I wanted, and then went and wrote some stuff inside Unity. It’s all generated from white noise in Unity, using low-pass filters mainly.
Hmm… I’d rather not get sidetracked into writing a custom audio generator at this point :-\
Well in case you or anyone’s interested, for the explosions and gunshots I basically did what you find in this youtube video using a butterworth filter and simple linear falloff etc. (I know, the video is almost unwatchable but I already knew my way around audacity).
Isn’t what you’re looking basically a Synth? There are a ton of VST ones.
Synth1 vst is pretty straightforward and free. You need a basic VST host though (or a DAW). No free ones come to mind though.
Ah, also try Audiomulch. Which may be more in line with what you’re describing.
There is no “substance designer equivalent for audio” as far as I’m aware. GTA V uses some hybrid/procedural approach I think, but that’s inhouse software iirc.
https://www.youtube.com/watch?v=L4GuM15QOFE
Like @AcidArrow said, classic sound synthesis using VST plugins is the closest you’ll get at the moment (that I’m aware of), but I would consider it approaching the realm of “entirely impractical”. I say that from the background of probably having spent over 100 hours trying to synthesize certain sounds. It’s hard!
I would approach this with an array of basic samples to start from. I would then make a volume router that would assign values from 0-1. I would then grab any number of DSP algorithms and drop each into a function and expose any tunable or variable parameter such as frequency such as fourier transforms, reverberation times, digital delay loop timing and falloff per sample and normalize their ranges from 0-1. I would then target those parameters with gameplay data so the soundscape was procedurally generated by the gameplay dynamics and responsive to player input. Perhaps create presets to Lerp between with those outputs sent to a further volume balancer. It would be reasonable to implement initially and can be grown as more useful DSP algorithms are tested and implemented. The choice of starting samples would lead to different outcomes based on same settings and eliminate the need to generate the initial sounds.
Well, here goes idea of programmer-friendly Synth :-\ (for some reaosn every vst plugin I saw seems to be centered around “box with knobs” paradigm) “substance designer for sounds” is exactly what I’d like to have.
Sounddesign is a really deep topic and the “box with knobs” approach already allows for more complex setups than most people can remotely make use of (myself included). You usually don’t just use one synth or effect, in a proper DAW you can build complex setups and for example use the amplitude from one signal to drive parameters on another channel. If you want to get even more “low-level”, then maybe NI Reaktor is your thing?
https://www.native-instruments.com/de/products/komplete/synths/reaktor-6/
You can build synths from scratch with this. The product page at a glance looks like you just use more “building blocks with knobs” in a simplified fashion, but I believe that’s a recent feature in v6 that they are promoting now. I played around a bit with Reaktor 5 some years ago, and you really can go low-level with this. NI built some of their own synths with it and there is a big user library where you can download stuff.
https://www.native-instruments.com/de/reaktor-community/reaktor-user-library/
I think Razor was made entirely in Reaktor:
https://www.native-instruments.com/de/products/komplete/synths/razor/
I’m not sure if it has any code based programming features or not and couldn’t find out with a quick google because “programming” often seems to mean something else for audio people.
Is it worth it? Imho no, unless you enjoy the process of playing around with it and don’t need to “get” anything else out of this. If you are into the tinkering aspect, cool, go for it! If you just want audio for your game, I suggest to do freelance work and buy high quality samples. That will be more time effective (and most likely higher quality too) by orders of magnitude imho.
You can use sunvox - WarmPlace.ru. SunVox Modular Music Studio
It’s free and very powerful. Seem intimidating at first but it’s actually fairly easy to use.
Some tuts on making sounds with it:
if you want a simpler solution, you can try http://www.bfxr.net/ which is pretty good too.
Full procedural SFX is a complete waste of time. You will be writing tons of complex, experimental code while you could spend that time browsing freesound.org and get better results 100x faster.
Hmm… I see node graphs on the screenshots. I’ll check it out, thanks.