Jaime here, and today I’m going to talk about the sound design process for lil’ Sparky’s roars, grunts and general vocal noises. If you’ve not seen him before, Sparky’s our test monster that’ll not be in the game. But we love him all the same.
Sound design, especially for vocal work, requires a few things. First is a decent recording space and set up, and not a jury rigged vocalbooth made out of duvets, chairs, clotheshorses and various other sound absorbing things…
The other is the ability to record yourself being a bit of an idiot and not getting all strung up and panicked about it, which sounds easier than it actually is. Recording sounds is a bit like a performance, you have to think “what do I want this to sound like?” before even going in front of the mic and then try and hit that, or something that you can then manipulate into the final article.
Case in point, here’s the “spotted” roar as it was recorded:
It’s…. something. Usable eventually. The next stage was pitch shifting, lowering it down so that it resembled something a bit more monstrous.
The Holy Grail of sound design: transpose pitch
Generally speaking when I record something I usually mess around with the pitch just to see what else I can get out of it, even if I’ve already got the result I wanted from it. I recorded some debris sounds, grains being dropped and swirled around a metal container. I didn’t really get anything usable out of it, so dropped the pitch down 36 semitones and tried that instead. Here’s a before and after for you:
It went from being something admittedly crap to a fairly usable atmosphere. So even if you’ve got something that you might consider a mistake or useless always take a quick 5 minutes to mess around with it! Anyway…
The next step with the roars was breaking it down into 3 components: the main “body” of the sound, the lows and the highs. This is pretty easy, just duplicate the track 2 times then use EQ to filter out the ranges.
Top to bottom: The three tracks, the low roar EQ, the high roar EQ
The main reason for this is to allow me to fine tune the different elements of the sound. There are psychoacoustic properties in low frequency sounds that trigger a fear response in humans, so as I was working on a monster for a horror game it made sense to separate it out. The highs were also separated out so I could fine tune the air and breathing noise, especially in quieter grunts. Here’s an example of each element with the roar, then all those layers combined:
Now for the effects chains. They’re mostly similar, but I’ll highlight a few differences in each beyond the EQ.
Post EQ it goes: Vocoder > Saturator > Chorus > Compressor
First thing there is the vocoder. In Ableton you can set the carrier wave to be another audio file, such as lion roars or whatever you fancy. After some experimenting with those I decided to use the vocoder to add a bit of grainy-ness to the sound, so used white noise as the carrier wave and kept it fairly low in the mix. It also sucked a bit of air out of the sound, giving the other layers a bit more presence. Here’s a before and after:
Next on the list is the saturator. It might seem a little counter intuitive to have something boosting the high end on this track, but it brings out the vocoder a bit more without messing around with EQ. A little lazy but it works, and again it’s blended in a subtly.
Next is the chorus and compressor. The chorus is simply to make the roar a bit bigger, fattening it up to match the size of the monster, and the compressor is set to have a fairly fast attack and longer release. This is based off a recommendation I was given when dealing with dialogue, as it allows it to pop in quickly but decay fairly naturally. Here’s the whole of the main roar:
Next is the low roar element.
EQ > Saturator > Chorus > Multiband Compressor
The saturator and chorus were used for a similar reason to those mentioned above, but the multiband compressor is probably standing out as the odd thing here. If you’ve used them before you’ll probably be wondering “But you’re only using low frequencies, why use a multiband?”. Well, despite the harsh EQ there the sound still stretches beyond the 1kHz range, and I was happy with the way the EQ was set up. a normal compressor would affect everything in the sound, but a multiband lets me compress different frequency ranges (or bands, hence the name). The result is a bit of a boost to the low end, while everything else is kept balanced. Here’s a before and after:
Finally there’s the high end element. There’s nothing too new going on in here, although the compressor has a shorter release compared to the main body.
EQ > Saturator > Chorus > Compressor
And here’s how it sounds.
Now we can stick them together and…
It’s not bad! But missing a little something. I also added a few pitched down animal roars (leopards in this roar only, and elephants on all of them) fairly low in the mix. Here’s how each sounds, and then under the roar itself.
A bit better, but still doesn’t sound quite right. The sounds need a bit of space. While most game engines will have audio reverb zones in them, I still think that it’s important to use them as part of the sound design process as well. Not every reverb has to be a cathedral!
I used a spring reverb via Ableton’s convolution plugin, partly as I like the way spring reverbs sound. It’s a subtle thing but I just think they add a little bit more interest, so no real rhyme or reason! Here’s the before and after:
The next send is a little more interesting…
Ring Modulator > Chorus > Reverb
Ring modulators are… strange. It would take me an age to explain them so go to Wikipedia and read up on it (Wikipedia). They can sound really nice when tuned properly, or absolutely nasty when the frequencies used are out of sync. If you ever get the chance to mess around with them, especially on a synth (Korg’s MS-20’s ring mod is gorgeous) then do so! The chorus and reverb are there to fatten out the sound a bit more. But rather than prattle on this is probably best left to your ears to judge:
Last but not least, mastering:
EQ > Limiter
This is a pretty basic chain, with a little corrective EQ going into a limiter. Sometimes I stick a multiband compressor here, but it seemed fairly pointless in this instance given how I’d broken down and mixed the sound. Here’s the final result:
Tadah! That’s a a monster for sure. Not bad considering it started life as a guy roaring into a mic surrounded by blankets, eh?
Of course, context is everything. So imagine stumbling on this dark, dilapidated ship like this:
Only to hear Sparky spotting you from behind. Fun times.
That was a bit long, so hopefully it was worth it! If anyone wants to ask anything the drop a comment or give me a shout on Twitter on either @teamjunkfish or @speedyjx :). Here’s a playlist of the sounds too so you don’t need to scroll though the page.
Over and out!