How I Built a Sample Pack In An Afternoon

To set the scene, put yourself in my approximate position: it’s been 4 months since someone asked you, with minimal direction and no pay, contract, or due date, to make a sample pack for their game mod, with the promise of rev share later. Not a great idea, and you barely understand what samples are wanted or what they should sound like, but you said yes regardless. They’ve checked in occasionally, and you kept saying you’re slowly working on it in the background, which you sorta have but not in earnest. Other projects got in the way that had set deadlines and paid up front, and you respect your personal time so you don’t work after hours or anything. Then you learn your client has other clients that are in part relying on this pack, and despite being very much their problem, you get a sense of urgency and decide to just (for lack of better word) shit it out as fast as possible. Maintain your position as a half-decent designer, client is as happy as they can be, all’s well. Let’s talk about how you might do that.

Step 1: Research

Go through your library to see if you have sounds that kinda follow what the spec sheet (if available) asks for. Check YouTube, Freesound, etc. to find material that sounds like it might work. One of the big parts of this set was the sounds of a Yeti, so I looked up other examples of yetis in the past, as well as Minecraft humanoid mob noises. Listen to the sounds closely, figure out how different recordings/creations of said sounds differ and how they’re similar. Pretty much all doves sound the same, but a typewriter sounds different from a modern rubber dome keyboard. Write down (or type like I do because I literally don’t own a notepad or pen) what you need for each sound and variations of said sounds. Then, build a template!

Step 2: Template

I use Phaseplant a lot, even for sample processing, so I tend to set up template presets including samplers, oscillators, filters, distortions, effects, etc. You can do this with most DAWs natively as well. It makes making many similar sounds easier, especially if you set up macros so variations of single sounds are easier. I usually note in the preset how to use it and what it’s for.

As an example, a template I used for this afternoon’s pack consisted of 2 samplers, one loaded with a “Data Loop” in PhasePlant and one that I swapped out from a folder of similar noises, a noise generator, a distortion, a lowpass filter, and envelope controlling total volume, as well as 2 convolvers and a limiter. I had an envelope controlling noise volume separately, and a stepped random source controlling noise colour and distortion amount. There was also a macro to adjust lowpass cutoff and decay time so shorter times would be lowpassed more. In use, all I had to do was align the samplers, adjust the macro, and switch between convolvers (used for reverb and delay).

Step 3: Apply

Using your template, build the new sounds. Ideally, use sounds you’ve made in the past that aren’t exclusive to anything else as platforms for your new sounds, be they mere layers or core components. These past sounds can also just be sounds you made for the pack you’re currently working on – mangle it enough and no one will know you just re-used assets. You can stretch, pitch, distort, filter, FM, AM, PM, ringmod, etc to your premade sound to help get a new sound from it. Bolster it with synthesis (especially in my case because that’s kinda my whole thing). If you have an ultrawide monitor (like I do #21by9lyfe), set up your spec sheet, notes, any external sources like samples, and DAW such that all are visible at all times. You can use PowerToys FancyZones on Windows to automate this and use custom zone shapes.

Step 4: Deliver!

I’ve been told being a sound designer is an artform. I see it more as playing with complex electronics to make weird noises, but I also see photography and cinematography as optical science and exposure balancing rather than an artform so what do I know? The end result is creation of new media; in our case, a boatload of new, slightly similar sounds. I recorded 51 raw files which ended up being split into about 500 raw assets, about half ended up in the final pack. I made sure each file was 192KHz 24-bit (highest my system can manage) and was named according to the spec sheet (literally just copy/pasting what the sheet said, including errors). This was the longest part, just working in Wavepad to trim and export each sound and naming them to spec. Zip them up (using .zip because for some reason no one knows about .7z which makes for WAY smaller files), upload to Dropbox, and send the client a link. They’re happy, I’m happy.

Once you get all set up and build your own workflow, this should be a fairly quick, efficient process for mass sound creation like for a sample pack. It’s also useful in a fast-paced environment, like on-the-fly sound design for TV shows for when your libraries don’t quite cut it. It’s useful in your down time to set up templates and play with ideas for things just in case you need them later. Just make them searchable in your preset browser.