How do you "make" sound effects?
Author
I can draw graphics, compose music, write code.. but how on earth do I create a sound effect? The sound that Mario makes when he jumps or the sound the Tetris block makes when you flip it. I have no idea where to even start.
I did a search on Gamedev.net and all I got were articles talking about advanced philosophies or implimenting surround sound. But where do those sound effects come from to begin with?
Do you have to record them from something? Are they generated entirely on the computer? Are there any tutorials or examples online for getting started?
Granted I'm more just a hobbyist who has picked up things along the way, but I figure I can steer you in some generally right direction. You can generally make sound effects using techniques that fall under one of a couple of general overarching categories: synthesis-based techniques or sample-based techniques.
In synthesis, essentially you're creating sound from scratch. You'll start off with some simple waveform, e.g., a sine, sawtooth, triangle, pulse, or even white/pink noise, and using different synthesis techniques to create sound. Probably the most pervasive method is subtractive synthesis where you take your original waveform and use filters to "subtract out" parts of the frequency spectrum you don't want. For example, the Mario jump sound you mentioned could be done using this technique. Check out the Triforce VST or any of the other Tweakbench VST instruments to see some subtractive synths geared towards recreating old NES sounds.
Other synthesis techniques you'll run across are additive synthesis (take a bunch of fixed pitch oscillators and sum them together to create a timbre), amplitude modulation (vary the amplitude of a carrier signal by a modulating signal... if the modulator's frequency is low enough, the effect heard is tremolo), frequency modulation (popular technique starting off in the 80s.. it's similar to amplitude modulation except the *frequency* of the carrier is varied by the modulating signal.. a modulator with a low enough frequency creates a vibrato effect), granular synthesis (breaks a sound up into very small snippets with varying durations called grains, and plays them back at, possibly, varying intervals, pitches, stereo positioning, ADSR envelopes, etc.. check out KTGranulator for an example VST plugin), physical modeling (recreate a sound by modeling the physical resonating bodies that produce the sound), and many others.
In sample-based techniques, you'll go out and actually record sounds and then, after recording your samples, doing some post-processing to mash and mold them into the desired effects that you want. I find that part of the fun of this method is that I won't always have the actual thing whose sound I want (or I'll have it but it sounds weak when recorded or just doesn't lend itself to making a strong effect), so then I have to go out and improv by finding things that sound like the sound I want to capture. Then I'll take them to my computer and layer the samples together to make the final sound. When you're doing the post-processing, you'll most likely find yourself playing around with things like EQ, pitch shifting, time stretching, and reverb to sculpt the sound you want from your raw recordings.
However, there are some gray areas in between. For instance, some programs (like NI's Absynth) allows you to load samples you've taken and then treat them as waveforms that you'd use in a typical synth. This can potentially give you a new array of timbres that would otherwise be difficult to acquire using simpler waveforms. Another instance where the two will typically cross is in vocoding. For instance, you could vocode a recording of some speech with some synth patch you've created to make it sound like the synth is talking (can get some pretty cool vocal effects this way).
Above all, aside from knowing some of the techniques, it's just a matter of going out, listening closely to what you hear, and trying things out to try to recreate that sound. You may end up using some combination of the above techniques or something completely different that allows you to get some really interesting sounds. Like anything else, there's really no substitute for hands on experience.
In synthesis, essentially you're creating sound from scratch. You'll start off with some simple waveform, e.g., a sine, sawtooth, triangle, pulse, or even white/pink noise, and using different synthesis techniques to create sound. Probably the most pervasive method is subtractive synthesis where you take your original waveform and use filters to "subtract out" parts of the frequency spectrum you don't want. For example, the Mario jump sound you mentioned could be done using this technique. Check out the Triforce VST or any of the other Tweakbench VST instruments to see some subtractive synths geared towards recreating old NES sounds.
Other synthesis techniques you'll run across are additive synthesis (take a bunch of fixed pitch oscillators and sum them together to create a timbre), amplitude modulation (vary the amplitude of a carrier signal by a modulating signal... if the modulator's frequency is low enough, the effect heard is tremolo), frequency modulation (popular technique starting off in the 80s.. it's similar to amplitude modulation except the *frequency* of the carrier is varied by the modulating signal.. a modulator with a low enough frequency creates a vibrato effect), granular synthesis (breaks a sound up into very small snippets with varying durations called grains, and plays them back at, possibly, varying intervals, pitches, stereo positioning, ADSR envelopes, etc.. check out KTGranulator for an example VST plugin), physical modeling (recreate a sound by modeling the physical resonating bodies that produce the sound), and many others.
In sample-based techniques, you'll go out and actually record sounds and then, after recording your samples, doing some post-processing to mash and mold them into the desired effects that you want. I find that part of the fun of this method is that I won't always have the actual thing whose sound I want (or I'll have it but it sounds weak when recorded or just doesn't lend itself to making a strong effect), so then I have to go out and improv by finding things that sound like the sound I want to capture. Then I'll take them to my computer and layer the samples together to make the final sound. When you're doing the post-processing, you'll most likely find yourself playing around with things like EQ, pitch shifting, time stretching, and reverb to sculpt the sound you want from your raw recordings.
However, there are some gray areas in between. For instance, some programs (like NI's Absynth) allows you to load samples you've taken and then treat them as waveforms that you'd use in a typical synth. This can potentially give you a new array of timbres that would otherwise be difficult to acquire using simpler waveforms. Another instance where the two will typically cross is in vocoding. For instance, you could vocode a recording of some speech with some synth patch you've created to make it sound like the synth is talking (can get some pretty cool vocal effects this way).
Above all, aside from knowing some of the techniques, it's just a matter of going out, listening closely to what you hear, and trying things out to try to recreate that sound. You may end up using some combination of the above techniques or something completely different that allows you to get some really interesting sounds. Like anything else, there's really no substitute for hands on experience.
Wow, that was a great response.
I've only worked with the recorded sounds bit, which can return some great results. While I strongly believe that most sounds can be created with a single coffee cup, so long as you are creative enough, going around and making recordings with various objects is a great way to go. The star wars laser sound? A significant part of that was created by smacking a loose chain-link fence.
I never did use any tutorials, but I bet they would have been very helpful if I had found some. My methods involve a whole lot of trial and error until I finally reach something similar to what I want (by, largely, doing exactly what romer explained).
I've only worked with the recorded sounds bit, which can return some great results. While I strongly believe that most sounds can be created with a single coffee cup, so long as you are creative enough, going around and making recordings with various objects is a great way to go. The star wars laser sound? A significant part of that was created by smacking a loose chain-link fence.
I never did use any tutorials, but I bet they would have been very helpful if I had found some. My methods involve a whole lot of trial and error until I finally reach something similar to what I want (by, largely, doing exactly what romer explained).
Stomper Hyperion is an interesting program for making peculiar sound effects from synth sounds (from scratch more or less), and it is also a good place to start - very simple interface. (you'll find the download if you search for it on google)
Laser sounds, "jumping sounds" as we know them, explosion sounds - all rather easy to make in programs like Stomper, once you fool around for a bit.
Laser sounds, "jumping sounds" as we know them, explosion sounds - all rather easy to make in programs like Stomper, once you fool around for a bit.
Nice post, romer. I'll have to bookmark this for when I get around to making sounds for my game.
Wow, Romer's said more than I could possibly hope to say, but I do have one little piece of advice for getting the right sound:
Make sounds that you expect to hear, rather than do hear. Real gunshots, for example, sound terrible in a game, and sound infinitely better if you overlay several other booming sounds on top of it.
Also remember that graphics and sound go hand in hand. Make sure for a big, unique effect you have a big, unique sound, and if you have a small, regular effect make sure you've got an identifiable yet unobtrusive sound.
And for the love of god, avoid those 1001 free sample CDs like the plague.
Make sounds that you expect to hear, rather than do hear. Real gunshots, for example, sound terrible in a game, and sound infinitely better if you overlay several other booming sounds on top of it.
Also remember that graphics and sound go hand in hand. Make sure for a big, unique effect you have a big, unique sound, and if you have a small, regular effect make sure you've got an identifiable yet unobtrusive sound.
And for the love of god, avoid those 1001 free sample CDs like the plague.
Quote:Original post by romerVery informative! Thanks for this.
Granted I'm more just a hobbyist who has picked up things along the way, but I figure I can steer you in some generally right direction. You can generally make sound effects using techniques that fall under one of a couple of general overarching categories: synthesis-based techniques or sample-based techniques...
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
