How to get into synth / executable music?
category: music [glöplog]
Hi All
I've been thinking about entering a synth / executable music competition at a party next year. Was wondering how you go about getting into this? Is this kind of competition more about the code trickery in a small exe, the music it plays or a bit of both?
What is a good starting point? any advice would be ace :)
I've been thinking about entering a synth / executable music competition at a party next year. Was wondering how you go about getting into this? Is this kind of competition more about the code trickery in a small exe, the music it plays or a bit of both?
What is a good starting point? any advice would be ace :)
It's a bit of both, in my experience - for me it's always been about having an okay-ish synth, but driving it to the level where it doesn't feel like an executable. A lot of people fall into the mistake of trying to make "standard demomusic" with 64k-synths, which is fine on one hand because that's what they enjoy (I guess), but it does leave a lot of headroom. A lot of the possibilities I've been exploiting is simply just doing something "alternative" with the synth, e.g. genres that are generally not done (hiphop, breaks), and I often model my synth patches after (commercial) stuff I listen to.
The code bit, for the most part, is fairly trivial - it's how the whole synth "feels", is what makes or breaks it.
The code bit, for the most part, is fairly trivial - it's how the whole synth "feels", is what makes or breaks it.
I really wonder about how people vote in general... seriously, each time I'm listening to 4k synth tunes, I feel it's been made by a JMJ fan, too much cheesy patches on cheesy chords but hey, people love it so basically, you know what to expect if you're going to enter a compo with something different.
kaneel: Actually I feel that most 4k/16k-synth music compos nowadays aren't that different from normal music compos, in the way that the winners usually have done something extra to make their music sound nice. In the softsynth-world, that's making patches that sound less and less like traditional "simple" JMJ-sounds. Listen to the stuff by Gargaj, Loaderror and Puryx for example.
In what sense? That is sounds like chiptunes processed through 4k synths?
(Won't talk about gargaj and others, thats all about tastes actually. For example, I really dislike the soundtrack of Elevation. That must be me.)
(Won't talk about gargaj and others, thats all about tastes actually. For example, I really dislike the soundtrack of Elevation. That must be me.)
Are you going to code the synth yourself? I'll assume you are.
The kebby articles are a good starting point, just to get a feel for how a softsynth is structured internally and lots of other small things: http://conspiracy.hu/articles/
It's not that difficult to make synth that does a lots of different things in 16k, using C/C++ and compressing with crinkler/kkrunchy.
Think about how you're going to use the synth. I'd suggest writing it as a VST plugin, so you can use it in your favourite sequencer. This also gives you a semi-usable UI for very little cost.
Think about how you're going to get the music data into the executable. Renoise has a parsed XML file format that is easy to convert to a smaller binary version.
Also, I'd say that if you're writing a synth for the first time, it's not neccesary to go all in on a full-fledged subtractive synth with advanced oscillators, multiple filters, modulation matrix, and chorus, delay, reverb, equalizer, and all the other standard stuff. Because that's a lot of work. IMO, you'd be much better off doing something a bit off-beat that sounds unusual and interesting. Also, the music is much more important than the code. I really like Teos Sunbeam from this years Breakpoint as an example of this, or listen to the music in cdak, which was abstract and wonderful.
The kebby articles are a good starting point, just to get a feel for how a softsynth is structured internally and lots of other small things: http://conspiracy.hu/articles/
It's not that difficult to make synth that does a lots of different things in 16k, using C/C++ and compressing with crinkler/kkrunchy.
Think about how you're going to use the synth. I'd suggest writing it as a VST plugin, so you can use it in your favourite sequencer. This also gives you a semi-usable UI for very little cost.
Think about how you're going to get the music data into the executable. Renoise has a parsed XML file format that is easy to convert to a smaller binary version.
Also, I'd say that if you're writing a synth for the first time, it's not neccesary to go all in on a full-fledged subtractive synth with advanced oscillators, multiple filters, modulation matrix, and chorus, delay, reverb, equalizer, and all the other standard stuff. Because that's a lot of work. IMO, you'd be much better off doing something a bit off-beat that sounds unusual and interesting. Also, the music is much more important than the code. I really like Teos Sunbeam from this years Breakpoint as an example of this, or listen to the music in cdak, which was abstract and wonderful.
Or, knowing nothing about the synth Teo used, perhaps I should restate that: the music is much more important than the complexity of the patches.
Also, here's a treasure chest of possibly applicable DSP code: http://www.musicdsp.org/. YMMV.
Also, here's a treasure chest of possibly applicable DSP code: http://www.musicdsp.org/. YMMV.
kaneel: In the sense of "wow, this bass sounds like it is generated by a $2.500 piece of hardware".
"Reverb is the SSAO of musicians." /kb/
@gargaj: That's pretty insightful. I think it says more about the importance of AO than it does about reverbs, actually.
@revival: I couldn't even begin to start coding a soft synth!! I do C# code for a job and I ain't very good at that either!
I'm more interested in writing the music to be honest, maybe hooking up with a coder who has already written a synth?
I'm more interested in writing the music to be honest, maybe hooking up with a coder who has already written a synth?
gargaj: I would add a condition about where to set the dry/wet-slider.
@gloom: oh please come on. not touching it, slowly moving out of it... and uduntseemehereanylonger.
@h0ffman
maybe take a look at something like this then :)
maybe take a look at something like this then :)
kaneel: I'm sorry, but what is your point? That 4k synths can't make sounds that sound not-softsynth-like? Or that it doesn't matter because the people using them will still make bad music? If it's the former, then I'd say you haven't listened to enough 4k music, and if it's the latter then it is, as you say, a matter of taste. :)
I'll try to avoid the fact you just told me I should just watch enough 4k in hope it's not going to bring a "you should try learning more about synthesizer" remark.
But seriously, I thought I made quite much sense by saying what I just said. I'll rephrase: Many 4kb synths are great but most of the time, people just use them with the worst synth patches ever that sound from the 80s, again, you're right, thats a matter of taste.
(arggh, why did I touch it... WHY?)
But seriously, I thought I made quite much sense by saying what I just said. I'll rephrase: Many 4kb synths are great but most of the time, people just use them with the worst synth patches ever that sound from the 80s, again, you're right, thats a matter of taste.
(arggh, why did I touch it... WHY?)
knl, so change the game! get 4klang or sth- and give it a go! i'd be rather curious what you'd do with it, in fact.
Well, the usual, bunnies and stuff you know.
skrbl: twit me your addy.
the big problem with soft synths is that everyone except loaderror, who rules, does gay funk with them! make mooooar aaangry stuff!
also, the Elevation track REALLY needs a guitar solo!
also, the Elevation track REALLY needs a guitar solo!
it was all downhill since they abandoned modules anyway.
What are the effects processors like in these soft synths? Any good amp simulators, filters? What about side-chain compression? The later is (imho) essential for making phat tunes!
@nic0: you can rest assure that there will be no gay funk if I finally get around to this!
@nic0: you can rest assure that there will be no gay funk if I finally get around to this!
4klang has something to you can send a signal to an other patchso basically, you could do sidechaining but it has no compressor.
4klang has delay/verb and flanger/phaser/chorus, distortion and filters, you gotta keep in mind it's 4k synth, it's not going to sound like your topnotch vsts... and you have no polyphony.
4klang has delay/verb and flanger/phaser/chorus, distortion and filters, you gotta keep in mind it's 4k synth, it's not going to sound like your topnotch vsts... and you have no polyphony.
Actually, 4k synths suck.
It's no miracle that most ppl only get JMJ like sounds out of it because 4k softsynths are hardly more complex/capable than the stuff Jarre had in the late 70s/early 80s, only with more aliasing. Luckily most of them are so modular that you can compensate for that by adding layer upon layer of weirdness to the patches - but this directly affects file size. As would making the synth components more interesting or *gasp* good.
No disrespect to 4k synth coders - many of them are brilliant examples of size coding, data organization and algorithm design, and it's amazing how much music you get in so few bytes. They're perfect for what they do. It's only that I've yet to hear one that does DSP stuff that I'd consider even remotely interesting. (And no, Karplus-Strong strings aren't.)
In 64k it's less bad - there's enough filesize to do either complex patches or complex DSP algorithms. Only problem is: Nobody seems to give a fuck. That darn V2 of mine is almost 10 years old now - and apart from adding sample playback (oooh how exciting) and some basic granular stuff (read: delay line with outer loop that fucks things up) there have been exactly zero tries to surpass it quality-wise, let alone coding a synthesizer that actually sounds good. People either write good DSP routines and then let them rot in an abandened subdirectory on their 2nd machine they only use for surfing, or they write a fully functional synthesizer that sadly doesn't get better than "sawout+=freqency;".
Perhaps that's because people get discouraged when they find out that only 30% of the stuff on musicdsp.org actually works, or there's simply way less scientific papers to steal from than there is in graphics. I don't know.
In the end, Gargaj and Gloom are kind of right. What you need is lots of knowledge and love regarding sound design. Don't try to imitate anything (in the real world, synths are over that point since the early 80s :), just find a unique sound and go for it. Don't expect anything good to come out of any 4k/64k synth by itself.
It's no miracle that most ppl only get JMJ like sounds out of it because 4k softsynths are hardly more complex/capable than the stuff Jarre had in the late 70s/early 80s, only with more aliasing. Luckily most of them are so modular that you can compensate for that by adding layer upon layer of weirdness to the patches - but this directly affects file size. As would making the synth components more interesting or *gasp* good.
No disrespect to 4k synth coders - many of them are brilliant examples of size coding, data organization and algorithm design, and it's amazing how much music you get in so few bytes. They're perfect for what they do. It's only that I've yet to hear one that does DSP stuff that I'd consider even remotely interesting. (And no, Karplus-Strong strings aren't.)
In 64k it's less bad - there's enough filesize to do either complex patches or complex DSP algorithms. Only problem is: Nobody seems to give a fuck. That darn V2 of mine is almost 10 years old now - and apart from adding sample playback (oooh how exciting) and some basic granular stuff (read: delay line with outer loop that fucks things up) there have been exactly zero tries to surpass it quality-wise, let alone coding a synthesizer that actually sounds good. People either write good DSP routines and then let them rot in an abandened subdirectory on their 2nd machine they only use for surfing, or they write a fully functional synthesizer that sadly doesn't get better than "sawout+=freqency;".
Perhaps that's because people get discouraged when they find out that only 30% of the stuff on musicdsp.org actually works, or there's simply way less scientific papers to steal from than there is in graphics. I don't know.
In the end, Gargaj and Gloom are kind of right. What you need is lots of knowledge and love regarding sound design. Don't try to imitate anything (in the real world, synths are over that point since the early 80s :), just find a unique sound and go for it. Don't expect anything good to come out of any 4k/64k synth by itself.
Also, yes, SSAO and Reverb are the same. Crude, tiresome approximations to essential global effects. :)