KVR Developer Challenge Any Year Wish List

[harryupbabble]

A complete tutorial program on how to be a better programmer than you. Kidding.
But seriously, this thread should exist waaaaaaaaay before the actual start date of any Developer Challenge.

I am drawing a blank at the moment though. Pardon me.

[harryupbabble]

Songwriting tools:

Better Chord Progression Generators
Better Riff Generators
Better Drum Pattern Generators
Better Lyrics Generators


Note: some may say "what a lazy music-maker you are". What's so lazy about spending so much time rejecting 99 percent of what generators generate? I figure it's a lot like noodling on the piano looking for nice sequences of notes. The difference is that humans have muscle memory but if the programmer is using "absolute true randomizers" then maybe generators don't have "muscle memory"?

[harryupbabble]

A software that would let me know which track in my DAW caused the master track to go over the volume level limit.

That way I could manually lower that track's volume level, the one that is too-loud-by-a-minuscule-amount-but-is-still-clipping-the-master-track track.

[Functional]

Songwriting tools:

Better Chord Progression Generators
Better Riff Generators
Better Drum Pattern Generators
Better Lyrics Generators

There isn't really such a thing as a "true randomizer", because that would imply non-deterministic algorithm and we don't know any AFAIK, none of the "truly non-deterministic" ones at least. Well, kind of. For example, with prime numbers you can come up with an algorithm that, to us, has characteristics of a non-deterministic algorithm (is why prime numbers are really loved in encryption), but this is far from conventional for this purpose. And they're not really non-deterministic (they follow from the axioms themselves), they're just PITA to calculate. Otherwise usually you can use stuff like weather data to have a random algorithm. It's still non-deterministic because weather is perceived as chaotic system, rather than truly based on probability.

NOTE: I'm using terminology here a bit wrong. By non-deterministic I actually am talking about RNG, which all of non-deterministic algorithms use. The "truly non-deterministic algorithms" should be RNG that spreads values across 50/50 on n tries. Such thing doesn't exist. Even weather data shoots more towards big numbers than smaller ones, so it's more like 49/51.

Problem with these pattern generators is that either they seem random and produce non-cohesive results more often than not or they arent and instead produce cohesive, yet boring results. There are also second kind of generators which take phrases as input and produce some kind of phrase based on that, these sort of are "muscle memory" ones. If you truly feel like you want this sort of stuff, then I would suggest go looking at Synfire by Cognitone. But problem here is, from what I've heard, that it takes a LOT of effort to learn how Synfire works and so on. It doesn't produce magic results on its own. So one might as well just do what it does; look at others compositions and mimic their phrases on a piano keyboard yourself.

I was kind of interested in this stuff some years ago but after giving it all a try, it wasn't really interesting for me. Whenever I lacked inspiration, I just looked at others compositions and tried to mimic them one way or another. I got better results. This may not be true for everyone, but it was true for me.

[xoxos]

I figure it's a lot like noodling on the piano looking for nice sequences of notes.

the thing about noddling teh piano is theoretically your sensibility improves. hitting the render button, you cannot petition the lord with prayer. but someone else may answer.

[harryupbabble]

There isn't really such a thing as a "true randomizer"

I have faith in randomizers, even pseudo-randomizers, especially the ones where the user get to feed it whatever they want. For example, there is this DC16 entry, it's called Random Midi Arranger. The user drags and drops midi patterns onto it. The midi patterns are 4 bars long. The user creates the midi patterns and there can be hundreds, maybe even thousands of different midi patterns and RMA randomly picks a song's worth of midi clips from those hundreds/thousands clips that the user created and randomizes the order of those midi clips/patterns and also randomizes the song structure. It's up to you what the content of those 4-bar long midi clips/patterns are to be. You can fill it up with rock music phrases or EDM phrases or flamenco music phrases or whatever. Or or or you can use RMA's built-in randomizer to randomize those 4-bar clips too.

The way I see it is that the randomizers or generators never creates nor decides anything. In the end it's you, the composer, that gets to decide what stays and what goes. Again, it's a lot like noodling on the piano and deciding what sequences to keep and what to discard. Lets say the randomizer (not necessarily the RMA randomizer itself but one that is designed to only randomize midi clips) produces 100 results, and let's say I decide to keep result #89 and reject the rest. What's the chance that you will also decide to keep result #89? And that could just be a decision involving 4 bars of the bass drum part.

So even if "true random" is not doable, other randomizers, not just RMA, could still expose you to music sequences you might never ever stumble upon compared to if you are just noodling on the piano hearing note sequences that your fingers pressed or in other words, sequences that are influenced by muscle memory. I figure that deciding what to keep and what to discard from the results that randomizers produces could be a kind of a skill similar to what casting directors and other professions that require judging of what/who stays and what/who goes?

But I guess at this point I'm just babbling because so far all my songs suck, especially the ones that I used randomizers on, just like you said you weren't happy with what the randomizers presented to you when you tried them out. But the thing is, maybe it takes a while, just like composing using the traditional ways also takes time. I heard that Brian Eno and a lot of other musicians did/do use randomizers, and who knows maybe they did get results that they did like.