Fresh idea, feedback needed

[Q_x]

After a cleanup needed in Graphics section, I wanted to start to make this crazy sound playback engine, proposed here:
http://wiki.parpg.net/Feature_list#Pie_in_the_sky

I wanted to make the engine based on three parameters passed from the game to the playback engine:
Mood - to handle information of the base sound: day, night in the wilderness, big ruined city, the underground
Intensity - for handling information of how dynamic the situation actually is
Mode - few off/on switches that will change the music accordingly if the PC will be fighting, will have low HP or interact socially.

I can prepare the working prototype in PureData, which needs an external binary to play, but it should be easy to code it to Python later, when the prototype will be finished.

I'm thinking of a two-layer sequencer. I think I will be able to handle making this type of software.
About three tracks should be played in a loop as an ambient sound, so that the playing engine will be able to change intensity by reducing the volume of one track (lets say some natural sounds for night mood) and increase the volume of another (like drums). This is the ambient layer.
The other layer is for effects - I wanted to make it out of randomly chosen samples, playing from time to time. The set of samples will be determined by the mood, like wild didgeridoo solo, few aggressive bass or guitar sounds when fighting or blood rushing and heartbeats when low on HP. Both when fighting and loosing. This samples will be played to the rhythm (more often with high intensity), so that hopefully it will all blend into audible soundtrack.

I'm aware of all the hard moments ahead. Matching sounds saying different things in one language, tracks staying consistent with each other and matching resonant frequencies across few art pieces to have it all really blending and mixing well.

I'm looking forward for a constructive criticism of this idea.
It would be also very helpful if you know where I can get a set of tracks (separate tracks in flac or wav) to use for testing. I'm not a musician (at least at this moment, but "never say never" ), I'm unable to make the music samples to use when developing and testing. I thought of using the tracks released by Nine Inch Nails for the sake of remixing it, I'm pretty sure Trent would appreciate the idea of making semi-automated software for remixing his work in pretty much random way, but its not the best suited set of tracks for this purpose - what I would need is rather an easy, even and smooth ambient on three or four tracks that could be easy to increase intensity of the music with changing the volume of the separate tracks. Effect samples matching a particular set of tracks should be easy to find.

[timong]

I think it's a pretty cool idea what we have here. The tracks of the game are already quite ambientish so far, so I bet this couldn't go too wrong given the right 'tracks' mixed in good rhythm and probably as you mentioned on IRC with some post process in game reverb or precalculated reverbs on the tracks for lesser CPU burden. I can't promise i'll jump on this wagon soon coz' my timeline is too much restrained, but if things are shaping up I'll try to provide some simpler tracks for this system.

[Q_x]

Its cool you think it does have any sense With this kind of ideas I'm always afraid that it has been made or it will be totally useless. Or that I'm not clear enough and sound like a prophet.
Its not urgent. I did not even start to code. I hope to have it working quite fast though, it looks simple on paper.
I'll hold myself with applying any pre-calculated type of intervention cause I need to be sure when doing this type of thing. Until there will be need for actually freeing some CPU - probably until implementing it in Python - I can fiddle with various approaches to making the sound smoother in the end. I've seen PureData rotating, noising, bluring, inverting and delaying video at once, all in software and all in realtime, adding few effects to a single stereo stream should be much easier
I'm still afraid a little bit and worried of this very lats stage - when all should sound really like from the same story.
It would be good if you help me with such kind of effort in other way than talking about the details (talking is very good thing to do regardless of state of things). But I have to work on my own until finishing the software part. I think I will be shouting really loud when that will happen Making screencast will be a must.

[Q_x]

I think I have ended this project. Or rather: ended a part of work and stuck, but I have made some serious conclusions.

I have several problems which I was unable to solve, and they are related directly to crippled capabilities of software (Pure Data) in storing any flexible-size type of list. I probably can avoid this, but, since I had only making a prototype in mind (that is partially ready), there is no sense in continuing work, as this part will look totally different when implemented in something that has full-blown syntax (allowing adding and removing items from list, giving back list length and stuff like that).

So I basically hit the wall with prototyping.
However, I have more or less clear view of how things should work and behave, I did some math too.

Conclusions are:
Playback machine alone is pretty simple piece of software, roughly resembing "remote controlled" sequencer. Needed set of functions is pretty similar: playing loops, some shorter things to the rhythm, changing volume accordingly to what is in game. Approach of using it needs much more work from artists. It needs some extra sound engineering as well. It needs extra input from the game engine itself - this are not things that are implemented now.

I have used NIN samples when prototyping, I can upload path I have made without samples (if needed, there is not much "in it") or just flesh things out on wiki somewhere or, which I'd do gladly, wait until some other persons will be happy cooperating to make this feature a part of reality in some distant future. If, of course, it will be needed/wanted feature at all.