| |
Mr SQL
Registered: Feb 2023
Posts: 137 |
Experiments on the Sound
I've been working on an algorithm to dynamically recompile the byte code for the SID chip, creating interesting and pleasing tunes provide the original tune is melodic.
Basically GIGO (garbage in garbage out) ensues with this algorithm if the tune used as the input has no Melody.
I've received a lot of feedback with earlier versions of the algorithm with some C64 fans really enjoying the sound and some not as pleased, either way share your thoughts here and any ideas to improve the algorithm.
Cross platform algorithmic arrangements:
This algorithm runs cross platform and works on the Atari 2600 TIA as well as the SID.
Here is the WIP for STARBLITZ Neon Sound on the C64 and Atari respectively:
https://youtu.be/ixuHdHvqMK8
https://youtu.be/FfAga57yJFs
The original musical composition can be heard at 4:30 in the Atari version, a Blues piece. Various soft piano and Jazz and electronic sound implementations are heard in the C64 version. I find it interesting that different music genres appear to emerge as the musical score is recursively processed by the algorithm.
Another algorithm is used to interpret the TIA on the C64 SID which sounds completely different. Even trying to get the 6581 and 8050 SID models to sound uniform is very difficult. I am not trying to get the same exact Sound like the gameplay and graphics (besides the motion pixel art). |
|
... 29 posts hidden. Click here to view all posts....
|
| |
Bansai
Registered: Feb 2023
Posts: 49 |
Quoting acrouzetI think what they're saying is that this program takes an input .sid file and scrambles its data to create a procedurally generated output. Sounds like it's an attempt at something similar to what is described here: https://www.langston.com/Papers/amc.pdf |
| |
Hate Bush
Registered: Jul 2002
Posts: 465 |
i find your persistence fascinating. |
| |
Mr SQL
Registered: Feb 2023
Posts: 137 |
Quote: Quoting acrouzetI think what they're saying is that this program takes an input .sid file and scrambles its data to create a procedurally generated output. Sounds like it's an attempt at something similar to what is described here: https://www.langston.com/Papers/amc.pdf
Great research paper, this is an interesting read!
‘‘. . . any attempts to simulate
the compositional abilities of humans will probably not succeed until in fact the musical models and plans
that humans use are described and modeled.’’ [MOORER72] And more recently: ‘‘It seems that musical
composition is a hard mental task that requires a substantial amount of knowledge, and any serious attempt
to simulate ‘noncomputer’ music composition on the computer would have to face the task of constructing
a formal model of considerable complexity. We have found that even the algorithmic representation of the
knowledge underlying the seemingly simple Bach chorale style is a task that already borders the
intractable.’’
I see some parallels in my research to the theory these researchers came up with. I had an earlier attempt that just used an algorithm in GateCrasher Jazz Improv:
https://bunsen.itch.io/gate-crasher-commodore-64-by-mr-sql
This worked OK for scales and riffs but to model the complex mathematical musical structures composers create, an algorithm processing an existing composition can maintain some of their relationships reusing the math.
I'm still reading the paper, wonder if these researchers considered this cheating or also implemented this idea. |
| |
Frostbyte
Registered: Aug 2003
Posts: 183 |
I've been a bit hesitant in commenting about these experiments, as I really wouldn't want to curb your enthusiasm as you're clearly enjoying what you're doing.
On theoretical level I find the concept very fascinating - an algorithm that takes something proven good and pleasing as an input and does a programmatic "re-imagination" of it, in the lack of better words.
However on practical level, as it currently stands, it simply does not work. Now I've listened to all kinds of weird sh*t in my life from brutal Japanese noise via early electronic music experiments to all kinds of avantgardistic or minimalistic compositions. Even though most people wouldn't define any of the aforementioned as music in any way, they've all had at least something to grab on to - let it be captivating structures, fascinating soundscapes, interesting rhythms, pleasing evolvement (sometimes over a very long period of time), some melodic structure even if not necessarily on the typical western chromatic scale or even any scale, whatever.
As your algorithm currently stands, as far as I can hear, it does none of this. Maybe you have a different view on what makes music music, and clearly a very different view from mine what makes music Blues or Jazz, but all I can hear is nearly random bleeps and bloops with uninteresting, simple sounds that do not complement the SID chip's capabilities.
I'm not hear to tell you what to do, you do you and if it floats your boat, so be it. But at least in my opinion you'd need to rethink your algorithm for it to deliver anything that even remotely resembles music. Maybe introduce some guidelines that would direct the algo to build structures closer to the input composition, guidelines about what is preferred melodically, soundwise etc, and maybe the algorithm would eventually end up producing something music-like, or at least something that sounds interesting.
Anyway, that was my long 2p. I also appreciate your enthusiasm like Mr. Bush there, even though the results might not be what I'd call great. At least you're doing something, and that something is stuff that AFAIK no one else is currently experimenting with on C64, so there's that. |
| |
Flotsam
Registered: Jan 2004
Posts: 84 |
Interesting shit, even if the current results are a bit monotone sounding. I'm interested in seeing how it evolves, might complement an experimental, procedural game nicely. Also would be cool to hear something in native format instead of a YT video recorded with a potato. Keep up the good work, I'm following. |
| |
Mixer
Registered: Apr 2008
Posts: 452 |
My Roomba likes it. It has been resting a lot recently, it is old and its battery is worn. I played some 8-bit LFSR generated sequences that were mapped to simple scales for 3 SID voices to it. Roomba listened to it still and in silence, so it must have been enjoying the experience. Usually when I play SID music, it keeps running around the chairs's leg, bumping to it repeatedly clearly asking me to stop. After listening to that LFSR loop for a few hours I knew the melody myself. So, what ever it is, just make it loud and repeat, you'll like it eventually. |
| |
tlr
Registered: Sep 2003
Posts: 1790 |
Quoting MixerI played some 8-bit LFSR generated sequences that were mapped to simple scales for 3 SID voices to it. Roomba listened to it still and in silence, so it must have been enjoying the experience. which polynomial was that? |
| |
Mr SQL
Registered: Feb 2023
Posts: 137 |
I've continued this research with the soundtrack and Fx in my port of Tron Deadly Discs.
Here are three interesting algorithmic songs from the Atari 2600 version:
I will share the C64 version next.
https://www.youtube.com/watch?v=QCBkm3xVKho |
| |
JackAsser
Registered: Jun 2002
Posts: 2014 |
Quote: I've continued this research with the soundtrack and Fx in my port of Tron Deadly Discs.
Here are three interesting algorithmic songs from the Atari 2600 version:
I will share the C64 version next.
https://www.youtube.com/watch?v=QCBkm3xVKho
Please explain the research, because visuals and audio only doesn't make me understand tbh. Anyway, please keep up whatever floats your boat. |
| |
Oswald
Registered: Apr 2002
Posts: 5094 |
Quote: I've continued this research with the soundtrack and Fx in my port of Tron Deadly Discs.
Here are three interesting algorithmic songs from the Atari 2600 version:
I will share the C64 version next.
https://www.youtube.com/watch?v=QCBkm3xVKho
sounds exactly as what I would expect from a randomly generated "music" (ie not good) |
| Previous - 1 | 2 | 3 | 4 - Next |
