glass-bead game

Inventor · Posted: Sun Jul 06, 2008 7:03 pm Post subject:

At some point perhaps we should notify the USCF and FIDE about this effort. They might be responsive and popularize it somehow, perhaps a brief mention in Chess Life Magazine or on their web page would be nice.

Kassen · Posted: Mon Jul 07, 2008 10:36 am Post subject:

http://www.news.com.au/story/0,23599,23979955-23109,00.html

Just add pressure sensors to the gloves. And ropes. And boots.

Acoustic Interloper · Posted: Mon Jul 07, 2008 10:53 am Post subject:

Acoustic Interloper · Posted: Sun Jul 13, 2008 7:06 pm Post subject:

Attached is the latest revision, a few new features and numerous bug fixes. chorusmanytones.ck, like manytones.ck, has 4 banks x 64 oscillators per bank, and is designed to run with testf(). While manytones.ck does OK under miniAudicle, chorusmanytones.ck hangs miniAudicle, but runs OK under command line ChucK.

EDIT: chess.zip attachment remove, see updated code from August 2 for latest version.

dewdrop_world · Posted: Mon Jul 28, 2008 8:07 pm Post subject:

Realizing that EM 2008 is coming up, I thought I'd better update my supercollider contribution for the new testf() protocol. So here's a first go... I also want to add variations on the synthdefs to represent the conflict relationship, but that isn't done yet. Right now there is a different sound for each piece type (except King -- I had an idea what I wanted to do, but it isn't a good one so I'll have to go back and revisit -- right now that piece will be silent).

I also plan to throw in a touch of reverb to try to glue the sounds together better.

The steps to start it up are the same. I did only a quick test on Mac (successful after working out some stupid bugs), not sure how well it will work in Windows. I might have some time on my lunch hour tomorrow to give it a quick try. I'm worried mostly about loading the buffers -- I've seen Windows drop UDP messages under heavy traffic, where I don't think I've ever seen that on my Mac.

Unfortunately, things have been... complicated... here lately and I won't be able to attend. It would have been fun but under the circumstances, I couldn't possibly have arranged it.

I'd love to hear comments or suggestions, and of course let me know if something doesn't work.

James

chess_music_2.txt

Description:

SuperCollider chess music update.

Download (listen)

Filename:

chess_music_2.txt

Filesize:

8.26 KB

Downloaded:

369 Time(s)

Acoustic Interloper · Posted: Tue Jul 29, 2008 6:33 am Post subject:

Thanks, James. It'll be a few days until I do another round of playing with it -- I gave my current long-distance employer 2 weeks notice yesterday, and there's something that they want me to hurry up and do. (So of course I am catching up on the Forum first thing!) I want to stay in their good graces, and keep a few hours a week of contracting work after I start teaching in the fall.

I made some changes to my parts of the code as well, changed XML/RPC to UDP because XML/RPC on OSX was having horrible delays when used over a LAN (as opposed to within a single machine). Fixed some bugs, improved my ChucK patch, improved the GUI. If you want that now, I can post it, otherwise I'll wait until I get to try your patch out.

Did the UDP droppage you saw occur within a single machine, or across a LAN? I always run the chess server and the sound generator in the same machine, both to minimize packet loss and to minimize latency, and have not seen a problem (actually, haven't heard a problem). The GUIs can be distributed, and actually have fewer problems with UDP than with XML/RPC/TCP, which should be more reliable because it uses TCP, but the Python XML/RPC library on OSX is rather suspect.

Since you are still tweaking your patch, I'll wait a couple days to see if you post any changes. Will try it out by the weekend at latest, and put it into the next "release."

Have a good week!

dewdrop_world · Posted: Tue Jul 29, 2008 10:08 am Post subject:

Cool - I do have a couple of tweaks. I found a bug with the bishop sound and of course the other improvements I mentioned last night.

I tested just now in Windows and all the synthdefs load properly.

I only noticed UDP droppage when filling wavetable buffers with client data using OSC. I could avoid the problem by adding a short pause between buffers, which I guess allowed some buffer to clear. I haven't noticed it during sequencing in Windows at all -- compared to buffer population, where dozens of big datagrams might be sent in a fraction of a second, triggering even 20-30 notes at the same time is relatively cheap.

James

dewdrop_world · Posted: Tue Jul 29, 2008 8:30 pm Post subject:

Maybe another bug?

Doing some more testing tonight, I noticed that my supercollider client is receiving only 0 or 1 for piece type. Even if I bring the bishop out and directly threaten another piece, the bishop synth never gets called -- only pawn or knight.

Or maybe I misunderstood something in the protocol?
James

Acoustic Interloper · Posted: Wed Jul 30, 2008 6:46 am Post subject:

I haven't seen a problem with this, but you never know whether I broke something. Here's the ChucK code for comparison. It includes a print statement for debugging.

dewdrop_world · Posted: Wed Jul 30, 2008 7:54 am Post subject:

I see... the protocol as described here says bank, osc, freq, phase, leftamp, rightamp, piece type, player but your code flips player and piece type.

Documentation is out of sync with the code then Smile

I tracked down another problem this morning. When a king is put in check, lots of oscillators get triggered for just a millisecond or two. That's causing "zombie nodes" in sc that never get released, eventually overwhelming the CPU.

I can work around it by delaying all releases by, oh, I don't know, 20-30 milliseconds. But the shortest of my envelopes have a 10ms ramp-in so it seems to me that notes are generated that are not perceptually significant.

Is this intended? What does it sound like in ChucK?

James

Acoustic Interloper · Posted: Wed Jul 30, 2008 8:35 am Post subject:

Acoustic Interloper · Posted: Wed Jul 30, 2008 8:39 am Post subject:

The PROTOCOL_2.txt description is up to date and has the correct ordering for the OSC parameters. I am attaching the latest update from maybe a week ago, to make sure we are in sync.

I won't get to look at the check noise timing for a few days.

Have fun!

EDIT: chess.zip attachment remove, see updated code from August 2 for latest version.

dewdrop_world · Posted: Fri Aug 01, 2008 7:15 am Post subject:

Here's an update from my side. I haven't tried it with the latest python stuff yet, though. Reverb is working now, synths are correctly chosen by piecetype, and there is a little mixer to tweak synth and reverb levels.

I haven't done the King sound yet. But, I did delay note releases so there are no more stuck synths. (It was easy because of message timestamps to the scsynth server - where the note onsets go out with a latency of 0.2, releases now use 0.3 so every note must be at least 0.1 second. This is easy to adjust once the game server fixes the too-short notes.)

hjh

chess_music_2.txt

Description:

SuperCollider chess music code, 1 August 2008.

Download (listen)

Filename:

chess_music_2.txt

Filesize:

11.19 KB

Downloaded:

293 Time(s)

Acoustic Interloper · Posted: Fri Aug 01, 2008 7:49 am Post subject:

Thanks, James! I'll try it out and copy it into the EM2008 distribution (as it were) this weekend.

I took a look at the CHECK! Python code the other day, and the sequence of events should be:

1. Shut off all currently played notes.

2. Delay according to tempo.

3. Turn on two notes for every piecetype, one for support and one for conflict, across 5 octaves. This would give 2 x 6 x 5 = 60 simultaneous notes.

4. Delay according to tempo.

Repeat above until a move out of check is made.

I would expect all manner of clipping. I do scale the amplitudes back so that their sum should not exceed unity gain by much! So, I expect that it would be ugly, but once all 60 notes are turned on, I wouldn't expect any timing difference from a non-check board configuration. I'll do some debugging, but I am glad you found a workaround.

There was a ChucK release the other week, and a friend of mine at work said the former ChucK release sounded better (i.e., uglier) on CHECK than the new ChucK release.

Have a good weekend.

Acoustic Interloper · Posted: Sat Aug 02, 2008 10:18 am Post subject:

Sounds very nice! I like the contrasts between my ChucK tone generator and your SC generator! They probably reflect differences between your approach to composition and mine, and between our perspective on the chess pieces/relationships/moves mapping to music, and between ChucK and SC. Obviously, I get bored with my own compositions after a while, and it's a nice change to listen to yours while playing. I am probably going to hook up 2 computers at some point, one running ChucK and one SC, and send OSC packets to both, just to see how they sound together.

I fixed the King CHECK panic sound timing issue. But, you should definitely KEEP your code in place for dealing with notes that are too short in time. The player can set both the main tempo scaling factor and the factor that scales application of the users' move speed to tempo, to 0.0, so essentially SC or ChucK or Max can get hammered constantly with notes that steal working oscillators in a bank. They should not crash in this condition. They are free to go silent or drop notes or whatever is needed in this case, and obviously if the player does not like what he or she hears, the player can increase these timing scale factors. I played your SC patch for a while with these both set to 0.0, and it sounded just fine!

The CHECK bug itself was due to the fact that I was asking for more oscillators at a time than exist in a bank. When this happens, the code looks at running oscillators that have sustains set by the player, and steals the one with the lowest sustain. Since by default the sustain is 0, it would always just keep stealing the same oscillator, effectively turning it on and off at different frequencies. I put a fix in testg() only (which runs plugin pannerSustainInterpreter.py) to run only as many oscillators as it has in a bank at one time. It then waits according to tempo, zeroes all those oscillators, and does it again with some other notes. I also added some randomness here to make CHECK more dissonant.

I did not change this behavior in the other plugins, and, as noted above, it's a good idea for the sound generator to be able to deal with very rapid notes on-then-off, even if it deals with them by dropping them. I still occasionally see one of these appearing from somewhere else, presumably when some oscillator is arbitrary selected for reuse. Besides laziness on my part, other people may write plugins, possible some of my future students, so it's probably a good idea to be prepared for this condition.

You can now start this by entering './gamestart' in one terminal window from within the chess/chess/ directory, and entering './guistart' in another. The guistart can optionally be followed by an IP address if the game server process is on another machine on the lan, e.g., './guistart 192.168.1.3'. I plan to have two GUI terminals set up at EM2008.

The updated code is attached, including your patch. Thanks again! Have fun.

EDIT: See August 11 post below for the EM2008 Festival update of the chess program.

Inventor · Posted: Sat Aug 02, 2008 11:40 am Post subject:

i sent the following to the USCF:

Acoustic Interloper · Posted: Sat Aug 02, 2008 1:00 pm Post subject:

Thanks. It'll be interesting to see if anyone takes an interest.

I'll probably write a Scrabble server for this at some point. I wonder if I am allowed to say Scrabble(tm)?

A friend of mine pointed me at XBoard and GNU Chess. The former is a GUI that can talk to the latter; my friend thinks that they use stdin/stdout rather than pipes or sockets, but whatever they use, one could undoubtedly insert an interceptor to spy on the game and generate music. Presumaby this would entail both human-vs-human and human-vs-machine games.

I am teaching a grad course in UNIX System Programming for the first time this fall. Depending on the difficulty, I suspect I will have students integrate those above 2 programs with this chess-to-music program as one of their major programming projects.

dewdrop_world · Posted: Sun Aug 03, 2008 9:23 am Post subject:

With this update, I think I'm going to have to consider my contribution finished. Each piece now has a "conflict" variation (using a ringing filter + distortion for a nastier sound), and the king makes noise now too.

(By the way, for the next release -- supercollider reserves the .sc extension for class library files. We've been using .scd for plaintext code files to run interactively, but .txt works just as well.)

I finally got around to listening to the MP3 game posted a while back -- interesting difference indeed. I get the feeling you're doing something different with sending a new frequency to an already-active oscillator. Right now I'm dropping the old synth and starting a new one for the new frequency (re-attacking the note). It wouldn't be hard to keep the same note but just change frequency. No time to try it at the moment, though...

James

chess_music_sc.txt

Description:

Last one?

Download (listen)

Filename:

chess_music_sc.txt

Filesize:

10.83 KB

Downloaded:

284 Time(s)

Acoustic Interloper · Posted: Sun Aug 03, 2008 11:33 am Post subject:

Acoustic Interloper · Posted: Sun Aug 03, 2008 12:11 pm Post subject:

Acoustic Interloper · Posted: Mon Aug 11, 2008 5:27 pm Post subject: Official EM2008 copy

Attached is the official electro-music 2008 festival copy of the chess-to-music program to be demoed Thursday evening. See above for background, README for instructions. Thanks to all who helped! I'll let you know how it goes.

EDIT: Updated README.txt in this copy. testh() is the latest, fanciest plugin configuration. Gotta go pack for em2008.

chess12aug_em2008.zip

Description:

Download (listen)

Filename:

chess12aug_em2008.zip

Filesize:

347.58 KB

Downloaded:

274 Time(s)

Acoustic Interloper · Posted: Mon Aug 18, 2008 12:12 pm Post subject:

Attendance at the chess game demo was pretty light, late Thursday evening, although I did get three people willing to play who were interested in what was going on, including both ChucK and SC sound generators. I think just setting this up as an installation piece would have been better, but I needed the equipment for other activities on other days.

Acoustic Interloper · Posted: Mon Mar 16, 2009 4:08 pm Post subject:

blue hell · Posted: Mon Mar 16, 2009 4:22 pm Post subject:

Ah, good news! And looking forward to examples from the scrabble machine.

Acoustic Interloper

I see that I am not the only person from this thread who will be attending NIME2009. Way to go, Dew Drop! Are you staying for the whole shebang? I'll be driving out the 3rd from Philly area, maybe with my son Jeremy along with me. I got 2 pages in the proceedings and a poster session that may become a hands-on demo of the chess game and some work my students and I have been doing with Scrabble in Java generating MIDI.

See you soon!