2. Muziek en instrumenten

Demonstructie van een handgemaakt pijporgel

Sun, 01 May 2011 11:50:00 GMT

Fijne stemmechanieken

Wed, 16 Mar 2011 22:16:00 GMT

Atarischennis

Thu, 30 Dec 2010 22:53:00 GMT

For reference purposes: Atari 800xl pitches for the PAL version

Thu, 16 Dec 2010 00:46:00 GMT

The Atari 800xl (as well as the 400, 800, 600xl and XE) has a very simple sound chip. It is not 8 bit like your sound interface is 16 bit or 24 bit in resolution now. It cannot make a sine-like wave with an 8 bit resolution. It has 4 sound channels, each can tap into one of these two clocks: At 64kHz, or if you poke a memory location, 15kHz. Two of the three can tap into the main clock, which is slightly lower for the PAL version: ±1.77Mhz. I assume the 64kHz and 15kHz clocks are also related to the clocking of the video, hence different for the PAL and NTSC versions.

The clocks are then devided by a number, which is expressed in 8 bit. The result is a square wave with a 1 bit resolution. This square wave can then be changed in volume in 16 steps, 4 bits deep. So the endproduct is technically 4 bits deep, but this 4 bit resolution is not available to the four individual streams!

The devided clock frequency results in a simple square wave. This can be made noisy. In the Atari, there are 3 random number generators. They are not really random, they repeat a noisy series of zeros and ones. These series of zeros and ones each have different lengths: 16 numbers, 32 numbers, and the third can be switched: 512 numbers or about 128.000 (17bits). The last one sounds the most like white noise.

The devided clock can be used as a sample and hold for these noise generators. This way, the maximum frequency in the noise can not be higher than the pitch by which you sample it. If you use the low numbers, you will notice the result is pitchy, much like a ring modulator. This is especially true for the shortest noise generator, called distortion number 12 in Basic.

You can then patch in X-OR circuits, by a poke command, combining channel 0+2 and 1+3 respectively. This results in a sort of crude high frequency filter, as the result can never be lower than the pitch of devider 2 and 4 (which then are, I assume, not used to sample noise)

Another way of combining audio channels is by patching two deviders in succession. So the output of devider 0, is then devided again by devider 1, thus resulting in a lower pitch. I will need to test if any or both of the deviders can be used for the noise sampling. At least bass material can be made this way.

In atari music, the 4 channels are often used as follows:

Channel 0 devided again by channel 1, for bass.
Channel 2, used to sample the large noise generator, for rhythms.
Channel 3, used for a high pitched melody.

Please note that deviding pitches will never ever produce an equal temperament scale. In fact, you will get some very pure fifths, but some others will be WAY off. Also, the higher your pitch at the output, the smaller the number is you devided with. So at high pitches, your pitch resolution is too small. The upside of this system, is that you will find lower octaves of the same value in cents as the ones you will find in higher pitches. So I looked into this, and worked out the Atari temperament down to the tenths of cents. This is something you will not find in any manual, which is why I post it here. I measured the standard 64kHz clock devisions, so no poking yet. Stay tuned for an update on those.

note, difference with 12-tone-ET, devision for several octaves.

B2 +3.2 cent, 255
C3 +0.6 cent, 241 120
C# +3.6 cent, 227 113 56
D3 -2.6 cent, 215 107 53 26
Es -3.6 cent, 203 101 50
E3 +1.3 cent, 191 95 47 23 11
F3 -6.1 cent, 181 90
F# -8.3 cent, 171 85 42
G3 -4.6 cent, 161 80 39
G# +5.6 cent, 151 75 37 18
A3 -0.7 cent, 143 71 35 17 8
Bb -1.7 cent, 135 67 33 16
B3 +3.2 cent, 127 63 31 15 7

Note that F, F# and G# are the furthest away from 12 tone ET. The rest is usable.

These are the distances between the notes in cents:

N second minor3 major3 forth- fifth-- majsix minsept

C 196.8 - 295.8 - 400.7 - 493.3 - 694.8 - 898.7 - 997.7
# 192.8 - 297.7 - 390.3 - 488.1 - 702.0 - 894.7 - 999.6
D 203.9 - 296.5 - 394.3 - 498.0 - 701.9 - 905.8 - 1003.2
# 197.5 - 295.3 - 399.0 - 509.2 - 701.9 - 904.2 - 1007.2
E 190.4 - 294.1 - 404.3 - 498.0 - 701.9 - 902.3 - 996.1
F 201.5 - 311.7 - 405.4 - 504.4 - 706.7 - 903.5 - 1002.5
# 213.9 - 307.6 - 406.6 - 511.5 - 711.9 - 904.7 - 1009.6
G 203.9 - 302.9 - 407.8 - 505.2 - 702.0 - 905.9 - 998.5
# 192.7 - 297.6 - 395.0 - 498.0 - 690.8 - 888.3 - 986.1
A 203.9 - 301.3 - 404.3 - 498.1 - 702.0 - 892.4 - 996.1
# 202.3 - 305.3 - 399.1 - 498.1 - 695.6 - 897.1 - 1007.3
B 200.4 - 294.2 - 393.2 - 498.1 - 688.5 - 902.4 - 996.1

The purest notes:
Second: 204cent. I'm happy to find these.
Minor3: 298cent or 316. There are many different thirds here, but some are really close to pure.
forths: 498.1cent There are, logically very pure forths.
fifth: 701.9cent Note that some are way off, but many are very pure Pythagorean fifths!

Since I prefer to play in Dorian: What is the purest minor key? It's Dm

D3 -2.6 215 107 53 26
E3 +1.3 191 95 47 23 11
F3 -6.1 181 90
G3 -4.6 161 80 39
A3 -0.7 143 71 35 17 8
B3 +3.2 127 63 31 15 7
C3 +0.6 241 120 60*

My Joystick mapping:

10 JOY=STICK(0)
20 IF JOY=15 THEN POKE 53760, 160 : GOTO 10
30 IF JOY=13 THEN N=215
40 IF JOY=9 THEN N=191
50 IF JOY=5 THEN N=181
60 IF JOY=11 THEN N=161
70 IF JOY=7 THEN N=143
80 IF JOY=10 THEN N=127
90 IF JOY=6 THEN N=120
100 IF JOY=14 THEN N=107
110 IF STRIG(0)=0 THEN N=INT(N/2)
120 POKE 53761, N : POKE 53760, 175
140 GOTO 10
150 REM RUN turns the sound off, GOTO does not!
160 REM Note 120/2=60 this is 14cents too low, but usable in this role.

Notes:
The SOUND command in BASIC resets all interesting POKEs you did into 53768 (Audio Control)
RUN shuts off audio that is currently playing, GOTO does not.

POKE 53768 controls the audio patch:
128 : Makes the long 17 bits pseudo random "poly noise" word 9 bits long.
64 : clock Channel 0 with 1.76 MHz CPU clock, for higher pitches.
32 : clock Channel 2 with 1.76 MHz CPU clock, for higher pitches.
16 : clock Channel 1 with Channel 0 for lower frequencies, preciser pitches.
8 : clock Channel 3 with Channel 2 for lower frequencies, preciser pitches.
4 : Feed Channel 0 and Channel 2 through an X-OR as an highpass filter.
2 : Feed Channel 1 and Channel 3 through an X-OR as an highpass filter.
1 : Set all 4 clocks to 15 kHz (horizontal refresh rate) for 2 octaves lower sound.
Add the numbers of the options you want, and poke these into 53768.

POKE addresses per Channel:
Channel, frequency - distorion and volume
0, 53760 - 53761
1, 53762 - 53763
2, 53764 - 53765
3, 53766 - 53767

The distortion value in Basis is not really distortion. It's a 1 bit stream and can't be distorted like an analog signal can be, by definition. This is the noise sampler. The following table is the numbers to POKE (from silence to high volume) => Binairy code => then the number you would have used in the sound command => : function. I left out similar fuctions, and ordered it by noisyness.
160 - 175 => 1010xxxx =>10 => : Pure square wave
192 - 207 => 1100xxxx =>12 => : 4-bit poly noise, sounds like a ringmodulator
032 - 047 => 0010xxxx => 2 => : 5-bit poly noise, sounds a bit like a ringmodulator
064 - 079 => 0100xxxx => 4 => : 5-bit/4-bit poly noise
000 - 015 => 0000xxxx => 0 => : 5-bit/9 or 17-bit poly noise
128 - 143 => 1000xxxx => 8 => : 9 or 17-bit poly noise, 17 sounds really noisy.
240 - 255 => 1111xxxx =>15 => : The input into the volume is always one, making the volume a control voltage, 4 bits deep. I have not yet checked of the output is DC coupled.

10 JOY=STICK(0)
11 P=(PADDLE(2)):Q=(PADDLE(3))
12 Q=160+INT(Q/15)
13 ? N,P,Q
15 IF PTRIG(2)=0 THEN POKE 53768, 5
16 IF PTRIG(3)=0 THEN POKE 53768, 1
20 IF JOY=15 THEN POKE 53761,160 :POKE 53765,160: GOTO 10
30 IF JOY=13 THEN N=215:GOTO 100
40 IF JOY=9 THEN N=191:GOTO 100
50 IF JOY=5 THEN N=181:GOTO 100
60 IF JOY=11 THEN N=161:GOTO 100
70 IF JOY=7 THEN N=143:GOTO 100
80 IF JOY=10 THEN N=127:GOTO 100
90 IF JOY=6 THEN N=120:GOTO 100
100 IF JOY=14 THEN N=107
101 IF N=NOUD THEN GOTO 10
102 NOUD=N
110 IF STRIG(0)=0 THEN N=INT(N/2)
120 POKE 53761,175: POKE 53760,N
130 POKE 53765,Q: POKE 53764,P
140 GOTO 10

Dansgeschiedenis

Sun, 03 Oct 2010 20:21:00 GMT

Dit gedigitaliseerde book over dansgeschiedenis is inmiddels zelf ook weer een ikoon uit de geschiedenis. Het is daardoor in het publieke domein, dus vrij up- en downloadbaar. klik! De plaatjes zijn aanklikbaar voor hoge resoluties.

Sweelinck's Fantasia Chromatica in 3 manieren van stemmen

Sun, 14 Feb 2010 02:48:00 GMT

Maxwell Steer heeft 3x hetzelfde stuk gespeeld, op hetzelfde instrument, dat op 3 verschillende manieren gestemd is.

In Equal Temperament, zoals u het kent. Eenvormig en flets, vind ik. De stemming was bekend in Sweelinks tijd, maar werd afgewezen. Bovendien kon het alleen bij benadering gestemd worden.

In Werckmeister III, een vroege wohl temperierte stemming, met een veel spannender pallet dan Equal Temperament, met een royalere klank dan meantone. Het klinkt me doorlopend correct in de oren. Toch is deze manier van stemmen pas zo'n 70 jaar na Sweelinks dood uitgevonden.

In 1/4 komma middentoonstemming. Mooie kwarten en drieklanken, maar af en toe wel erg rasperig een zwevend. Toch was dit een Sweelinks tijd een heel gebruikelijke manier van stemmen. Wellicht is het idee van deze muziek ook wel dat wringende klanken zich oplossen in prachtig liggende drieklanken. Vooral het snellere deel tegen het einde klink echt heel strak, tot het zachter wordt. Dan volgt een spannend eind.

Electric Tabla Jam

Mon, 31 Aug 2009 18:48:00 GMT

My electronic tabla (it says "Riyaz proffessional") controls my home built 32 step sequencer, of which 20 positions are used for this rhythm. The Sool rhythm has 10 beats per cycle.

The sequencer not only plays these voltages up or down, but it can also derive other patterns. I worked these out and implemented these in hardware. I would have to look up how I did this. I do not think of it anymore, as each of the patters is useful. So in a live situation, I just choose a different one at random.

The output of the sequencer is analog, and sends out 0-10 volt. This is then converted to midi by my new miduino, a circuit by Tom Scarff. The software was partly rewritten by me. To make it play in the key of my choice. Using a very secrative math trick called rounding, I made the sequencer play in a minor key, rather than in the 12 tone system. This allows for live tweaking of the knobs without ever being out of key. Freedom!

Then there's my Arcade joystick, which has an home baked circuit that converts the 4 switches and the fire button to 15 different notes, and connects right onto the keyboard matrix of an old casio childrens keyboard which cooks midi note messages. I would do this in the miduino too if I needed to do it again. A series of if/else statements is much easier and more flexible than soldering and debugging a circuit with 8 logic IC's and a over a meter of wire.

Then there's the large joystick. It has a Doepfer midiconvertor on board. The miduino has enough pins to do it all together, but I like the fact that it is a funtional unit on it's own. It controls all the effects of the Arcade channel.

The pitch of the tabla can be controlled by the sequencer as well as by the Atari paddle. I made this input on the tabla when I replaced the tuning pot with a precise ten-turn pot. It works with two vactrols.

Sound sources are: cymbals, a Raagini shruti box, the tabla box, a Waldorf Blofeld running 2 patches amplified by a 40 year old Italian tube amp.

I recorded it with my photocamera, which has a rediculously low audio sampling rate. They all have. It's just not a number the camera's are sold by. I should have checked weather I was in the image myself. Now I am a pair of hands on youtube too, looking like the Lorax of Dr. Seuss. Ah well. It's just a test drive to check if my work is getting somewhere.

EP gaat maandag stemmen of Brein je offline mag laten gooien.

Sun, 06 Jul 2008 21:26:00 GMT

EP stemt maandag over wet die het afsluiten van downloadende internetters regelt, een maatregel die kan worden opgelegd zonder dat er een rechter aan te pas komt. Als je ISP drie keer een klacht ontvangt over downloaden, moeten ze je offline gooien.
Lees er meer over op:
http://retecool.com/post/de-eu-gaat-het-mogelijk-maken-dat-brein-je-offline-gooit

Ik heb Konrad Boehmer van de BUMA gebeld, en die bleek fel gekant tegen het voorstel. Het druist tegen alle rechtsbeginselen in die we in de afgelopen 200 jaar hebben verworven. Lees er meer over op:
http://retecool.com/post/wat-vindt-boehmer-van-dit-voorstel

Met deze link kun je alle Nederlandse Europarlementariers tegelijk verzoeken tegen dit wetsvoorstel te stemmen.

Waar kan die microfoon nog meer in?

Wed, 31 Oct 2007 13:46:00 GMT

In het kader uitvindingen voor in het theater: Een danseres wil haar kauwgomgesmak uitversterken. Dus beschaafd met haar mond dicht kauwen, en toch luid gesmak. Een lavaliermicrofoon bij de mondhoek had geen zin, want daarmee hoorde je het gesmak alleen bij open mond. Bij gesloten mond was er niets te horen. Ik kreeg een idee: de microfoon in het oor. Dat werkte prima! Enthousiast geworden door het succes werd de microfoon ook nog in neus geprobeerd. In de neus was het geluid harder, maar daar ademt ze ook door en dat was een enorm kabaal. Om toch het luidere geluid uit de neus te kunnen gebruiken, heb ik de microfoons afgedicht met krimpfolie tussen het grid en de microfoon zelf. Dat blokkeert het ademlawaai en voorkomt een snot-out, de nasale variant van de sweat-out (microfoonsterfte door zweet) zoals die op het voorhoofd kan optreden. Een geslaagd experiment.
Ik ben nu benieuwd waar ze voor gaat, het harde geluid uit haar neus, of het schonere maar zachtere geluid uit haar oren. Aangezien de draadjes naar de oren helemaal weg te werken zijn in het haar en draadjes naar de neus er uit zien als een zuurstofvoorziening, zal het visuele aspect bij die keuze een grote rol spelen.

Wiitaar

Sat, 07 Jul 2007 19:15:00 GMT

Een Wii remote aan de kop van een gitaartje gebonden, darwiinremote OSC + MSP aangezet, even programmeren en uren en uren tweaken: Daar is de Wiitaar. Instant jiminess zonder gedoe met bandrecorders.