Circuitbenders Forum

Please login or register.

Login with username, password and session length
Advanced search  

News:

Author Topic: speak n spell theory  (Read 9667 times)

brewmachine

  • Newbie
  • *
  • Karma: 2
  • Offline Offline
  • Posts: 12
speak n spell theory
« on: March 20, 2009, 07:10:58 AM »

hi,
just wondering if anyone here knew any actual theory behind circuit bending, i know its pretty much all 'anti-theory' as coined by RG but if anyone has any useful links or info on the speak n spell n casio SA-2 bends and why they do what they do, it would greatly help my dissertation.
cheers
bru
Logged

Gordonjcp

  • This person is dangerously insane.
  • *******
  • Karma: 78
  • Offline Offline
  • Posts: 1005
    • http://www.nekosynth.co.uk
Re: speak n spell theory
« Reply #1 on: March 20, 2009, 09:56:14 AM »

Basically what happens is that you are altering the way in which the circuit works.  For instance, by adding another capacitor in parallel with a timing capacitor, you slow down the timing, which might make an oscillator run slower or slow down an envelope.  By shorting two pins on a chip you might put it into an odd mode, not used in that piece of equipment, or you might crash the CPU in a computer-controlled device causing it to send garbage to the sound chip.  With a ROM-based sample-player instruments (like the Alesis HR16) then if you earth one of the address lines, you will only play back half the sample.  Which one depends on which pin you earth - if you earth A0 then you will only ever play back even-numbered samples (half the bit-rate).  If you earth A3 then it will repeat groups of eight samples twice, and so on.  The really interesting thing about bending sample-based stuff is that it is entirely predictable - you can always tell what it's going to do, and have some idea of the effect you'll get.

Now, with the Speak'n'Spell you've got a simple microprocessor which slings a string of bits to a sound chip.  You'd be surprised at how simple it is.  Read some data (not samples, we'll come to that), push it out a serial port, and then the sound chip will turn it into Dalek-y speech.  Simple.  When you crash the CPU it runs off into the weeds and starts slinging odd fragments of words mixed with bits of the program code out to the sound chip, with hilarious consequences.

To pack recognisable speech into the low-capacity ROMs available in the 70s and 80s, Texas Instruments used a codec called LPC-10.  Now, this is not so much like a sample as an MP3 in that the data doesn't directly represent the output waveform, it describes how to generate it.  The data encodes pitch, timbre and volume information into packets that are turned into sound by what is more-or-less a complex hybrid digital/analogue synthesizer.  Now, one of the the interesting things is that you can get software implementations of the LPC-10 codec, and run them on your PC.  You can have a lot of fun compressing and uncompressing stuff, and introducing random bit errors to the data.  Or, you can be totally bizarre and use LPC-10 to play back utterly unrelated data, like this:
http://www.gjcp.net/~gordonjcp/lpc10.mp3

I'm planning on (finally) launching my new website in a couple of days, once I get a few more articles written.  There's quite a lot of technical information on how sounds are generated, which might help you.  I'll post a link later.
Logged
If at first you don't succeed, stick it through a fuzzbox.

ne7

  • Dayglo Volunteer of International Masturbation
  • ***
  • Karma: 17
  • Offline Offline
  • Posts: 142
    • ne7 web
Re: speak n spell theory
« Reply #2 on: March 23, 2009, 09:00:05 PM »

gordon thats so seriously brill info on the speak and spell codec stuff!!! :D
Logged
ne7/triad
------------------
http://ne7.untergrund.net

Gordonjcp

  • This person is dangerously insane.
  • *******
  • Karma: 78
  • Offline Offline
  • Posts: 1005
    • http://www.nekosynth.co.uk
Re: speak n spell theory
« Reply #3 on: March 23, 2009, 10:17:40 PM »

Feel free to use bits of that sample, btw.  It is (like all my samples) CC-BY-SA, but if you ask very nicely I might grant you CC-BY (typically if you want to use bits of my samples in commercial work, for which I'll want *something* - I like real ale).

It's particularly good if you timestretch it to slow it down with something like rubberband.  It's got an oddly compelling speech-like quality that grabs the ear, but makes no sense.
Logged
If at first you don't succeed, stick it through a fuzzbox.

computer at sea

  • Closer to the meat
  • *****
  • Karma: 20
  • Offline Offline
  • Posts: 301
Re: speak n spell theory
« Reply #4 on: March 23, 2009, 10:54:55 PM »

Quote
It's got an oddly compelling speech-like quality that grabs the ear, but makes no sense.

Can I get that on my tombstone?
Logged

brewmachine

  • Newbie
  • *
  • Karma: 2
  • Offline Offline
  • Posts: 12
Re: speak n spell theory
« Reply #5 on: March 30, 2009, 11:02:59 PM »

wow, thanks 4 that mate, that's quite a lot to digest in one sitting! All of it is a fantastic help to me though... your a star  ;)
keep me posted about your website too, sounds interesting.
thanks again
brew 
Logged

goldenbaby

  • Dayglo Volunteer of International Masturbation
  • ***
  • Karma: 1
  • Offline Offline
  • Posts: 112
Re: speak n spell theory
« Reply #6 on: April 07, 2009, 11:54:39 PM »

Hah, what luck!  I just picked up my first Speak & Spell at the thrift store today.  It was in great condition, and only cost $2.50 CAD.  What a deal.

It's of the "Aimless" variety shown on that website.  Are the bends written on the picture the only ones available?  I was already having visions of doing my first patchbay.
Logged