Digital Sampling And Guitar
By Frank Zappa
LAST MONTH WE were discussing the guitar-related aspects of New England Digital's Synclavier digital keyboard synthesizer with a guitar interface. Can you actually feed in a guitar sound if you want to store that?
Well. in order to make a real guitar sound, you have to use the part of the synthesizer that does sampling. That's where you store a note specimen from any sound source in the real world, and then that note gets transposed up and down by the keyboard.
Does it sound fishy when you retrieve the sample in distant octaves such as, say, at the outer ends of the keyboard?
It depends on how well you sample the specimen. In other words, if you're sampling a note over five octaves, it's going to sound fishy. But if you sample each string individually, it will sound better. For instance: You could take your low E string and play open E, and play a B, and sample both of those notes. If you dedicate that to just the real octave that string would play in, I don't think you would hear any difference over an octave span – if you sampled each of your strings in that way. It means storing a lot more samples, but if you want a realistic guitar sound, that's the best way to do it. Once the sample is in there, you can electronically add vibrato to it, you can add tremolo to it, you can add pre-planned pitch bends that stretch an octave – really ridiculous stuff.
We have a trumpet sample in the Synclavier now, and the guy who's programming for me, Steve [De Furia], did a tweeze on it that he calls "wazzing" the note. He set it up so that once the sample is initiated by the keyboard, it starts to rise an octave; and it will go all the way up an octave if you keep your finger on the key long enough. But if you take your finger off, or if you're playing fast lines, the result is this bizarre sort of thwarted linear desire: It keeps trying to rise, but it never gets all the way up to the octave. It makes the strangest effect on the trumpet sounds. I imagine it would do just as well on any string sample. So far we've sampled some great low notes on a Bösendorfer piano, huge bass drums, castanets, snorks, wu-han cymbals, chimes – all kinds of stuff. And you're able to play scales of these things after they're in there, and they sound completely real. It's not like the Fairlight [another digital synthesizer], which, to me, sounds kind of like cardboard when you put things in sample. Synclavier sampling is much more realistic because of the 50k [50,000 cycles per second] sampling rate.
Are you able to sit down and say, "I want a rhythm of a certain number of sixteenth-notes" and set other parameters, and then assign the sound to that scheme?
So you could actually spend three or four days putting in a peculiar set of rhythms and then then punch a button and let it run.
That's right. But it doesn't take three or four days.
What if you're doing, say, a long piece of music?
If you were doing a very long piece of music, it would. I've developed a sublanguage for the system that accommodates the type of chords and rhythms that I prefer to use. There is already some software in there that accommodates certain compositional functions, but it's all general-purpose stuff. So Steve and I figured out a fast way to store the basic elements that are most useful to my style, and stored it in building-block form in order to expedite the process of typing in a composition alphanumerically [using the letters of the alphabet and numbers 0 through 9] as opposed to keyboard loading. I can conjure up rhythmic and harmonic nifties very quickly this way, and save on some of the drudgerous aspects of compositional permutation.
Read by OCR software. If you spot errors, let me know afka (at) afka.net