I was born at a unique time (1968). I was born during the transitor age, but just before the silicon chip age.
Anyways, when I was still quite little, I noticed electronic organs were downright crappy. For example, you’d switch to clarinet. But the recreation sounded nothing like a clarinet. That is why when I was still a child I came up with an idea.
Magnetic tape recorders (cassettes, 8 tracks, etc.) were all the rage when I was still young. So I thought (when still a child), why not have the various notes of the electronic organ in the form an unending loop of magnetic tape, that recreated exactly whatever sound you are trying to reproduce? I was also recently thinking, you didn’t even have to do that. Just have it as a unending groove in ordinary record-player type recording.
Anyways, no one ever did this. So I have to wonder if my idea is/was even viable. I seriously wonder to this day. Could you actually do something like I just described? And why hasn’t anyone tried this yet? (BTW, FWIW modern electronic organs do recreate sounds exactly, without the tape or record loop though. My question, though, is just about my idea.)
Something like this?
Most people would recognize this instrument as providing the flute sounds for the Beatles’ Strawberry Fields Forever, but it was used on many, many albums in the 60s and 70s.
The problem with this is, it’s the chaotic sound when a note is first struck or sounded that gives a particular instrument much of it’s character. That’s one reason a basic synthesiser can sound very flat compared and uninteresting to a natural instrument. Just having a loop of of the recorded instrument would exhibit the same problem. That’s why the Mellotron doesn’t play in a loop, these attack sounds could be recorded and reproduced , but the drawback is that it will only play each note for about eight seconds before it runs out of tape.
…which itself was rendered ‘obsolete’ because of sampling technology.
Interestingly, as Alka Seltzer pointed out, the main thing all these instruments have shown is that the characteristic sound of an instrument has far more to do with phrasing, attack and nuances that the player of the real instrument has learned to control, but which require a ridiculous amount of fussing to bring out on an electronic instrument.
Sometimes playing them through tube amps close to the point of distortion can give them more character. I think that’s what’s happening in this White Stripes performance. I particularly like the percussive sound the organ is making through the opening bars.
And there are a whole host of other nuances. Most instruments sounds different depending on how hard they are played. Plucking a guitar string softly vs. loudly is not just a difference in volume/amplitude. You get different sounds, as the more loudly plucked note has more audible harmonics than the softly plucked string. Hit a piano key softly, and there’s this mellow, almost muted quality to it. The sound is soft and a little dampened. Bang on it and you get bright, sharp sounds will a full spectrum of overtones. If you took a passage played with a soft touch and simply increased the volume so it’s as loud as one played with a loud touch, they wouldn’t sound the same.
Most of this sort of stuff is modeled in synthesizers through multi-sampling, by recording each sample at a range of attacks, and then applying some flavor of ADSR (attack, decay, sustain, release) envelopes to them which basically govern how amplitude and frequency response change over time. And it wasn’t necessary to sample every single note. You could get a reasonable sounding piano with a single sample for every, I dunno, third or so note, and then just scale those notes to fill in the missing keys. And, if you don’t utilize multi-sampling for dynamic levels, you can get a reasonable approximation using just ADSR envelopes and mapping velocity (basically, how hard you strike the key) to them.
And that’s modeling something relatively simple with fixed, discrete pitches like a piano. With many other instruments, you have a lot more options as to how to play the note. You can bend it, you can slide it, you can add vibrato, you can bow it vs. pluck it, you can mute it, you can play it as a harmonic, you can hammer-on or pull-off, etc.
In other words, making an instrument sound like that instrument is not going to simply being a matter of recording the notes you need on tape, and then playing them back in whatever order you need.
That said, the sounds this crude sort of sampler makes are quite interesting in their own right–I just wouldn’t think of them as really trying to be a substitute for the real thing.
A few years back, there was a lot of talk about using physical modelling to synthesise sounds. If you have an accurate physics engine, it should be possible to emulate the characteristics of real or even imaginary instruments. I don’t know if it seriously took off though, anyone know the latest on this?
Most of the modern crop of synthesizers use a combination of sampling and modeling to create “authentic” reproductions of instruments.
So, it might take the samples of a piano, and then use modeling to smooth out the transition between samples of different attack velocities, or to add the harmonic complexities of the sympathetic vibrations of the other strings.
My electric piano (circa 2001) has hundreds of voices and models lots of things mentioned here. For example, on the voice Sweet Trumpet, if you hold a key, the trumpet sound quavers out after a suitable pause. if you don’t hold it that long, no quaver!
This instrument throws hammers, but they don’t hit strings, just some felt. There are sensors that detect how fast the hammer moved and both the attack and the loudness are adjusted.
There is some recording ability, too, not that I have used it. But there entire voices designed to be used for improving the realism of recordings. For example, there is an 88-key voice of guitar squeaks, usually made when a guitarists fingers slide down a string. You could record a track with a few squeaks to make your overall guitar recording sound more authentic.
