Samplers & sample based synth modules...
No.1 - The EMu Systems ESI 4000 Sampler
An addition to our kit list we would struggle to manage without is the EMu Systems ESI 4000 sampler or "Easy 4K " as it's also known. Purchased to replace our ageing Akai X7000 sampler (very ageing) the Easy had it's first outing at a mini gig way back in April '99 laying in Mellotron 4-4 choir because we were too idle to take the real thing along! It is now used to almost to it's full capacity, especially when we run out of real 'tron sounds in a given song but still need more. Indeed we have now installed the maximum possible amount of sample memory that the ESI can take - 128 megs - and our current sample bank is running at close on 118 megs!!! The grandest sound we use has to be one called "St. James' Bell", a sample we spent an age up a bell loft in Haslingden with a mini disc recorder trying to get right! If you want to hear that particular sound then listen in on "Bells" when we play it - I guarantee you won't miss it!!!
So what exactly is an ESI 4000 ?
So what's a sampler?
Erm..........? If you already know how a sampler works then this might be a good time to skip to the next instrument section of the page
But if you don't.......
Well..... traditional analogue recording media such as the magnetic tape found in cassettes or on older open reel machines uses minute particles of magnetic oxide - rust - to record the voltage fluctuations of an audio signal . Once captured onto the tape we can play them back again whenever we want. The fluctuations of the audio captured by the oxide on the tape is read back out of the playback machine as a varying amplitude voltage. Should there be some form of unwanted interference added to the recording then it is also played back along with the music. In addition tape recordings are subject to quite high levels of hiss due to the size of the magnetic oxide particles on the tape. Naturally such interference is not desirable so a method of encoding the music was needed before recording it onto tape in order to prevent the interference becoming too apparent. The system developed for the encoding process slices the audio waveform into samples several times per second - samples of the amplitude or loudness of the audio waveform.
So now we've sliced the sound into samples, what can we do with them? The next step in the process involves turning the sample values into some sort of code and for this we use a stream of ones and noughts. The reason for using ones and noughts is simply that an electronic device functions best if it is working like a switch, IE switched ON = ONE, and switched OFF = NOUGHT. Any state of "part switched-on-ness" is regarded as a ONE by the electrickery we are using so it's not half so important what might be happening amplitude wise to an electronic ONE compared to how much amplitude problems caused by noise and distortion would affect the un-encoded audio signal. But........ it doesn't take an idiot to realise that a sample which can only be a ONEor a NOUGHT can't tell you how big the slice of audio we are looking at actually is - whether it is just a volt high or a million volts high - so we need some way of assigning a stream of ONES and NOUGHTS to a given range of amplitudes. This is where BINARY code comes in.
Binary means nothing more complicated than doing sums with ONES and NOUGHTS rather than with numbers from nought to ten. Confused? Well think of it like this. If we write down on a piece of paper 1001 we would interpret that as a thousand and one. But we should actually think of it in quite a different way - we should say that it is actually one of a value "thousand" plus none of a value "hundred" plus none of value "ten" and one of value of "one". Add that together and we have one thousand + no hundreds + no tens + one ones - a thousand and one. So each time we move to the left we increase the value by ten times and we also have up to ten levels in each space. It is easy therefore to say what 2222 would be, 2 thousands + 2 hundreds + 2 tens + 2 ones, two thousand two hundred and twenty two. We have all been brought up with the decimal system so it's easy for us to recognise 1001or 2222 or what ever for what they are. Now supposing we had all but one of our fingers and all of our toes chopped off - we could only ever count to one couldn't we? So how are we going to implement a code which can count higher than one with only one finger? By using binary mathamatics!!!
One finger stuck up (your nose, your ear, where ever) must equal one. A binary NOUGHT equals....... surprisingly nothing....... so no finger stuck up equals nought, nothing, zilcho! But what if we say that a ONE indicates two? Hmmmmm............ and a NOUGHT equals no two? Hmmmmm even more. So if 1 = 2 and 0 = NO TWO where do we go from here?
Remember that figure 1001? Remember we said it equals one of a thousand, none of a hundred etc. etc.? Do you also remember that each time we move to the left we multiply the previous level by ten so we jump from 1 to 10 to 100 to 1000 etc.? Let's substitute a 2 for the 10 of our decimal system and then jump up by a multiplication of 2 each time we move to the left. Lets also say that we can only have two levels in each space so we can have the ONE as our quite incorruptable one level and a NOUGHT as the other, our on and off state for electronic devices.
How does this affect counting in binary then? Let's look at 1001 again. From the right, this time we have a one at space one. So that must mean we have 2. Not quite, this is the only odd ball on the snooker table, it actually means we have a one in this instance, Don't worry though, it's not going to happen again, I promise. The next slot up has now gone up by a factor of two so 1 x 2 = 2, surprise, surprise, this slot has a value of two but this time there's nothing in it so there can't be any twos. The next slot again goes up by two times so it must be the slot for fours but there's nothing in that either so it must mean we have no fours.The last slot has a one in it and by applying the same rules as before that slot must equal 2 times the previous slot so it equals 8. So we have one 8 + no 4 + no 2 + one 1. So our binary number 1001 = 8+1 = 9. Adding a nother ONE to the left of the binary number means we have doubled the range of the binary number. So if we assign 8 spaces in total we can give a binary number value to any level from 0 to 255. Further extending it to a sixteen slot number gives us a range of levels from zero to well over 65,500. This is where the phrase 16 bit audio comes from.
So this 16 level binary number - or binary word - will allow us to specify a level for each sample of the waveform in steps one sixtyfive thousandth apart - very small steps in amplitude on an original waveform which may only be swinging a few volts. And if we take those samples often enough throughout a period of a second we can slice the original sound up, encode it as a stream of ONES and NOUGHTS our computer or CD player can understand and then somehow reassemble them again to get back the original sound. And the fact that those ONES and NOUGHTS cannot be corrupted by spikes and splats and that the tape hiss is fairly continuous in level and is therefore not interpreted as a ONE means that we should be able to hear the same out as we encoded and put in to begin with.
That's how a CD player works but a sampler?
Well, if you sample the sound at say 42000 times per second, or 42Khz, and play it back again at the same speed the pitch of the original sound you recorded will play back exactly the same. But if you halve the speed of playback to 21Khz. then the pitch will halve so suddenly a note sampled at say middle C will play back at C an octave below middle C. Double the playback speed and the note will play back at C an octave higher than middle C. Apply slightly more complex divisions or multiplications to the playback rate and you can transpose a sample taken at middle C right across a 6 octave keyboard.
Mind you it does sound a little strange if you do that!!! As an example of this think of the "tddd" female vocal sound Jean Michel used for the melody played back via a Fairlight CMI - a very early sampler - on Zoolookologie. The reason for this effect is that any natural sound has an envelope of amplitude. Think of a bowed string; as the bow hits the string the sound snaps on. For the duration of it's being bowed the sound is of an almost constant amplitude but the second the bow is taken off the sound starts to die away, or decay, at a rate governed by the natural ability of the instruments body to sustain the note. So we have an attack period for a sound, a sustain period, a release point and a decay rate. If the attack period of our bowed string is X thousandths of a second, doubling the playback rate to pitch shift the sample up an octave would cause the attack time to be shortened to X/2, the attack would take only half as long now and the characteristic sound of a bowed instrument would start to be lost. In addition the sound would start to become "chipmunk-esque"! To overcome these shortfalls with samplers we do two things: 1. Sample several times within a given range of notes over which we wish to play back the sample, that way the transpositions of sample attacks are not so great in their effect on the output sound. Typically we sample at intervals of about a fifth for best results. 2. We fit a downstream envelope generator and voltage controlled amplifier to modify the output signal bringing it back within a recognisable sound envelope. This has a knock on bonus effect too in as much as we can mess around with the envelope settings make the sound of one instrument more like that of another should we so desire.
But why does it have no keyboard? Well, the Easy4K is a midi controlled, rack mount format instrument and as such it is designed to be played by an external midi controller keyboard or by a sequencer. We currently use our ESI in both modes, in the former by controlling it with a Roland AX1, the ultimate poseur's keyboard, and also with fixed, none portable MIDI mother controller keyboards such as the 5 octave Evolution model shown here. We also used to use a MidiMan Transmidi wireless transmission system with the AX so that the player had complete freedom to wander about the stage without fear of tripping over his/her MIDI cable however both our Transmidis have recently died, taking out the radio mic transmitters they also use in the process. We have now replaced one of these with the new Kenton system.
No.2 - The EMu Systems Virtuoso
& the rest of the Proteus family...
The earliest samplers such as the Fairlight CMI were incapable of capturing sounds of a duration much longer than a few seconds . Multi-sampling, the practice of placing several sound bytes sampled from the same instrument across the entire note range to avoid the chipmunk effect, was unheard of. This was due almost entirely to the cost of memory chips at that time. And the key to a sampler's ability to capture lengthy segments of sound is the amount of sample memory in the beast. Nowadays memory is incredibly cheap compared to the heyday of the Fairlight and it is not uncommon for an instrument to boast hundreds of megs of RAM where tens of kilobytes of memory were the quotas referred to with awe back in the heady days of 1981!
But those early days left us with a legacy we now make best use of - the ability to use sample memory in an economical manner. And no company is better at this than one of the founders of modern sampling, EMu Systems who were right there at the beginning with the EMulator. They have not just continued with their pure sampler line however, they have brought in a whole range of pre-loaded sample play back instruments which use short snips of sound taken from different instruments right across the full spectrum of modern (and not so modern) sound creation. This range is based around their Proteus which has evolved greatly since its birth some years ago.
The basic Proteus has a broad spread of instruments on board, similar in many ways to the GM (General MIDI) specification and it can best be likened to a general duties, all purpose work horse. But there are other far more specialised instruments within the Proteus family too. If hip hop is your bag or trance or dance, all these gendres of music are catered for with the Mo-Phatt, X-treme Lead etc. The B3 carries a host of electronic organs, both draw bar Hammonds and conventional electronic instruments like the Vox and Farfisa. And the Vintage Instruments carries everything from Mellotron sounds, through Simmons electronic drum kits to the ubiquitous Solina. Orchestral sounds abound on the model we use, the Virtuoso, shown above.
But the beauty of the entire Proteus family is the way that each instrument works around the four memory card slots which take expansion boards pre-populated with fresh sounds. Most of the range have only one board pre-installed when you buy them, the Virtuoso is the exception with two of the slots taken up to cover the full range of orchestral instruments and to provide many which have also been played in alternative styles - pizzicato and marcato strings for example. With 32 channel MIDI capability it is possible to create a very accurate sounding virtual orchestra. And best of all, additional boards can be purchased from EMu dealerships such as Arietta Music to turn your particular Proteus family instrument into any of the others within the range.
It is the Vintage Instruments card and the World Instruments card we have chosen to install into our Virtuoso's two free slots. Thus in one instrument we have a full orchestra, a host of vintage instruments and a vast library of ethnic instruments and percussion from all around the world. And we have the ability within our rig to dedicate an entire MIDI output port of 16 channels just to this one instrument to allow us to play with an orchestra at the same time as we may chose to use a Mellotron sample set and say an ethnic percussion ensemble! In view of Jean Michel's love of ethnic instruments and his use of practically every synthesiser since they first became main stream in the early seventies we are well provided for with this 1U rack mount instrument! .