Magazine Archive

Home -> Magazines -> Issues -> Articles in this issue -> View

Ensoniq Performance Sampler

The ability to load new samples whilst playing is only one of many performance-related features which make the 20-voice EPS stand out from the current crowd of sampling keyboards. Craig Anderton reveals the others...

This is not Mirage II: the EPS is a quantum improvement in sampling technology for Ensoniq and, in many ways, for sampling in general. This is the most impressive sampler I've seen at the sub-£2,000 price point, not just because it sounds good, but because it is deep. Brilliantly designed and filled with exceptional creative opportunities, I have a feeling this is the instrument that will break the sampling market wide open.

The keyboard is the same as that of the Ensoniq SQ-80, having polyphonic aftertouch, and all comments in that review [see page 8] apply to the EPS. But I'd also add that if you think having individual articulation for each note of a synth is fun, imagine what it can do for the normally overly-static sound of samples.


When you sample a sound with the EPS, you create what Ensoniq calls a wavesample. One or more wavesamples placed appropriately on the keyboard make up an instrument, and each wavesample can be panned anywhere within the stereo field. Let's dally for a bit at the instrument, which also includes eight layers. These layers are variations on the instrument - possibly consisting of some different wavesamples, different filter settings, different envelopes, etc. You can combine any number of these layers to make up a patch (up to four patches total); these patches are brought in with two patch select buttons located right above and to the left of the pitch wheel (as binary arithmetic fans will note, two buttons gives four combinations).

Up to eight instruments are available at any given time. Data defining the set of eight instruments, plus a sequenced song (described later), can be saved as a bank.

Doubling sounds live is simple. To double one instrument with a second instrument, just double-click on the second instrument button, and you're doubled. Want to add a third instrument? Double-click on its button, and now you have tripling, in fact, you can stack all eight instruments if desired.

In case you're concerned that all this doubling and such will gobble up the available polyphony, note that the EPS has 20 voices. Although you can play 'only' eight voices per key, per instrument, you can literally hold down eight keys, select a new instrument, play eight more keys on top of that with a different sound, let them sustain too, and select yet a third instrument and play four notes on top of the already sustaining notes from the other instruments!

When stacking sounds, this also means that you have full ten-voice polyphony if you layer two sounds, and five-voice polyphony if you layer four. With the EPS, you don't have to think too much about notes cutting each other off. And if you don't need all 20 voices, you can reduce the number of voices to increase the playback rate and thereby improve fidelity. For example, with 20 voices the playback rate is 31,2kHz, which yields a 15kHz frequency response. While this is quite respectable, selecting 12 voices allows a playback rate of 52kHz, thus allowing a full 20kHz bandwidth.

Figure 1.

The EPS 'keyboard range' is also an interesting concept. Each instrument can cover a specific range of the keyboard; that's nothing special. What is special occurs when you start loading different instruments on to the keyboard (up to eight maximum). Ensoniq uses the model of overlapping sheets of paper (each 'sheet' or keyboard range can span between 1 and 88 notes). For example, Figure 1 shows one possibility: a one-octave electric bass instrument, four-octave guitar instrument, one-octave bell instrument, and two-octave piano bass instrument. As shown, you would hear only the exposed top 'surfaces' of the layers - bass and guitar (Figure 1 shows a side view of the 'paper sheets').

Figure 2.

Now suppose you select the bell instrument. It now goes to the top of the pile, and replaces the top octave of the guitar sample. Finally, let's select the piano bass sound (Figure 2). It replaces the electric bass and the lower octave of the guitar, so our keyboard now provides two octaves of sampled piano bass, two octaves of guitar, and one octave of bells. This technique allows for up to eight split points of different instruments, with each split capable of playing up to 20 voices (voices are dynamically assigned) - pretty significant stuff, if you ask me. And I might add that changing an instrument's range is a very simple operation, as are almost all the performance-oriented features.


The EPS is a 13-bit machine, although all internal processing uses 24-bit resolution. Thanks to floating-point output conversion, the dynamic range is an excellent 96dB. Although 13 bits isn't 16 bits, Ensoniq need make no apologies. The EPS does seem cleaner than 12-bit machines, and thanks to some clever design elements discussed later (like sample interpolation), the sound quality is definitely pro level.

Regarding memory, the basic model includes 480K of internal RAM. This allows 4.95 seconds of sampling at a 52.1 kHz clock rate and 41.7 seconds at 6.25kHz. Memory can be expanded by the user to 896K o r2.1 Megabytes; in the latter case, you get 22.9 seconds of sampling time at 44.6kHz, and an optional SCSI port. The SCSI port will presumably be used for such tasks as dumping samples to hard disk (the EPS is set up to work with generic Apple-compatible hard disks). Apparently, the EPS is also capable of recording directly on to hard disk, but the folks at Ensoniq wouldn't elaborate except to say that they didn't want to emphasise that feature too much just yet. Hmmm...


Instruments can be saved to disk as instrument files, and deleted or (of course) loaded. There is one awkward aspect to disks, however. The EPS contains no operating system (OS) of its own, so before the EPS will do anything the OS must be loaded from disk. There are also times when, during the normal course of operation, the EPS will need to access the OS from disk. It would therefore seem that putting the OS on all your disks would be a good idea - except that the OS takes up about 10% of the available disk space. So if you want to cram the greatest number of instruments on to a disk and not include the OS, then boot from a disk that contains the OS, load the instruments from an instrument disk, and be prepared to re-insert the OS disk from time to time. By the way, as with the Ensoniq SQ-80 - the disk drive also stores System Exclusive data dumps of up to 256K - long enough for just about anything except extremely long sample dumps.


Good news: you don't have to think in hexadecimal any more when using an Ensoniq sampler! The EPS boasts a blue fluorescent display similar to the one on the ESQ-1 or SQ-80. However, this display has little words that light up when you're in certain modes, kind of like the displays used in video recorders and some microwave ovens.

The EPS uses the familiar page-driven user interface, where you select parameters with cursor buttons and use a data slider (or increment/decrement buttons) to change values. It's not hard to figure one's way around the machine; for example, the preliminary manual I received with the unit said nothing about voice editing, but I was able to figure out about 90% of the functions anyway. It's easy to catalogue files, easy to check memory, and generally very easy to use (although considering the sheer number of parameters, a computer-based editor would be welcome). I do have one suggestion, though: a 'cheat sheet' listing all the available menus, and options for those menus, would be of great help.


The EPS uses fixed sampling rate technology. Rather than shifting a clock to create transpositions, the number of samples is altered instead. To explain the ramifications of this in full would be rather time-consuming, so here's the bottom line: the samples sound really good when you transpose them. What's more, when you transpose down, the EPS interpolates new samples in between the existing samples, which produces exceptionally good sound quality - even when you take a note and transpose it down several octaves. You really can shift a cymbal down, say, two octaves and have it still sound good.


Sampling is simple. Select the instrument into which you want to sample, and go. There are 40 available sample rates between 52.1kHz (highest rate) and 6.25kHz (lowest rate). You can also choose the input filter cut-off frequency (when you change sample rate, the cut-off defaults to the optimum value for that rate). There's also a 'pre-trigger' function, where the sampler stores a bit of the signal prior to reaching the threshold and therefore always catches the very first transients of a sample. You can also select between Line or Mic input levels (although there is no associated sample input level control - oh well), and check on the available sample time (which takes sampling rate and amount of memory into account - delete instruments, or lower the sampling rate, to make more room if necessary).

What about multi-sampling? Again, it's simple. Specify the root key for a new sample, and the EPS will automatically choose a split point located midway between the root key and the root key of the nearest sample.

Regarding auto-loop, I thought I'd try it out, so I sang 'ah' into a mic and sampled it. I then enabled the auto-loop function and set an arbitrary loop start point. Next came four tries at finding the perfect loop; it wasn't necessary to do five. The loop was perfect - not a click, pop, or ping. But then came the coup de grace. Just for kicks, I tried a bunch of additional end points. For every two new end points I tried, one of them gave a perfect loop. Maybe I got lucky, but based on my other sampling experiences with this machine, it seems like this is one auto-loop function that really works. Incidentally, you can adjust the end (and start) points in a number of ways: as a percentage of the sample, in coarse sample increments, and (using the data slider) in fine sample increments.


We'll cover editing concepts in this section; for a blow-by-blow description of what's on each page, refer to the 'Pages' sidebar. Let's start with my favourite sub-module - the filter. The EPS has two fully digital filters connected in series, F1 and F2. There are four ways to set up these filter stages:

F1 F2
1. 3 lowpass filters; 1 lowpass filter
2. 2 lowpass filters; 2 lowpass filters
3. 3 lowpass filters; 1 highpass filter
4. 2 lowpass filters; 2 highpass filters

F1 and F2 can be set individually for cut-off frequency, modulation from Envelope 2, keyboard tracking, and a single modulation source/amount (amount can be positive or negative). So, if you choose filter option 4 (2 lowpass + 2 highpass) and set F1's cut-off higher than F2's cut-off, you'll create a bandpass filter - just like the old Moog modular synths, where coupling together a lowpass and highpass filter would give you a bandpass response.

The three 6-stage envelopes (nominally for Pitch, Filter, and DCA, but cross-assignments are possible) are equally innovative. (Also note that there is a second release stage for adding pseudo-reverb effects.) You can, of course, specify time and level parameters for each envelope, but the really interesting part is that you can define both a soft velocity amplitude shape and level, and a hard velocity shape and level. For velocities in between these two extremes, the EPS will actually interpolate (estimate) an appropriate value between the soft and hard envelopes! This dual envelope approach cries out for a computer editing program where you can grab the envelope breakpoints with a mouse, but I'll settle for programming them via the display until a computer editor comes along.

"But," you say, "programming 6-stage envelopes is a pain." Well, apparently someone at Ensoniq agrees, because you can save up to eight of your favourite envelopes as templates which can be recalled later. If you've worked for years getting that perfect plucked string envelope, then save it as a template. The idea of being able to save and load envelope templates is one of those 'why didn't someone think of this before?' ideas.

Another tricky technique occurs with the final DCA. When adding modulation, you don't just specify depth, but the (adjustable) modulation level at which the modulation reaches an (adjustable) maximum depth, as well as the (adjustable) level at which the modulation reaches a minimum depth (which is also adjustable). This allows for all kinds of velocity/pressure crossfades, velocity/aftertouch switching, and positional crossfade effects, as well as numerous special effects just waiting to be discovered by some musician with a lot of patience.


Just when you think there's nothing more that can be packed into a box this size, you find - the sequencer. This one is light years ahead of the one in the Mirage.

One of the ground rules is a sample time/sequence length trade-off; they both compete for memory space. However, if desired, you can use the EPS strictly as a MIDI sequencer with 48 tick/quarter-note resolution, and dedicate the entire available memory for sequencing other MIDI instruments. This gives about 80,000 available notes, even without memory expansion.

The EPS holds up to 80 nameable sequences, which can consist of up to eight polyphonic, dynamically allocated tracks. Each instrument plays over its associated track (eg. instrument 1 over track 1). A track can be muted or soloed, mixed down with the other tracks, and panned within the stereo field. Sequences can also be chained together into a single song, and you can also record in song mode.

The editing is pretty decent, too. You can use GOTO and jump to any bar number in the sequence; set the tempo; loop the sequence; set internal or external sync (Song Position Pointer is both transmitted and recognised); select metronome on/off, click rhythmic value, and click level - you can even pan the click, or send it to an individual output if you have Ensoniq's optional OEX output expander box (more on that later). You can change the sequence length, and programme a count-off when recording. There's also an entire additional family of sequencer commands, as described in the sidebar 'Command Edit Pages'.


We're talking True Genius here! You can indeed load a sound while another is playing, but there is a catch: there has to be enough memory to load the new instrument. If you try to load an instrument and there isn't enough memory, you have to select an instrument to delete. This might be a problem if the EPS was an ordinary sampler, but it's not. Usually, you want a lot of memory in your sampler so you don't have to constantly load stuff from disk. With the EPS, you get used to calling up instruments the way you'd call up patches on a regular synth; all you have to do is remember to start loading the new sound before you actually need it. And don't forget about the two patch select buttons by the pitch wheel that let you change sounds 'on the fly'.

As mentioned earlier, with this sampler, notes cutting each other off is almost a thing of the past. And this philosophy extends even to the optional multi-output expander (the OEX), which is, simply stated, what we've been waiting for all along when it comes to multiple outputs: you simply assign one of the eight instruments, or a single wavesample for that matter, to one of the eight outputs. Yes indeed, we're talking polyphonic, non-mono-hardwired outputs that work just like you want them to. Am I complaining that it costs extra? Not a chance.


Ensoniq have more than done it again. The Mirage made affordable sampling a reality; the EPS not only makes affordable high quality sampling a reality, but goes beyond what we've come to expect from an instrument with this kind of price tag. There's a lot of pent-up demand for a quality sampler that costs £1695 - well, here it is. This is one brilliant piece of musical engineering that richly deserves the success it will no doubt enjoy.

Price £1695 incVAT.

Contact Ensoniq UK Ltd, (Contact Details).


The EPS has a lot of pages, which I've arbitrarily broken up into 'Voice' editing pages and 'Command' editing pages. Here's the rundown on the voice edit options.


Choose from the following parameters: Mode (normal, cycle, and repeat), five hard velocity and soft velocity levels, five hard velocity and soft velocity times, 2nd release parameters, attack time velocity (ties attack time to velocity), keyboard time scaling (envelopes get shorter as you play further up the keyboard), and soft velocity curve on/off. You can also recall any of eight envelope templates.


In addition to the parameters mentioned in the main text, modulation sources include wheel and/or aftertouch, just aftertouch, external MIDI continuous controller, pedal, mod wheel, pitch (wheel as well as any other pitch modulators), keyboard, three separate velocity curves, one of three envelopes, random modulation, and LFO.


DCA-related options include wavesample volume, pan, A-B fade in, C-D fade out (the latter two provide the velocity crossfade/switching and special effects alluded to earlier), and the modulation source/amount choices as in the filter section. Each sample can have its own fade characteristics, which makes it easy to fade from one multisample to another to camouflage the split point.


This page adjusts the root key of the wavesample (semitones and fine tuning), LFO amount, Envelope 1 amount, randomise (adjustable frequency and depth), bend range (0-12 semitones or global), and modulation source/amount (same as the filter section).


Parameters include waveform (square, sawtooth, sine, triangle, and S/T - kind of a pointy sine wave), speed, depth, delay, LFO reset (on/off), and LFO modulation (although there is no amount parameter to go along with this).


Here is where we get into some more digital magic. You can set the sample start and end points, loop start and end, loop position (once you've set loop length, the loop can be moved around in the sample), loop modulation and amount (you can actually shift the loop position with the modulation options available to the filter, DCA, etc!), loop mode (forward with no loop, backwards with no loop, loop forward, bidirectional looping, and loop and release).


A number of parameters affect a given layer, including: monophonic or pedal glide (turns glide on/off with the footswitch) with variable glide time, legato layer (this chooses a layer that will act in a particular way when you play legato), layer velocity (with separately settable high and low parameters), pitch table (standard or non-transpose, but note that the entire pitch table can be edited if you're a microtonal fan), and layer name.


This page selects auto-loop on/off, memory remaining for both the internal memory and disk, master tuning, global bend range in semitones, your choice of 16 levels of touch-sensitivity (from light to 'white knuckles'), pedal mode (volume over MIDI controller 7 or modulation over MIDI controller 4), sustain footswitch mode (sustain or patch select), and auxiliary footswitch mode (patch select or sequencer start/stop).

Edit Instrument

On this page, you can choose the instrument's name, patch (this shows which layers are active) for each of the eight instruments, which layers will be activated on key up and which on key down, aftertouch/pressure mode (channel, poly, or off), and the MIDI Out channel over which the notes you play (or the sequencer plays) on that instrument will appear, as well as an associated Program Select number. You can also set whether keypresses will go out via MIDI, activate only the internal sounds, or do both. Finally, there's a readout for the amount of memory taken up by the instrument being edited.


You can select the base channel; whether data will be transmitted over the base channel or the individual instrument channels; whether the base channel pressure is channel, poly or off; MIDI mode (omni, poly, multi, and two types of mono response); controllers, Sys Ex, program change, and song select on/off; and external continuous controller number (this is the controller that appears as the 'external MIDI controller' in the modulation options).


Pressing the Command button, then one of the other page buttons, accesses even more pages; these are of a more system-oriented nature. For example, the Command/Instrument page (accessed by pressing the Command button then the Instrument button) takes care of instrument housekeeping, such as make/copy/delete/save instrument, and save bank. Command/Layer allows for similar layer housekeeping. Selecting the Command/Seq page performs the same function for sequences, as well as letting you change the sequence length and edit song steps. Command/MIDI turns the disk drive into a general purpose MIDI System Exclusive data recorder; Command/System sets the number of voices, lets you load Mirage DSK sounds, make a directory, change storage devices if something's connected to the SCSI port, copy the operating system to floppy disk, and so on. Here are what the other pages do:

Command/Env 1

This contains some diagnostic routines that are best left alone - don't try this at home!


This lets you re-define any key to any pitch, with one cent resolution, and is ideal for creating microtunings.


One of the most powerful Command pages, this lets you mix, merge, splice, fade in, and fade out wavesamples; normalise gain so that a wavesample hits its maximum available level; and smooth volume, a compression-like function that makes it easier to loop signals with varying volume levels.


This provides low-level wavesample editing functions, like add, scale, clear, copy, replicate, reverse, and invert data.


Now we get into some truly serious digital weirdness. You can make, copy, and delete wavesamples; truncate unneeded portions to save memory; try a crossfade loop, reverse crossfade loop, ensemble crossfade, or 'bow-tie' crossfade in search of the glitchless loop; convert the sample rate if you didn't need to sample that kick drum at 52.1 kHz after all; make the loop longer; or synthesize a loop if your real-world signal proves to be, despite all the digital processing options, unloopable.


This sequencer-related page provides a means to quantise, merge, copy, erase, or transpose a sequencer track (quantise goes from 1/4 notes to 1/64th note triplets, and like most of the following functions, can be restricted to a specific number of bars in the track); scale or erase controllers; erase pressure (this helps save memory); erase all notes within a particular keyboard range for the given track for a given number of bars; erase program changes; shift the track forwards or backwards up to 48 clocks (quarter-note), or event edit a track.

Also featuring gear in this article

Featuring related gear

Previous Article in this issue

MIDI Automation Systems

Next article in this issue

Studio Magic

Sound On Sound - Copyright: SOS Publications Ltd.
The contents of this magazine are re-published here with the kind permission of SOS Publications Ltd.


Sound On Sound - Feb 1988

Gear in this article:

Sampler > Ensoniq > EPS

Gear Tags:

Polyphonic Aftertouch
16-Bit Sampler

Review by Craig Anderton

Previous article in this issue:

> MIDI Automation Systems

Next article in this issue:

> Studio Magic

Help Support The Things You Love

mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.

If you value this resource, you can support this project - it really helps!

Donations for January 2022
Issues donated this month: 3

New issues that have been donated or scanned for us this month.

Funds donated this month: £141.00

All donations and support are gratefully appreciated - thank you.

Magazines Needed - Can You Help?

Do you have any of these magazine issues?

> See all issues we need

If so, and you can donate, lend or scan them to help complete our archive, please get in touch via the Contribute page - thanks!

If you're enjoying the site, please consider supporting me to help build this archive...

...with a one time Donation, or a recurring Donation of just £2 a month. It really helps - thank you!

Small Print

Terms of usePrivacy