16-Bit Stereo Sampler
Akai's S1000 is without doubt the industry standard professional sampler, so why try and improve on the best? Paul Wiffen finds that SMPTE cue listing, AES/EBU real-time output, and built-in effects are a good start (with hard disk recording to follow).
It is not really possible to evaluate a machine like the Akai S1100 without constant reference to the S1000. Certainly, anyone who has ever worked with the S1000 will feel immediately at home. As far as user operation from the front panel is concerned, it is more like working with an S1000 with a new software update than a new machine. All the buttons are in exactly the same places, and all the features which already existed on the S1000 are programmed in exactly the same manner. The only difference is that there are some extra screens which allow you to programme the additional features not found on the S1000. Furthermore, the S1100 loads S1000 and S900 disks directly, without any of that tedious mucking about with sample conversion or rebuilding of programs.
However, some pretty radical changes have taken place under the bonnet (and round the back) of the S1100, and we will be looking at these in greater detail later on in the review. But, as constant reference will be made to the S1000 throughout this review, I had better summarise its features and attributes for those who have been up the Amazon for the past two years (and who missed my review of the keyboard version last month!).
The S1000 itself was a continuation of the S900, which at the time of the S1000's release was the best-selling sampler worldwide. The S1000 was eagerly awaited, as most owners were already stretching their S900s to the limit of its 512K memory and 8-voice polyphony. However, nobody quite expected the huge step forward that the S1000 represented. Its 16-voice polyphony, 2Mb of memory (expandable initially to 8Mb), high density floppy disk drive, and polyphonic separate outputs were completely unheard of at this sort of price (indeed, some major American and Australian manufacturers found themselves wishing they had some of its attributes in their hardware — most noticeably the polyphonic outs).
But it wasn't just the paper spec of the S1000 that lead to its overnight success. Thanks to large LCD display, users found that for the first time they could see (and therefore understand) what was going on in complex operations like waveform editing, looping, and keyboard assignment of samples. User friendliness was its hallmark, but not at the expense of speed of operation for the expert user (a rare combination indeed).
The actual depth and complexity available in sample and preset programming was also unprecedented. You could see the sample as it was being made, adjust start and end points visually (with a permanent pre-record in case your trigger threshold had cut off the front of your sample), and loop it using a loop window which was as accurate as anything offered by a computer-based sample editor!
Once you had created your samples, you had unprecedented flexibility in assigning them to the keyboard. In each 'keygroup' you could mix and match up to four samples, and switch or crossfade between them using velocity, in an easy to comprehend visual display. Keygroups were assigned to the keyboard using another visual display, and overlapping areas were possible for further doubling of sounds or positional crossfading (to smooth out exaggerated changes in timbre from one keygroup to the next). Filtering and enveloping could be set up for each keygroup individually or for all of them simultaneously, which saved time, but meant that no compromises were necessary if you really wanted to tweak each keygroup.
Multiple presets could be held in memory at the same time (especially if it had been expanded) and, for the first time ever, it was child's play to assign them to different MIDI channels for simultaneous triggering from a sequencer. The S1000's disk operating system was the fastest and most flexible yet, able to load a full bank, an individual preset (with or without the associated samples) or individual samples, with the minimum of fuss from the new HD diskettes, each of which could hold 1.6Mb of sounds. It was without doubt this superb disk handling that enabled Akai to sell so many S1000s without a hard drive, when everybody else's machine seemed to demand one.
Regular software updates kept the S1000 at the forefront, as Akai added ground-breaking features like time compression/expansion (for which the market had been screaming) without the expense of new hardware purchases. The S1000 today is as valid as it was when it first appeared, because of this continuous upgrading of its operating system. Yet you are never left with a useless machine if you happen to lose your latest system disk — you can always fall back on the original system, which is permanently stored in ROM (Read-Only Memory) inside. Other sampler manufacturers please note — this is the only way to work. If I had a fiver for every time I have sat looking at an Emulator, Ensoniq or Roland sampler which has been separated from its operating system, or whose hard disk has crashed, turning it into a very expensive metal doorstop, I'd be a rich man by now!
The S1000 is the industry standard sampler, virtually every studio in the world has one, and 90% of pro keyboard players have one nestling in their rack, right? So why try and improve on it further? Well, unfortunately, there are all these wretched people out there who are never satisfied, who always want more (like yours truly, for example). To explain what this 'more' is that we're always wanting, perhaps we should start with a little history...
When the Akai S1000 first appeared on the market, it was in a class (and a price range) of its own. The only comparable unit was the Emulator III, which still had monophonic separate outputs, a low density floppy drive (really only suited to backing up hard drives), and a price tag nearly three times that of the Akai unit. (Whatever happened, by the way, to the upgrade for the EIII that Emu announced at last Winter NAMM, to give more voices, more memory, and polyphonic outputs?)
But since then, the S1000's clear lead has been slowly eroded. The Dynacord ADS arrived six months later and was audibly superior in terms of sample quality (it used an A-to-D convertor which wasn't available when the S1000 was released), although it had no filtering. It was also (in the UK at least) considerably cheaper. The Ensoniq EPS had nowhere near the same sample quality, but it did feature Play While Load and better filtering and enveloping than the S1000. It was a lot cheaper, too. But while amateurs and some semi-pro users may have been lured away by the cheaper pricing, the pro end of the market continued to buy the S1000 in large quantities, spurred on by the arrival of a SCSI interface and the digital I/O card. CD-ROM (previously one of Emu's major advantages) also became available as third party developers began to have confidence in the number of units in the field. The S1000 looked unassailable.
Next, Roland mounted a challenge. When first announced, the S770 looked like a really serious contender. The sound quality was (to my ears at least) the best ever offered on a sampler (not counting £20,000 plus systems) and its digital filters are still the only ones I've ever heard that come close to the warmth and roundness of analogue circuits. It offered 24 voices, and the memory could be expanded to 16Mb using off-the-shelf standard SIMMs. What's more, its built-in digital port could output everything the machine was doing in real time. But the price was a good deal higher than the S1000 (especially when you added the cost of an RGB monitor) and the operating system was nowhere as neat and quick as that of the S1000 (its disk handling was especially sluggish).
However, the S770 made a good showing in the all-digital field, where users never want to leave the digital domain until the final mix (something the S1000 couldn't do), but it never presented a threat in the middle ground that Akai occupy so successfully (especially when Akai came out with 8Mb cards that allowed the S1000 memory to expand to 32Mb).
Emax II, Emu Systems' latest contender, features the same sort of sample fidelity as the Emulator III for around the same price as the S1000. It has 16 stereo voices and a unique feature which means that if you use stereo samples, the polyphony of the unit is not halved. Emax II also has a great sounding set of filters, plus the SE synthesis technique.
The most recent sampler to hit the stores is the Ensoniq EPS16 Plus. It features considerably improved sample quality (it now has true 16-bit A-to-D conversion) and built-in effects. I'm amazed that it has taken this long for a manufacturer to put a DSP chip into a sampler. Let's face it, we have had built-in reverb, chorus, etc in synthesizers for years now, and I would have thought that samples were one of the first sounds you'd want to sprinkle a bit of fairy dust on! Anyway, the tail end of 1990 has finally seen the emergence of a sampler with built-in effects, in the shape of the EPS16 Plus, which, although still only a mono sampler, is considerably cheaper than the S1000.
None of these excellent products have managed to slow down the sales of the S1000. In fact, it still has many advantages over the other samplers out there. However, competition is a healthy stimulus and the end user is the principal beneficiary. Akai have obviously noticed the fact that other machines, although not better overall perhaps, were providing higher sample quality than the S1000 and offered certain additional features, like full digital output (which the top end of the market wants) and built-in effects (which appeals more to the lower end of the market), and have consequently decided to upgrade the design to implement the features which were not available (or not even thought of) when the S1000 was originally launched.
So Akai have come up with the S1100, which builds on the success of the S1000 and includes the latest developments to have hit the market in the two years since it was first launched. The first upgrade is to the input sampling section. Akai are now using the Crystal A-to-D convertor, featured in many of the other samplers released in the meantime. One of these little beauties does the same job as the two Burr-Brown chips used in the S1000, and it sounds a whole lot sweeter. The dynamic range is apparently 6dB higher at the input stage. It certainly sounds cleaner and quieter. Obviously, all 16-bit convertors are not equal!
The next addition to the S1100 is a DSP chip to allow 24-bit processing for built-in effects. There is room for 50 effect patches to be held in the S1100 at once, with 10 DSP algorithms available. Each program has its own separate send to the effects and the effected signal is available from audio outputs 7 and 8. The DSP algorithms are Large, Medium and Small Halls; Large and Small Rooms; two Plates; Chorus/Flange; Pitch Shift (+/-50 semitones/cents) and Echo (with three delay voices of up to 60 milliseconds each).
The effects sound extremely clean, thanks to the 24-bit floating point calculations for the DSP. People often wonder why it is necessary for the calculations to use more bits than the A-to-D or D-to-A convertors. This is because the huge number of calculations required to build up a reverb effect each produce a greater number of digits. Although the difference is insignificant in a single rounding off, if these extra digits are repeatedly rounded off each time, then errors build up to create noise in the effect. By using a larger register to accumulate the bigger numbers, the rounding off process can be left to the end of all the calculations, where it remains insignificant.
The final hardware addition to the S1100 is in the D-to-A department. The S1000 is actually a bit of a triumph in that all the outputs (11 in total) are actually driven from one D-to-A convertor. Maintaining sound quality when performing that kind of multiplexing is a real achievement, especially if there are a lot of samples being triggered at the same time. However, in the S1100 Akai have opted for overkill, to ensure that nobody ever has cause for complaint even in the most demanding applications — it has a 20-bit D-to-A convertor for each output! This means that the sound quality should never suffer, however much you are asking the machine to do at once, because there is no multiplexing between the different outputs.
The next improvement in the S1100 is designed to cater for the most demanding users in the digital domain. The initial excitement about the S1000's optional IB104 digital interface was somewhat dampened when it was found that, although the card was excellent for getting sounds that had already been digitised into the S1000 without any loss in fidelity, it could not be used to pass the music that was being triggered in the S1000 on to the next link in the digital chain (the digital reverb or digital recorder). The digital output was fine for backing up the memory or the hard disk to DAT, but was not capable of sending the multiplexed voices being played in real time.
All this has changed in the S1100. It comes with AES/EBU capability as standard, allowing it to transmit whatever is being triggered inside the machine over the digital bus as well as recording incoming samples in the digital domain (as with the IB 104). This means that an S1100 can be digitally connected directly to a professional DAT recorder, multitrack, or reverb (if similarly equipped with an AES/EBU digital port).
Another professional interface which comes as standard on the S1100 is SCSI (Small Computer Systems Interface). The fact that the SCSI card wasn't standard on the S1000 caused a few problems in its early days, as pro users found that they couldn't use mass storage devices until they could track down one of the all too elusive interfaces. It also meant that if, like me, you were a programmer or sound developer for the S1000, you had to carry around your entire sound library on floppy disk, in case the S1000 you were working on happened not to have the optional SCSI card fitted. With the S1100, you can be sure every one you come across will have a SCSI port.
Of course, if you prefer to have a hard disk inside your sampler, the optional HD84 hard disk allows 80Mb of sounds to be stored within the machine, in addition to whatever SCSI drives you may wish to connect externally. This may be a good idea if there is a central core within your library of sounds which you never want to be without. But in 1990, 80Mb represents a mere drop in the ocean of sounds available in S1000 and S900 format (which the S1100 can read directly, programs and all), so any 'power user' is going to be looking for a much larger SCSI format (either a huge capacity hard drive or some form of removable media — either Syquest or rewritable optical) to house the vast sound library that they can now draw on. Don't forget that there is a growing library of sounds on CD-ROM as well.
Another indication of the widening market that the S1100 is aimed at is the addition of SMPTE In and Out on the back panel. This is primarily for the Cue Play facility, a 'cue sheet' type software addition that allows you to trigger samples at given timecode points, which is primarily used for putting sound effects to picture. It is the first piece of software to appear under the Utilities menu; the poor old S1000 is still waiting for some software to appear in this menu.
Actually, with a bit of luck, this Cue Play would be an ideal feature to have available on the S1000 as well (next update?). "Aha," I hear the clever ones amongst you shout, "but then what would you trigger it from? There's no SMPTE input." Well, you would trigger it via something that should be added to the S1100 at the first opportunity — MIDI Time Code (MTC).
Samplers are a perfect example of a MIDI instrument that would really benefit from the inclusion of MIDI Time Code. Apart from anything else, it is much cheaper to implement MIDI Time Code than SMPTE on a MIDI instrument, as you don't need the extra hardware to read the SMPTE code; MTC can enter via a standard MIDI port (although you shouldn't try and use it for too much else at the same time!). Also, you can buy a SMPTE-to-MTC convertor for under £100 in the shape of a J.L Cooper PPS1, if you wanted to trigger your S1000 (or other non-SMPTE sampler) from a SMPTE timecode.
Back to the S1100... Cue Play takes a fairly standard cue list format. SMPTE times are expressed as hours, minutes, seconds, frames, and tenths of frames (eg. 02:36:47:21.6). Next to this you have the file name of the sample you want to trigger, and whether you want a Note On or Note Off to occur at that SMPTE time (I foresee lots of droning notes on looped samples if people can set Note Ons without Note Offs). Next is the pitch you want to trigger, and finally the velocity you want to trigger the note with.
This increasing use of SMPTE on musical instruments is all very well for people in the business of putting sound to picture, but there are other more musical methods of triggering, say, vocal files which I would like to see implemented on instruments like this. For example, MIDI Clock makes much more sense than times expressed in picture frames if you are working in a musical context. Instead of working out with a calculator that you need to place the vocal for Verse 1 at roughly 0 hours, 1 minute and 3 seconds, and then adjusting the number of frames until it is in exactly the right place, wouldn't it be much easier to look at your sequencer, see that the verse starts at Bar 25, and enter Bar 25 Beat 1 as the start point? Ah well, I guess someone will get around to doing it one day! Don't get me wrong, I've nothing against SMPTE. When you are putting sound to picture it's the only way to go, but MIDI Clock makes much more sense for music recording in isolation.
That pretty much covers all the hardware additions and changes in the S1100. Like the S1000 it can be expanded to 32Mb using the new 8Mb boards, and both are 16-note polyphonic. And there lies my only real complaint about the S1100. Like its predecessor, it sounds so great that you end up trying to use it to play everything in a sequence, and there just aren't enough voices (even with the clever polyphony and priority assignment). The problem gets worse if you are using stereo files which, of course, require two voices. Even if the overall polyphony couldn't have been increased, I would like to have seen the stereo voice concept that Emu use on the Emax II, to make sure that triggering stereo samples doesn't halve your available polyphony (otherwise you're better off playing a mono sample through the stereo effects and saving other voices to play back even more sounds). Still, you can't have everything in this world, and Akai haven't ignored the problem completely. Apparently, there will be an S1100EX which will act as a 16-voice expander for the S1100. The two machines will communicate via SCSI, and it too will be able to hold up to 32 megabytes, giving you a 32-voice, 64 megabyte system (get him a tissue someone, he's drooling!).
The future also looks pretty good for the S1100. Version 2.0 software (due next Summer) will allow the machine to act as a stereo direct-to-disk recorder, and will also allow programming to be carried out via an external monitor. Normally, I don't like to talk about 'vapourware' or future features (to use a more polite term), but Akai have consistently delivered their promised software upgrades on time, so I'm prepared to give them the extra plug (consider Version 2.0 on the S1000, with all that great time compression/expansion stuff).
Apart from the addition of pages like the SMPTE Cue Play and Effects parameters, very little has changed in the rest of the software from the latest versions of the S1000. One noticeable change (which was overdue) is that the Response page — which was confusing and badly ordered, and didn't allow you to see more than one program at a time — has been replaced by two pages, one called Mix and the other called MIDI, both of which allow the status of several programs to be compared. The Mix page features parameters for Level, Output, Stereo Volume, Pan, and FX Send (as a percentage), while toggling to the MIDI page (still under Play) allows you to set Channel, Key Range, Polyphony, Priority, and Octave Shift for the same programs. This whole arrangement makes working with multiple programs on different channels from a sequencer about 50 times easier (no exaggeration)!
All in all, the polyphony issue apart (let's face it, who ever has enough voices?), the Akai S1100 is pretty much everything that any S1000 owner could ever dream of: better sound quality (in and out), built-in effects, SMPTE Cue Play, and full real-time AES/EBU digital out (with hard disk recording and external monitor programming to come). When you look at how little extra money the S1100 costs, it's no wonder Akai couldn't afford to provide any extra voices. Clearly, the forthcoming S1100EX is designed to take care of that need.
A while back, it was common for journalists to express their complete approval of a product with the phrase "I bought one". I know it sounds a bit hackneyed now, but my defence is "I bought one before I knew I was going to be asked to review it". Anyone want to buy a second-hand S1000 with one not-so-careful owner? On second thoughts, I'll keep it for the extra polyphony!
£3499 inc VAT.
Akai UK, (Contact Details).
Review by Paul Wiffen
Previous article in this issue:
Next article in this issue:
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!