Monday 2 May 2022

Akai CD3000XL: Your Face in a Magazine, with a Big Red Limousine

Let's have a look at the Akai CD3000XL, a hardware sampler from 1996. Mine appears to have been manufactured in August 1996, if I'm decoding the serial number correctly. It's one of Akai's later models, with 16-bit stereo sampling at 22khz or 44khz.

Personal anecdote. Back at the turn of the millennium - which is spelled two-two, just like accommodate and occurrence - I lived in London. I remember a row of shops just outside Notting Hill Gate tube station. The Something And Something Exchange shops. Books and comics, clothes, music, possibly others. Staffed by students who needed a bit of pocket money.

Incidentally possession is also spelled two-two, and so is successful and committee. Committee is two-two-two. Philippines is one-two. Millennium and accommodate are two-two. That's how you remember. Two-two. You just have to mentally train yourself that millennium is two-two. How useful is the ability to spell "millennium" in 2022? Not very useful, but what if the dating system changes and suddenly 1023 becomes year zero and we have another millennium in 2023? What then?

The Exchange shops cling on in the face of high rents and eBay, and there are fewer of them now, but it's nice to know that they still exist. A little reminder of London in the 1990s, when there was still a place for weird, scuzzy little shops. The Music and Video Exchange was particularly interesting because it was a chance to see rare old synthesisers and tape delays in real life.

I remember that along one wall of the shop was a glass cabinet packed full of samplers. Stacks of them on top of each other. Rackmount samplers by E-Mu, Roland, Ensoniq, even Casio, but Akai was by far the most common, because Akai was king in the UK.

The interior. It's essentially a 1990s desktop computer in a tough metal case. Notice how the rack ears have reinforcing brackets that are screwed into removable side plates - Akai built their samplers well. Thankfully the capacitors appear to be in good condition. My CD3000XL has been upgraded from the stock 8mb to 16mb of memory with a pair of standard 72-pin 8mb SIMMs.

Samplers in general were extremely popular in the UK in the 1990s. Whole genres of music were created with them. Entire careers relied on them. Almost every electronic musician had one, or several. This is "Beat Dis" by Bomb the Bass, which was made with a bunch of Akai S900s. It got to number two in the pop charts in early 1988:

Sampling had an interesting history that went through several waves. The idea of making music with the audio equivalent of kitbashing dates to the invention of sound recording, but digital sampling didn't become practical outside universities until the late 1970s, with the Fairlight CMI, which was launched in 1979. It cost a fortune, and most were sold to recording studios, but the likes of Peter Gabriel, Jean Michel Jarre, The Art of Noise, Kate Bush etc were sufficiently enthused by the machine's creative potential to buy one.

They all released at least one album that used sampling heavily, but for the most part the novelty wore off, and by the middle of the decade they learned how to use sampling more subtly. Meanwhile the recording studios and record producers that owned Fairlights tended to use them as just another instrument in a big arsenal of instruments. Trevor Horn, for example, used sampling extensively on his mid-80s mega-productions, but that was because it was a cost-effective way of getting a big orchestral sound without having to pay for an orchestra. The Pet Shop Boys' first couple of albums were made almost entirely with a Fairlight, but it was the sequencer of the CMI II that they were drawn to, not sampling per se. And so forth. Sting, U2, Queen, Genesis etc - the rock aristocracy - all used samplers, but transparently. They were not interested in chopping drum loops into tiny slices so that they could play them back all skittery-like.

But as the 1980s progressed samplers dropped in price and came within the grasp of a wider audience. Akai's first sampler was launched in 1986 at around £900, which was extremely good value; within a few years it became possible to emulate the functionality of a Fairlight or Synclavier sampling workstation by combining an Akai rackmount with a MIDI-equipped Atari ST running Cubase (or a Commodore Amiga with a MIDI interface and a tracker that could use it), at which point sampling opened up to a new batch of producers who had fresh ideas.

This happened rapidly, so that for example De La Soul's Three Feet High and Rising and the Beastie Boys' Paul's Boutique were released only a year after "Theme From S'Express" and "Jack to the Sound of the Underground", and only two years after MARRS' chart-topping "Pump Up the Volume". In the late 1980s there was a wave of dense, sample-heavy records, which was largely snuffed out by the need to pay for the use of samples, but e.g. DJ Shadow's Endtroducing..... with five dots two-two five dots two-two came out as late as 1996.

By the end of the 1990s sampling had become normalised to a point where it was hard to tell whether a record used samples or not. I'm willing to bet that the soundtrack for WipeOut 2097, released in 1997, is 90% samples, but none of the tracks really use samples in a show-offy way:

Apart from the very first track, by The Future Sound of London, which does use sampling in a show-offy way, but that's because FSOL were awkward. Beyond upbeat electronic dance music the entire genre of drum and bass was built with samples, as was the ambient house of the early 1990s, ditto trip-hop. The Future Sound of London's Lifeforms, Portishead's Dummy, Roni Size / Reprazent's New Forms were all sample-heavy, although New Forms pointed to a future in which drum and bass artists incorporated live instruments into their repertoire. New Forms in particularly illustrates a popular workflow of the period, where musicians borrowed analogue synths and acoustic instruments and built up a personal sample library, with everything ultimately being played back from a sampler, because that was easier and more reproducible than messing around with MIDI-CV converters and setting up a bunch of analogue gear.

One of the artists that appears on the soundtrack for WipeOut 2097 is Daft Punk, who were nothing special in 1997. Within a couple of years however they spearheaded a sample-based disco revival, and I think of that period as the last triumph of Akai rackmounts. I've always associated the next sample-based fad, the mashup, with software samplers and laptops. By the time of "Freak Like Me" and The Grey Album sample-based musicians were using software plugins to mix MP3s. From that point onwards the rackmounts withered and died.

Still, although Akai's sampling heyday only lasted a decade and a bit, that decade produced Logical Progression, Endtroducing...., Lifeforms, U F Orb, Second Toughest in the Infants, sample-heavy rock albums by U2, David Bowie and the like - and scads of trance and jungle 12" singles that only appeared on compilations. About the only artists who didn't use samplers extensively were the blippy experimenters on Warp and Rephlex.

Was there a definite point when it all came to an end? In 1998 Norman Cook got to number six with "Rockafeller Skank", which showcased a new type of timestretching. Akai samplers could timestretch, but Cook used a software plugin instead, and I like to think that was the point when Akai samplers were doomed. That and the 1996 release of Gigasampler, a software sampler.

USB existed in August 1996, but it was in its infancy. Instead the CD3000XL connects to external hard drives and desktop computers with SCSI, which is deeper magic from before the dawn of time. In the 1990s musical instrument manufacturers generally targeted the Apple market, hence the prevalence of SCSI and FireWire in musical gear of the period. Sadly the individual outs are unbalanced.

Earlier Akai rackmounts had multiple outputs as an optional expansion, but by the time of the CD3000XL multiple outputs came as standard. If you look to the top-left of the lower motherboard you can see the EPROM that holds the operating system - mine has version 1.52. As far as I can tell that's the latest version that came installed in the CD3000XL. eBay has adverts for a V2.0 upgrade, but I can't find a single person on the internet who has installed or used it.

Gigasampler (latterly Gigastudio) was quickly followed by HALion, Kontakt, EXS24, Battery and so forth in the first years of the 2000s. Hardware samplers still had a place in a live setting, but they couldn't compete with the storage space and ease of use of a desktop PC, and when laptops became powerful enough - roughly 2001, with the PowerBook G4 - they didn't make a lot of sense in a live context either.

My CD3000XL can be expanded to use 32mb of RAM, with an optional 510mb hard drive, but by the end of the 1990s desktop PC hard drives were in the gigabyte range and software samplers could stream samples from disk, so RAM wasn't a huge issue. Furthermore as time went on it became practical to run multiple virtual instruments and effects in real time on a computer, at which point sampling itself lost some of its raison d'etre. Why sample an analogue synth when you could run an analogue modelling plugin in real time?

The L7-A1045 DSP was apparently designed by Roger Linn. Beyond its use by Akai it also appeared in a bunch of arcade machines. The other big name in Akai-land was David Cockerell, developer of the EMS VCS3, who in the 1980s moved to Akai and helped design the S612, S900 and its heirs.

Some vintage musical gear comes back into fashion, but as of 2022 rackmount samplers are big, heavy, unwanted pieces of 1990s studio gear that no-one wants. The MPC units fetch high prices on the used market because of their ease-of-use, built-in sequencer, and their punchy sound quality, but the samplers are generally shunned. Especially the later, CD-quality models, because they were supposed to sound transparent.

Incidentally there's some debate as to the concordances between the S-series rackmounts and contemporary MPCs, because they shared hardware. The CD3000XL appears to use the same D/A converters and DSP as the MPC-3000, but does it sound the same? I don't know.

On a practical level a hardware sampler doesn't make a lot of rational sense nowadays, but I've always wanted to own an Akai rackmount just to see what the fuss was about, and when an Akai CD3000XL popped up at the 2022 VEMIA synth auction for pennies I put in a bid and won, because I was the only bidder.

There were essentially four waves of Akai rackmounts. The range began in the mid-1980s with the Akai S612, which had a black faceplate, 128kb of memory, six voices of polyphony, a sampling resolution of 12 bits at a maximum frequency of 32khz. It used 2.8" floppy discs, which are now staggeringly rare.

However when people think of Akai samplers they think of the three waves that followed. They all had beige faceplates and a similar interface dominated by an LCD and a jog wheel. The 1986-onwards S900, S1000, S1100 in that order (with the budget S950 coming slightly later) progressed from 12-bit to 16-bit sampling at 40khz to 44khz, with optional SPDIF connectors, SMPTE adapters, and multi-output options. They came with anything from 750kb to 8mb of memory, expandable to around 32mb. On the used market they're the most sought-after of the bunch on account of the 12-bit sound quality of the S900 and S950. The 16-bit S1000 and S1100 remained current for several years because even in the late 1990s they were "good enough" for most musicians. Akai also released keyboard versions of some of their early samplers, but they were huge and very expensive.

The 1992-onwards S3000, S3000XL, CD3000, CD3000XL and so forth - with the S2000 budget model - really split into two groups, but I think of them as the samplers that were brand-new during the heyday of hardware samplers. They were 16-bit, 44khz machines with digital filters, envelopes, timestretching, looping, multiple LFOs, flexible modulation routing, digital I/O. The manual for my CD3000XL compares it to a modular synthesiser. The latter models could be expanded with optional boards that added multi-effects and hard disk recording.

My CD3000XL is relatively stock. The effects board is rare and expensive nowadays, and also slightly pointless because there are better options. If you want to recreate the sound of 1992 you really need an Alesis Quadraverb, or no effects at all, because your samples already have reverb. Drum and bass producers back then often used clever sampling tricks to make it sound as if they had more effects than they did, such as this track, Johnny Jungle's "Flammable", which used timestretching to give the drums an echoey flange effect:

Akai continued the range with the 1998-onward S5000 and S6000, which were very powerful, with pull-off control panels and tonnes of analogue and digital I/O. Unfortunately they were released just before Akai declared bankruptcy. Not because of poor sales but because one of the owners of the company had apparently been embezzling money. Akai's final rackmount samplers were the 2002-onward Z4 and Z8, which were the first with USB as standard, but they appear to have sold poorly and I've never seen one in real life.

As mentioned above Akai went bust in 2000, in what was at the time Hong Kong's biggest bankruptcy, so I assume their later models were not widely available. Press reports from 2000 talk about a forthcoming Akai software sampler, the Akai VZ8, but it was never released. The Akai that exists today is essentially a completely different company that emerged in 2004, although it still sells Akai MPCs. It's ironic than the MPC outlasted the supposedly more professional rackmounts.

What's it like to use an Akai sampler in 2022, in England, in the spring? Labour-intensive. The first problem is storage. My CD3000XL's CD doesn't work, but that's not a huge problem. Akai used its own special format for CD-ROM data, and I don't have any Akai sample CDs. The CD drive will also play standard audio CDs, but it connects to the D/A converters with an analogue cable, so sampling from the internal CD is slightly pointless. For maximum sound quality it's a better idea to connect a CD player optically to the SPDIF inputs instead. The manual even recommends this. Incidentally the CD-equipped Akai samplers had a system whereby you could edit the samples by recording only your edit commands to a floppy disc, with the original audio files left on the CD.

The floppy drive does however work, so I dug out some old floppy discs and used them instead, which limited me to 1.44mb of samples. Which is plenty for individual notes and short loops in mono at 22khz. So I decided to make a couple of tracks by sampling things with the CD3000XL, viz:

That track uses a mixture of noises sampled from some synthesisers I had to hand - my modular synth, a Korg MicroKorg, a Volca FM, some sampled beats from Norman Cook's Skip to My Loops sample CD from 1992. I never thought I would use a floppy disc drive in 2022, but here we are. Or SCSI, for that matter. I never thought I would use a floppy disc drive and a SCSI interface in 2022, but etc.

I've written before about the emotional aspect of computer interfaces. In the early 2000s FireWire was the thing Apple people used instead of USB, so I've always associated FireWire with iPods and Steve Jobs and successful people who work in the media. FireWire is tasteful, urbane, stylish, unlike USB. When I think of USB I think of no-name USB memory sticks. Cheap promotional giveaways, broken peripherals that don't work, horrible plastic knock-offs of Apple products.

On a rational level USB is fantastic, but emotionally I will never warm to it. USB is like videotape in that respect. Whatever the merits of videotape I will always associate it with the sight of Elisabeth Brooks taking off her leather coat in The Howling, and that memory makes me feel ooshy, especially given that technically she was the villain. Not technically a bad memory, but it colours my perception of VHS. It was the medium of pornography and films taped from your friend's telly who had Sky.

On a technical level the CD3000XL is considerably more sophisticated than the Volca Sample, albeit that it doesn't have a sequencer. The MIDI implementation is a lot more sensible. It is however a lot harder to carry around.

But what of SCSI? I must have used SCSI peripherals as a child, because I have a distinct memory of using pre-ATA Apple Macintoshes. But I was unaware that their hard drives connected to the machine with SCSI, so ultimately I feel nothing for the standard. Neither love nor hate.

SCSI was an early attempt at a universal interfacing standard. In practice the only devices that used SCSI were hard drives, tape drives, latterly CD-ROMs, but at least they tried ("they" being the long-gone Shugart Associates, who advocated for it). It's one of many computer interface standards that was technically smarter than the competition but also more expensive, so ultimately it died off in favour of ATA, although it was kept afloat in the late 1990s by Akai samplers and Zip drives. It strikes me as one of those things that probably lingers on in a few niche markets, e.g. aviation, the military, NASA.

SCSI stood for "small computer systems interface". One problem facing interface standards in the 1980s and 1990s was the fact that home users tended to buy a computer as a single unit, perhaps with a printer and a disk drive, so there wasn't much need for an expansion port that could support mass storage in the consumer market. As a consequence there was very little incentive to lower the price of SCSI peripherals. Furthermore the standard was a pain to use. It dates from a time when you didn't just have to install drivers to make a peripheral work, you also had to fiddle around with jumpers and bus settings as well. They were not happy days.

But I was talking about music. For the track above I sampled everything through the CD3000XL's analogue input at a mixture of 44khz and 22khz and sequenced it all with Logic. The track is being performed live, with the multiple outputs plugged into different channels of my MOTU 828 so that I can apply effects. Musically the shift in the middle owes a lot to Orbital's "Impact" and the rolling breakbeats aren't a million miles from early Prodigy. The sampled voices are slightly anachronistic - they belong to an earlier age of sampling, circa 1984 - which is why I've called it "Cross-Section".

Even at 44khz there's noticeable aliasing, and the CD3000XL's analogue i/o is surprisingly noisy, although this is only really a factor if you use heavy compression. I tried to play the samples outside their range as much as I could, because I wanted to hear how Akai dealt with transposition. The "Impact" break with the chords at 01:57 is the most obvious example. It sounds surprisingly warm. I was impressed with the punchy bass.

For the following track I did things differently, as if it really was 1996:

That time I used the SPDIF digital inputs and outputs. They're essentially noise-free. I used my MOTU 828 as a D/A front-end for the CD3000XL, transmitting samples to it as digital audio with the SPDIF connectors. After making an arrangement with Logic I recorded each part separately with the CD3000XL's digital outputs - taking care to set the clock source appropriately. If you connect two pieces of audio equipment with SPDIF one of them has to be a clock source for the other, or else the audio has pops and crackles. Luckily the CD3000XL picks up the MOTU 828's clock and vice-versa, so that wasn't a problem.

This time most of the samples were made with Mutable Instruments Plaits, including the guitars, which were created with Plaits' inharmonic string model. For the rest of the samples I raided my library of compact discs, including A Storm of Drones, Sonic Boom, and Ken Ikeda's Tzuki, amongst others.

How does the CD3000XL work? The overall flow of sound isn't a million miles from a contemporary ROMpler. At the lowest level are individual audio SAMPLES. The CD3000XL will store 255 of them. It has limited but decent editing options, essentially trimming, cutting, fading, normalising, and simple timestretching. You can't automatically split drum loops into individual hits. Propellerhead's ReCycle predates the CD3000XL - 1994! - but as far as I know Akai never added beatslicing to its rackmount samplers. About the only automatic function is looping, which will try to find a good loop point when you hit the F7 key. It also has EQ with a mixture of filter types, although they don't work in realtime and there's no way to preview the results. The low shelf is useful for getting rid of low-pitched rumbles.

In order to use the samples you have to assign them to a PROGRAM, which is a little synthesiser voice, with filter and volume envelopes, velocity sensitivity, modulation, multiple key ranges etc. Surprisingly looping is done at the sample level. You'd think it would be a program-level feature, but no. The filter is a weak 12db/octave model with resonance. It's more of a muffler than a squelchy resonant filter, but it sounds surprisingly good for a digital filter from 1996. At some point I'll have to do a video in which I use the LFO to modulate a sample's filter.

You can have four different samples in each keygroup. How many keygroups? I don't know. The manual is vague on that matter. In theory keygroup splits exist so that you can make a lovely-sounding dynamic piano with velocity fades, but they're also useful for squeezing more than sixteen sounds out of a MULTI setup.

Programs are in turn part of a MULTI setup that sets the MIDI channel, panning, level, and individual output assignments of each of the sixteen multitimbral parts. The CD3000XL is 16-part multitimbral, with 32-note polyphony. One limitation is that it can only store one MULTI setup in memory at a time, although thankfully the LOAD page will load the MULTI plus all of its programs and samples in one go, which is handy because Akai's file system is flat, without folders. It just puts all of the samples in the root directory.

You know how modern operating systems ask you if you want to save your work before shutting everything down? The CD3000XL doesn't do that, so it's good practice to save frequently, and remember that saving a MULTI setup only saves the programs and samples in that setup, not the whole memory, unless you select SAVE MULTI + PROGS +SAMPLES.

In terms of ease-of-use the MULTI setup page is easy. PROGRAMs aren't too bad. After a while I developed a rhythm for editing a keygroup, hitting F6 to assign a sample, renumbering the program, then setting up a new program with a new sample. I found it easier to remember where everything was if I gave the programs and their associated samples the same names. Cutting the silences from the beginning and end of SAMPLEs is however irritating. A "trim silence" command would have been very useful. Setting loop points is unintuitive; instead of simply moving the start and end point of the loop, you define the length of a loop and use that as a window imposed over the sample. Does that make sense?

I've only used mono samples, so I have no idea how well the CD3000XL loops things in stereo. One thing that struck me is that the CD3000XL is surprisingly playable. By default each program has a simple velocity curve and envelope, and you can play samples way out of their range without them sounding awful. A simple synthesiser beep at middle C sounds lovely fed through a delay and played across the keyboard. I was expecting masses of digital noise, hiss, aliasing etc, but the overall sound quality is fine. It's very different from the typical "bitcrusher" plugin effect, warmer, smoother.

A V5.5 SCSI2SD SCSI emulator. It draws power from the CD3000XL.

Mass storage? The CD3000XL has lots of internal space for a hard drive, but there are no mounting brackets. I assume Akai sold a kit that screwed into the case somewhere. The CD3000XL takes SCSI hard drives up to a capacity of 510mb, but it partitions them into 60mb chunks. The other period-correct options are SCSI Zip drives and SCSI external hard drives, but they're all at least twenty years old now, and Zip drives in particular were notoriously flaky even when they were new. At least one of Akai's samplers supported magneto-optical drives, and I suppose given that SCSI was supposed to be universal a tape drive might work, but good luck finding media for them in 2022.

Is it practical to have no storage at all, and just sample in the sounds you need at the beginning of a session? No, it's far too time-consuming. Could you leave the sampler turned on all the time? According to my electricity meter the CD3000XL draws 22 watts of power, and after leaving it on for a couple of hours it doesn't get very warm, so presumably there's no reason why not, but what if there's a power cut? The S-series samplers apparently support sample transfer via MIDI, but I haven't tried it, and it's probably very slow, besides which you would need to run Akai's MESA software to do that - in which case why not buy a SCSI card and plug the sampler in with SCSI instead? Hmm?

Therefore modern options essentially consist of a USB floppy disc emulator or a SCSI-to-SD-card device, as pictured above. My CD3000XL's floppy drive still works, so I opted for a SCSI emulator instead. Specifically a V5.5 SCSI2SD adapter from AmigaKit, but there are other options. SCSI2SD uses a MicroSD card for storage. At first I thought it didn't work, but there's a bit of setup - you have to load some firmware into the MicroSD card to make it pretend to be a SCSI peripheral. By far the hardest part of this process involves finding a USB-to-teeny-tiny-USB cable that transmits power and data. USB-to-teeny-tiny-USB cables are often used by mobile phones, but their cables tend to only carry power.

After setting up the SCSI2SD card I have to remember to switch the internal SCSI ID from 4 (the internal CDROM drive) to 0, which is the number I picked for my SCSI2SD card. SCSI supports up to seven peripherals on a chain, so there's no reason why you can't give a device any arbitrary number, as long as it doesn't overlap with another device.


An early example of melodic sound sampling, with magnetic tape rather than digital playback.

I have a 32gb MicroSD card. In theory this can be split into different volumes, but for now I'm content to use just one 60mb partition. That's about twelve minutes of samples at 44khz mono, which is plenty for individual notes.

The CD3000XL loads everything into memory rather than streaming it from disc. The procedure for backing up data simply involves loading samples into the machine, removing the SCSI2SD drive, inserting a second drive, and saving everything out of the sample again. There's no easy way to back up the CD3000XL's data. When it was new you were supposed to connect the sampler to your computer with SCSI and use Akai's MESA software. I dug out my old Toshiba laptop with a built-in floppy drive and managed to use ImageDisk to clone the CD3000XL's floppies, but beyond that I haven't looked into transferring or editing the CD3000XL's files to PC in a usable form. MESA should work on my laptop, but I would need a PCMCIA SCSI card of some kind.

Talk of computer editing does raise the awkward question of why anybody would bother with an Akai hardware sampler in 2022. It doesn't make a lot of sense to send samples from a computer to a sampler when I could load them into Logic's capable built-in sampler instead. The CD3000XL's sound quality isn't that distinctive - listening to the WipeOut 2097 soundtrack I have no idea which tracks were recorded with S1000s, which were recorded with S3000s, and which were recorded without a sampler. However it's a fun toy, and it has encouraged me to put some effort into sampling. I have now built up a little Akai sample library.

Oh yes, there was another thing. The screen. The LCDs in old Akai Samplers tend to go dim over time. Specifically the backlight fades. Mine was originally very dim, to a point where I had to use a bright front-light to see anything. The video examples in this post were filmed with the original, dim backlight.

The two options are a replacement backlight - although the CD3000XL's measurements are slightly different to the other Akai rackmounts, and homemade replacements involve soldering and hot glue - or a complete replacement LCD assembly, which has the advantage of being futureproof, but is two-thirds the cost of the sampler. In the end I opted for a replacement backlight from a chap on eBay called saint_baz, which is probably not his real name. The result is pictured above. It's still not that easy to see in daylight, but it's much better than it was.

Sunday 1 May 2022

Alesis NanoCompressor


Let's have a look at the Alesis NanoCompressor, a cute little hardware compressor from the 1990s. Sound on Sound reviewed it back in March 1997, but the manual is copyright (c) Alesis 1996, so who can you trust?

Who can you trust. The front panel is vintage 1990s, just as the MicroGate and MicroVerb look like the 1980s. Look at the cheatline running through the Micro effects' logotypes. It's supposed to look like chrome, but without the chrome. You know what makes me sad? My memories of graphic design in the 1990s are of David Carson and The Designer's Republic, because I was alive then and I knew what was hip.

But if you Google "1990s font design" the internet's examples are all purple and yellow, like Miami Vice, and the fonts resemble the logotype from Moonlighting, which finished in 1989. The 1980s bled into the 1990s, but by the middle of the decade Miami Vice was dead as a dodo, a big joke. The hip new thing was Se7en, which came out in 1995. Remember the title sequence, with the jumpy text and the deep blacks? That was my 1990s, not pastel colours. The internet is a bald-faced liar.

More of the 1990s.

The NanoCompressor was sold alongside the NanoVerb, which was a reverb unit, and the NanoPiano, NanoBass, and NanoSynth, which were all sample-based synth modules. They used the same case design, one-third of a 19" rack. As of 2022 they're all available on the used market for peanuts, because there's not much demand for them, so they make for a fun collector's project.

Historically they replaced the Micro- effects, and they were in turn replaced by a small range of Pico- effects, although as far as I can tell the only Pico effect was the PicoVerb, which came out in 2003 for the low, low price of £70. By that time Alesis was trying to recover from bankruptcy, and the market wanted plugins, so as far as I can tell Alesis gave up on effects in the mid-2000s to concentrate on drums and the Ion/Micron instead.

I bought the NanoCompressor on a whim. I've never owned a hardware compressor, and I wanted to see what it was like to use one. I was secretly hoping that it would be rubbish but characterful, like a Holga camera or a toy piano. Noisy, distorted, sick, but awesome. On the other hand the NanoCompressor is a simple analogue VCA unit, a type known for its transparent sound; unlike the contemporary Alesis 3630 it isn't infamous for its noisy pump, in fact there isn't much about it on the internet at all.

My experience with Alesis' earlier Micro effects is that they're surprisingly okay. I was expecting masses of hiss and distortion, but apart from a little bit of low-level aliasing and some scratchy pots the MicroVerb II sounds like a decent software plugin. Compared to my Strymon BigSky it sounds thinner, because the reverb isn't modulated, and it's less glassy-smooth, but a couple of the presets are really good. SMALL 1 thickens up the sound unobtrusively and LARGE 4 is epic. The MicroGate is similarly transparent - it gates things, there's not much to go wrong - and having used the NanoCompressor on a couple of songs I can say that, yes, it's just as "surprisingly okay" as the others.

Which is slightly disappointing. I was hoping for grungy sound but instead the NanoCompressor is a low-key volume leveller stroke attack-adder-er. Even set to maximum pump it's surprisingly restrained.


It's a stereo unit, although the panel gives the impression that it's mono. The controls operate on both channels at the same time. It can do peak compression - which acts quickly on transient signals such as bass and drums - and RMS compression, which is gentler and frequently used on the final mix in order to keep the level nice and steady. Here's an example of the difference between the two, plus a little bit of pumping:



What is compression? It's one of those things. On the surface it's simple, just a form of automated volume control, but compressors are among the most sought-after, most legendary pieces of audio equipment. A modern reissue of the 1960s Teletronix LA-2A sells for almost £4,000, and people are prepared to pay even more for the originals. It's a mysterious world of vari-mu and "all-button mode".

On the surface compression is simple. A compressor listens to an incoming audio signal, and if the signal goes above a certain level the compressor turns the sound down. Not instantly - there's a little bit of a delay so that it feels natural. Ultra-rapid compression is generally called limiting, and to complicate matters there are levelling amplifiers that also boost quiet sounds so that everything is the same volume, and software plugins called volume maximisers that look ahead a few microseconds, so that the signal is as loud as it can be without clipping, or without clipping too much.

Compressors were developed in the before times, primarily for the radio industry, although they filtered through to recording studios. They were particularly useful in radio studios, where a programme might have had a mixture of studio interviews, live music, outside broadcasts, pre-recorded music played from shellac discs, all kinds of different sounds recorded at different levels.

All of those things had to be broadcast as loudly as possible in order to cut through the static and hiss, but not too loud otherwise the signal would distort. In theory the producer could move the volume slider up and down while the programme was playing back, but a compressor could do it automatically, without having to listen to the whole programme in advance, and it could do so quicker.

This is essentially what RMS compression is like. It smoothly keeps the music from getting out of hand. But paradoxically compression can be used to increase the overall volume of a piece of music. Contemplate the following four waveforms:

I downloaded a book a while back. Stockhausen Serves Imperialism, it was called. I was expecting a nice big dose of imperialism, but instead it was a bunch of moaning about conceptual music.

The first waveform is a little bit of music without any compression. It sounds fine, although you have to strain a bit to hear the quiet bits. Classical music is often recorded like this, with minimal or no compression, because classical producers want to capture the original performance without colouring the sound.

With the second waveform I've applied a really harsh compression algorithm. The loud bits are now the same volume as the quiet bits. With the third waveform I've raised the overall volume so that the signal is as loud as possible without distorting. It sounds punchier, more in-your-face than the first waveform. Perhaps too punchy, because the compression was really hard.

It could be worse, though. With the fourth waveform I've turned the volume up until it distorts, and then I've turned the volume up even more, until it sounds obviously distorted. This is a technique that became popular in the 1990s. How better to make a record stand out on the radio than by making it really, really loud?

The fourth waveform would probably sound great coming out of the speakers in McDonalds, where no-one cares about sonic fidelity. It's torture to listen to with headphones though, or at least it's torture to listen to for very long.

And that is RMS compression. A tool for good. A tool for evil. Like pliers. A tool. The NanoCompressor is perfectly anonymous in RMS mode. The attack and decay controls are disabled, so you only have control over the threshold at which compression occurs and the ratio of compressed to non-compressed signal. What does RMS stand for? Root mean square. What does that mean? "It's the rough average". I don't know, I'm not a mathematician.

There's another kind of compression called peak compression, which is often used on individual instruments, commonly the bass and the drum. Those two instruments are key components of modern beat music. The NanoCompressor's peak compression mode has attack and decay controls, which add an envelope to the compression.

If you're used to synthesiser ADSR envelopes peak compression is a little bit paradoxical, because a slow attack actually makes the sound more aggressive. Think of a compressor's attack and decay controls as an inverted ADSR envelope, because they add quietness.

Contemplate the following two waveforms, which are a bass note put through the NanoCompressor:


At the top the NanoCompressor has a fast attack. It instantly clamps down on the bass note and keeps it quiet. At the bottom I've slowed down the attack, which has the effect of slowing down the onset of compression. Notice how there's a little peak at the beginning of the note. That's the original signal, before the NanoCompressor clamped down on the volume. The result sounds punchy, aggressive, with a little POP at the beginning of the note. If it's confusing, just train yourself that slow attack = punchy sound and not the other way around.

Compressors can do a third thing. Look:


That's an example of side chain compression. If you put a signal through the NanoCompressor's side chain input, it uses that signal as a trigger for the onset of compression. In this case I've fed a simple pop-pop-pop bass drum into the side chain input. Notice how the bassline seems to go ooowop-ooowop-ooowop, as if the bass drum was punching holes into it.

When used subtly the side chain input is handy for mixing the bass and the drum. They occupy the lower frequency range, and if you've not careful it's easy to have the bassline overlap the kick drum. The end result just sounds muddy and indistinct. Side chaining the kick drum to the bass separates the two so that the bass quietens down when the kick drum sounds, so that you still hear the pop-pop-pop of the kick drum.

Side chain compression can be used unsubtly, as in the video above. Dance music producers of the late 1990s latched on to it, but the example I'm going to post is much more modern. This is Low, from their 2018 album Double Negative:


The whole album is masterclass in the creative use of compression. Notice how the crunchy not-quite-bass-drum seems to punch holes in the rest of the music, and then it takes a split-second for the background music to recover? That's side chain compression.

As mentioned up the page I found the NanoCompressor slightly disappointing, because it's neither good nor bad. It doesn't have much character, and it's less flexible than a software plugin. There's nothing wrong with it, but what's the point? Nonetheless I decided to make a whole track with it:


I played each part with a modular synthesiser fed through the NanoCompressor in peak mode, with the modular plugged directly into the NanoCompressor, which in turn was plugged into the audio interface. I didn't use it as a send effect. This setup bakes the compression effect into the recorded audio, but this just meant that I had an incentive to get it right first time and stick by my decision.

I added delay and reverb with Logic, and in the video I'm running the entire mix through the compressor in RMS mode. I didn't use the sidechain input, and all of the voices were recorded onto individual tracks. From what I remember the knob settings pictured in the video are more or less what I used throughout when it was in peak mode.

I haven't mentioned the controls, have I? From left to right the NanoCompressor has THRESHOLD, which determines the level at which compression begins; RATIO, which determines the strength of the compression; a tiny HARD/SOFT button that determines the steepness of the volume slope, although I could barely detect a difference; a PEAK/RMS button; ATTACK and DECAY knobs; a button that determines whether the level meter shows the input or the output level, and finally a bypass button that turns off the compression. And a volume knob.

As you can hear, when used subtly the result is mostly transparent, although there's a noticeable punch to the bass, a kind of aggressive thwap sound. Musically I was inspired by Manuel Göttsching's E2-E4, which is a forty-minute-long synthesiser jam from the early 1980s that was apparently a favourite of top New York DJ Larry Levan, because he could play it while going to the toilet.

And that's the NanoCompressor. Is it better than Logic's built-in compressor plugins? No. It doesn't even save much CPU load, because compression plugins aren't particularly CPU-heavy.

Is it sonically characterful? No. It's a technically competent, transparent compressor. You might appreciate it more if you're playing live without a computer, or if you're a guitarist and you want it as part of your pedalboard, assuming you can get it to fit. Perhaps you want to make a big chain of analogue effects. On the positive side of things it does nothing wrong, and it's cute, so there is that.