In Part 1, we saw how manufacturers realised that putting DSP effects on synths made for great sales. Subsequently, they twigged that it was also a good idea to let us take them off again (selectively), and route and adjust them ourselves.
Like opening Pandora's Box, once the delights of built‑in DSP had been released on to the marketplace at the end of the '80s, there was no going back: the Roland D50 and Korg M1 made it almost impossible for any manufacturer to sell a synth without effects ever again. Side‑by‑side with one of these machines, other keyboards sounded small, thin and dry. People reached in their pockets time and time again for a machine which drowned all the sounds in reverb, choosing them over better instruments that didn't have any onboard DSP. The only exceptions were specialist machines like the Novation BassStation which strove to recreate older instruments. But even they probably lost sales because they weren't heard in every demo with effects plastered liberally all over their sounds!
Certainly the D50 and then the M1 captured synth market shares which hadn't been seen since the days of the DX7. But putting effects in a synth wasn't covered by licences and patents, so there was nothing to stop the other manufacturers following suit, and they did. This didn't happen overnight, as some companies didn't have DSP chips or effects algorithms ready to go, because they'd never written such algorithms before. Companies like Yamaha were better prepared to respond, because they had sold stand‑alone digital reverbs like the REV1 for years, and at the time were enjoyed unparalleled success with one of the first big‑selling multi‑effects units, the SPX90.
So when Yamaha came up with a keyboard with effects in it, it was a killer: the SY77 had both looped PCM samples and FM synthesis in it, but was that why it sold, or was it the effects? Well, the sample‑based sounds were good before the effects were applied, but they were positively gorgeous afterwards, and adding effects made FM sound warm for the first time (prior to this you could only have a reasonable stab by de‑tuning eight lots of the same program on the TX816, but this reduced it to the role of the most expensive chorus unit in history). The only thing which stops the SY77 being my favourite Yamaha product of all time is the SY99, which came out the following year and which was more of the same (but quite a lot more).
Within a couple of years (ie. around the turn of the '90s), 95 percent of keyboards on the market (and 98 percent of those sold) had built‑in effects as standard. But as we saw last month, straight away manufacturers' tech‑support lines started to clog up with uncomprehending users complaining that "program 37 doesn't sound the same when I use it in Multi Mode", or "when I combine my two favourite sounds in Combi Mode, they don't sound right" or (my personal favourite) "Multi Mode is broken". As detailed in the first part of this feature, the growing multitimbrality of contemporary effects‑laden synths was exposing a weakness they all shared: they offered only global effects. The synths sounded great when the salesman played them in the store one program at a time, and the multitimbral demo sequences had been carefully programmed so that the effects complemented all the selected sounds, but when you got them home and started sequencing yourself or making your own Combinations, you soon discovered that one effects processor (however many effects it could do at once) didn't go far when you had 16 source timbres to work with, many of which could sound inappropriately processed when the effects were on and the rest of which would inevitably sound too thin and weedy with the effects off.
We looked last time at some of the things you can do to avoid the most obvious problems — for example, bass drums and basslines turning into inaudible mush because they were being processed by the same expansive reverb that worked a treat on your string pads. You can always work around this by purchasing something cheap to play back the most problematic sounds, but this does rather defeat the 'all‑in‑one' workstation concept.
Faced with these complaints from users, it was clear that manufacturers had to do something to help. Evolutionary products followed the revolutionary products — in Korg's case, the T series and 01/W followed the M1.
The second‑generation products dealt with the effects problem in various ways. First of all, individual sounds could now often bypass the effects altogether. Alternatively, you could sometimes reduce to zero a newly added individual effects send parameter (see the 'Most Effective' box for more on this and the reasoning behind it). This helped enormously in rudimentary mixdowns, but it wasn't everything. Very few of these synths had separate outputs, so you didn't have the freedom to add external effects to individual sounds via an external mixer. Another common solution to the problem of different sounds needing differing amounts of effects was the wet/dry mix parameter but this tended to work better on a single Program or Combi than in Multi mode, where the real problems and conflicts between the simultaneous effects requirements of different programs occurred.
EMU's Master Performance
When Yamaha weighed in with a synth featuring internal effects, the company were able to take advantage of the strides they had already taken in developing stand‑alone digital reverbs. Unsurprisingly, the SY77 sounded great and sold well.
The real breakthrough came when manufacturers started to look at how engineers use effects in conjunction with a mixing desk, sometimes inserting them between source and input, other times sending a differing amount from each channel to the DSP unit and then introducing the returned signal into the final mix output. Thus were born synths with separate insert and global (or master) effects (again, see the 'Most Effective' box for more on this).
Of course this development didn't happen overnight. The first example I came across of just how much difference a great effects structure could make was on Emu's Proteus MPS keyboard. At first I thought it was no big deal, just the best of Proteus 1 and 2 sounds together in a keyboard together with the by‑then obligatory effects (which the original Proteuses — Proteii? — hadn't had), but the more I explored the MPS the more I realised that a properly designed effects structure can make a huge difference to the character of an instrument. As with the original Proteuses, the MPS offered an effects buss to which you could assign varying amounts of different sounds via an effects send parameter, and you could also route individual sounds to a 'Submix' output on the rear panel of the synth for processing by an external effects unit if you wished. Where the MPS scored was that, in addition to this flexible method of working with external effects, it also offered the internal effects the original Proteuses had lacked.
Over at Korg, there were a few more intermediary products with X prefixes, in which they continued to refine the effect structure before they eventually came up with Trinity, which to my ears has the best built‑in effects on the market. This opinion isn't, incidentally, because I recently had an 18‑month spell working for Korg (something that is now over); indeed, I was very cynical about Trinity when it first came out, preferring the much more cutting‑edge Technics WSA1, which could make far more outlandish sounds. But to explain the way manufacturers have moved on from the early implementation of effects on keyboards to the current state of the art, I need a concrete example to refer to, and Trinity is simply the example I know and (after getting to know the instrument inside out) have come to like best.
Wholly Trinity Effects
Emu's Proteus MPS featured an internal effects buss, so that effects could be used in a send/return arrangement, or disabled in a multitimbral setup. Usefully, it was also blessed with assignable Submix hardware outputs, so you could 'break out' a signal from a multitimbral setup for external processing if you wanted to.
A major component of Trinity's high‑quality sound is the quality and flexibility of the effects (which run at 48kHz, like its sound samples). Never was this made more clear to me than when I first heard an external signal passing through Trinity and having the onboard effects applied to it. This for me is the most important aspect of the HDR‑TRI Trinity Hard Disk Recording Option. By the time Korg shipped it, four tracks of hard disk recording was fairly old hat, and now, when computers are doing 30 or more tracks of hard disk recording without even breathing hard, only four tracks is pretty lame. But I still think this option is worth its weight in gold because it gives you an external stereo input (in both analogue and digital) to Trinity's effects (something first pioneered on the Wavestation AD). This way, it really is possible to record an entire album on the Trinity if you wish, passing acoustic sources through the external inputs to the HDR tracks and mixing the eventual output direct to DAT from the S/PDIF Out. Indeed, I did exactly this when recording an album for Steve Fairclough (Korg's guitar product specialist) during my spell with Korg.
Trinity's internal processors refined the concept of insert and master effects (Korg weren't alone, however; Roland used a similar system on the JV1080 and improved it further on the JV2080). What sets Trinity apart in my opinion is the number of different effects that are available, the way that their differing DSP requirements can be mixed and matched for the best possible combination and how the audio pathways inside the machine optimise the way each is used in the final result. However, such complexity can lead to confusion. I will therefore try and give a brief overview of the way things are structured, so that those of you who do have access to Trinity can follow better how things are managed, and those who don't should be to apply at least some of the theory involved to some of Trinity's contemporaries.
Falling In Line
Korg's Trinity features one of the most comprehensive effects sections of any keyboard workstation to date, and its implementation of these effects, even in DSP‑hungry multitimbral setups, takes some beating.
As I explain in the 'Most Effective' box elsewhere in this article, some effects work best by being simply placed between the source instrument or mic and the recording/mixdown medium. None of the original signal is required at the output, so you just take the effected result and pass it on to the next stage. However, any other sound in your mix is probably not going to benefit from the same treatment. So you need to put the effect on the individual sound and make sure it doesn't affect any of the others. Trinity lets you do this by assigning different insert effects to each sound you are using in the Combi or Seq (Multi) modes. In fact, on Trinity, you can assign up to three insert effects to each sound, providing you have enough DSP horsepower left. The only thing that Trinity doesn't allow you to do is put insert effects on the HDR‑TRI's analogue inputs.
"The real breakthrough came when manufacturers started to look at how engineers use effects in conjunction with a mixing desk..."
Size Matters
The problem with offering people loads of different insert effects which they can assign at will is that different effects use up differing amounts of processing power — and however much processing power you put in a keyboard these days, it is never enough for some combination or other of the possible effects. Korg's solution in Trinity is to tell you exactly how much DSP each effect requires. This parameter is measured in arbitrary units from one to four and is called, for want of a better word, Size, the idea being that however many effects you use, total Size cannot exceed eight. Within this one limitation, you are then left to mix and match the effects available. Each effect is grouped with the others of that Size, so you set the insert size and then a pop‑up menu offers you all the effects of that Size to choose from. If you find you can't get the effects you need because the total Size exceeds eight, you have to reduce the number on one of the other effects to allow you to increase the size of the one you are trying to set. The number one tech‑support call is "why can't I increase the size of the effect on part 3 so I can make it stereo chorus?" The answer is usually, "because you have too many effects already on parts 1 and 2". You have to learn to think of the DSP power available as a cake of a certain size: if you make one slice bigger, you have to reduce the size of the other, because the cake is of finite size.
Fortunately, this available size of cake for the insert effects is completely separate from the Master effects section. This means you can assign Trinity's reverb and chorus to the master effects send/return structure without worrying about losing power for the insert effects. On most synths with decent effects nowadays, it is fairly common for master and insert effects to be handled by separate chips, so that you don't have to rob Peter to pay Paul.
Once you have decided which algorithms to assign to the two master effects, you can further determine which of the sounds in your multitimbral setup you want to send to which effect and how much of each sound you want to send, using a slider or rotary control. Having struggled to mix and match effect sizes in the Insert effects to achieve the results you want, you should find the assignment of master effects and setting the amount of each send a breeze in comparison. If you do have access to a Trinity, try taking the most uninspiring group of sounds you come across and see just how far you can take them just using the effects architecture. You'll be amazed at the transformations that can be achieved, though I am sure that any other contemporary workstation or keyboard will allow you the same sort of experimentation and results, even if not to quite the same extent.
Next time, we will start to look at individual effects and in particular the kind of keyboard sounds that you can apply them to for the best cosmetic results.
Most Effective — Insert & Master (Global) Effects
In a traditional recording studio, some effects are applied to the signal before it is recorded, because they have such a fundamental impact on the sound that it cannot be played or recorded properly without the effect being 'printed' to tape (or hard disk these days) along with the original signal. Into this category fall distortion, overdrive and the other various electric guitar effects, the more transparent compression that is used to record vocals, drums and other acoustic sources at a good constant signal level and a host of other effects which radically change the character of the sound.
Guitar effects are a classic case in point; how the pathetic noise that emerges from the average electric guitar pickup caught on, I will never understand, and no‑one is foolish enough to want to mix any of the original signal back in once they have rendered it palatable with distortion/overdrive, chorus, flanging, wah‑wah and the like.
Such effects, when used within synths, are generally known as Insert Effects (because the effect is inserted into the signal path between source and recording medium), although terminology does vary from manufacturer to manufacturer, and the term is also increasingly used generally to describe effects which can be applied exclusively to single sounds in a multitimbral setup without affecting (or indeed effecting) the other sounds in the setup, as we shall see.
Not all effects can be used in the same way, however. As anyone who has ever played through a reverb unit with no wet/dry balance parameter will tell you, you can't throw away your original signal and just use the effected one unless you want it to sound as though you are 10 miles away at the bottom of the Grand Canyon. Because in the natural world we tend to hear a sound directly from the source first and then the reverberated signal bouncing back from the environment it is in next, we need to emulate that in the signal path. As a result, engineers have developed a system for sending a portion of the source sound to the reverb, and then mixing the effected result back in with the source, with the balance between the two giving you a rough impression of the distance between you and the source. All professional mixing desks therefore have several 'send' controls on each channel, where the differing amounts from each signal to be effected are set up, send outputs where the amounts from each channel are collected and sent out to a reverb (or similar effect) and then return inputs and amount controls which allow the effected result to be mixed back into the overall track. As time went on this 'send & return' arrangement, so common with mixers, was duplicated on the internal effects architecture of synths, as detailed in the main part of this article.