The third instalment in our series on creative percussion programming looks at trigger-to-MIDI converters, and explores their usefulness for triggering drum samples and/or sounds.
WE STARTED AT the ear and are now working backwards. In the first instalment, we explained why drum machines sound like machines, and offered a few quick tips on how to get around it if you happen to own one. In the second, we suggested using samplers as replacement sound generators for drum machines, and gave tips on how to program those
more effectively. We also hinted that playing plastic piano keys may not be the ideal method of triggering these sounds.
In this instalment we'll be looking at the next step upstream - trigger-to-MIDI converters, the things that turn a physical, or in this case electrical, "thonk!" into something which can play our wonderfully programmed sounds.
We got our hands on seven different trigger-to-MIDI converters from six different manufacturers, representing a variety of features and costs - the Translator 2 from US company Phi Tech, the DZ1 from Casio, the Octapad from Roland, the TMI and MTM from Simmons, the SC40 from Pearl, and the PMC1 from Yamaha.
It's worth mentioning up front that while we originally approached this set of reviews from the angle of a keyboardist (or other non-rhythmic musician) looking for good ways to trigger percussion sounds, we ended up being fascinated with the keyboard-oriented possibilities these devices opened up for drummers. Fledgling young Bill Brufords, sit up and read on.
What follows is a capsule review of each conversion system, in price order, from the bottom to the top (or vice versa, depending on your current financial status).
Phi Tech's Translator 2
($250 in the States) was the least expensive and the most basic of the interfaces we tested.
Front panel "controls" consist of a power LED, a single "trigger happening" LED (alas, not one per input), and a "panic button" to kill droning notes. The back panel consists of six ¼" input jacks, six ¼" output jacks, MIDI In, MIDI Out, and a jack for the wall-mounting power supply. Notice anything missing? Well, there are the "internally adjustable 'set and forget' parameter controls over individual pad sensitivity, separate MIDI In/MIDI Out channel assignment, velocity tracking, and selection from a choice of eight pre-programmed MIDI note assignment options". In other words, you have to open the small, sturdy metal case to get at a set of miniature trimmers and DIP switches to set up the device. There are sensitivity trimmers for each input, plus a master velocity curve trimmer (ie. how subsequently stronger thonks translate to various MIDI velocity levels), MIDI channel switches, and the aforementioned note assignment options. These options are pre-programmed for certain note values to match up with a handful of drum machines, along with forming a C Major or C# Pentatonic scale.
Hint: have some small tools available when setting up this box - those controls are small
. And even though hitting the Panic button is supposed to reset the box after changing an internal DIP switch, we found we had to unplug and replug the box more than half the time to "unfreeze" it after making a change. Once the lid's back on, things are cool.
If it's not obvious already, this box is more suited to fixed setups where not much variation is expected (the controls are fiddly, and who wants to keep opening their box?) such as a live gigging situation. However, if you are indeed using a sampler for your playback device, you can map and transpose your various drum and percussion samples to match up with the preset note configurations. An additional bonus is the ability to translate MIDI In to trigger outs - nice for driving an existing electronic drum kit (the designers' intention) or perhaps a cheap monophonic analogue synthesiser (our idea); we've heard some great electronic toms come out of a Pro One...
The Translator 2's most important feature is its price. It doesn't present the same over-the-top creative options as some of the other boxes we'll be looking at, but it does its job with little hassle, and the manual has a good, sardonic sense of humour - what more can one ask of a drummer?
Casio seem dedicated to making technology affordable. True, occasional corners are cut from the "pro" point of view (when are they going to make something - anything
- in rack-mount form?), but on the other hand, they also manage to throw in a surprising number of well thought-out features.
The Casio DZ1
(£199) is another such example. On first encounter, a "pro" sees things to gripe about - the rather nice metallic grey and aquamarine case is not
rack-mountable, but is about the size of a hard cover book and about the weight of a football (not very confidence-inspiring). It comes with no power supply, but six AA batteries are included, and the manual also gives instructions on how to run the unit with a car battery. As a matter of fact, a lot
of Casio equipment is battery-powered, which brings us Californians to wonder when their battery-powered amplifier/speaker is coming out, so that we can actually take these things to the beach.
OK, let's remind ourselves of the price, and keep going. The back panel is generous, with room for eight trigger ins, four preset selector footswitches, and a hi-hat switch (all ¼"). Two wired-together MIDI Outs are provided (that's one less MIDI Thru box to purchase), along with a nine-volt power input, power switch, and overall sensitivity control.
The top panel includes eight sensitivity sliders, a selector button per "translator" (Casioan for "trigger input"), four preset selectors, four mode selectors, a three-digit LED display, and a pair of scroll keys (used for parameter editing only - thankfully, they are not needed to access parameters, as they are on so many synthesisers today). Each trigger input and edit mode switch has an associated LED, telling the user precisely what's going on - a basic human need overlooked on other, more expensive devices.
The front panel is virtually the manual - there are four presets, and each "translator" may have a MIDI channel, note number, and program change number selected for each one. When a new preset is selected, the program changes are sent, and various modulations are zeroed and the pitch wheel recentred.
But reading the inoffensive manual reveals a couple of other thoughtful features. The first is that hi-hat switch. While the footswitch is left alone, one translator's note number is sent per thonk at input 8. While the footswitch is held down, a different note is sent. At the time it is actually depressed, it also sends a note, at a selectable velocity. In other words, it can work just like a hi-hat (or at least as close as any drum machine has got so far). Just line up one note number with a closed and another with an open hi-hat sample. Of course, other samples (slap bass, for example) may be selected just as easily - or this switch could be used to extend your kit out to nine sounds (the footswitch can be programmed to be silent when struck).
Another thoughtful feature is a mode in which the velocity of a selected translator is displayed when struck, including an overload indicator, which is nice for the initial setup of a trigger-to-MIDI interface (and for when you're unsure as to just which velocity range should be producing which results).
Finally, while you're editing a particular trigger's parameters, all others are blocked from being transmitted. This is a quick and easy way to figure out which pad is which, even though it would have been nicer if the trigger number followed the last pad hit instead. Sigh. Almost nobody does this.
Disadvantages? Not many. The biggest is being restricted to only four patches. Once you start using any of these interfaces beyond the drum machine or drum kit level, it quickly becomes clear that four presets are not enough. And the DZ1 is the only converter not protected by a metal case. Don't Casio realise that drummers are... are... animals?
(Of course, more sedate keyboardists would be safer owners.) And coming with batteries instead of a proper power supply borders on the insane.
But just remember: this is the second cheapest device of its kind, competent, and by far the easiest to use.
The Simmons TMI
, which retails for £249, is a rack-mounting module with eight sensitivity controls, a twin seven-segment LED, and an otherwise minimal front panel. The back panel consists of eight trigger in and eight trigger out jacks, along with MIDI In, footswitch, cassette (an unusual multi-pin DIN connector), and "suitcase kit" jacks. The trilingual manual is a mite cryptic and wordy in places. For example, "channel" is used to refer to trigger/pad inputs, not MIDI channels, resulting in sentences like "each channel in each patch can also be sent down a midi (sic) channel, to see which midi channel has been programmed for each of the TMI channels, press the DISPLAY button (A), you will see that the channel led lights up in the display (B) and the display has changed to show one of the 16 midi channels that the particular TMI channel is being sent down (C)." Reading manuals like this one contributes to the supposed illiteracy of musicians.
Additional negative points include the fact that numbers greater than 99 cause a decimal point to come on in the display to indicate 100 onwards, and the autoscroll (for value adjustment) seemed a bit overeager - we often overshot values. Aside from that, the manual's illustrations are helpful and, in general, the TMI is easy to learn.
Having trigger sensitivity knobs on the front panel is a big plus, and there is an LED under each one showing current trigger activity. Despite that, we still had a couple of problems with crosstalk rejection, where hitting one pad on a stand caused another one to trigger.
Each trigger input (not "channel", damnit) can be sent as a particular MIDI note on a particular MIDI channel or be silenced altogether in a patch. (Because the trigger ins are echoed back out, presumably, in their case, to an electronic drum kit - you may not want
MIDI doubling up your thonks on some patches.) There are 50 of these patches (each of which may send out a MIDI program change) that can be chained in various "sequences" (not to be confused with sequences of notes - language problems again) and stepped through, back and forth.
A real meat-and-potatoes sort of machine, this. Nothing fancy, but all of the basics are there times 50 patches, and you don't
have to open the chassis.
It's tempting to think of the venerable Roland Octapad
(£399 list) as eight rectangular pads in one case with a MIDI Out. And indeed, we'll be looking at it in that
context next time around. But what you may not realise is that it also has six trigger inputs that it converts to MIDI - in essence making it a 14-"channel" thonk-to-MIDI converter. The additional trigger inputs behave identically to the built-in pads with respect to programming options, and allow you to use other outboard pads (or electronic kick pedals, or whatever) at the same time.
One immediate and obvious advantage of the built-in pads is that you save the money that normal pads would cost you (not to mention stands to support them). These would be a necessary expense because the other trigger brains pretty much demand them to get started. The Roland pads, although small, feel just about as good as most of the free-standing ones, and (thanks to some clever crosstalk-cancelling electronics inside the box) are entirely
isolated from one another - not all pad/brain combinations can make this claim.
While banging out hard rock rhythms in a space the size of a large TV dinner may not be your idea of an ideal performance situation, there is something to be said for the technique-related aspects of having all your pads so close together - whether you're a drummer or not. And anyway, you just try carrying anyone else's trigger converter and eight of their pads in one hand (juggling counts).
The Octapad stands alone among this group in its very simple and elegantly-implemented programming system. All you have to do is choose a parameter for editing, then hit a pad (internal or external), and that pad's values are displayed and may be changed. Once you've worked with it, it seems incomprehensible that no other thonk-to-MIDI converter is set up this way.
The edit mode is either global (edit all pads at once) or per individual pad, indicated by slow or fast flashing of the two-digit LED display. There are but six editable features - unfortunately, there are only four parameter buttons to hit. It is somewhat annoying that two of them are accessed as hidden controls (hit the "A" and "B" preset buttons while in edit mode), although there are few MIDI percussion interfaces which are not
guilty of some equally heinous trespass against reasonable programming methods. At least these two, pad sustain (time between note-on and note-off from a pad hit) and minimum MIDI output velocity, are less subject to frequent adjustment than the other parameters, which include MIDI channel, MIDI note number, trigger velocity curve (a generous five choices), and pad sensitivity.
Being able to set up different sensitivities for each pad is useful, even though there are built-in pads. This is particularly true, for example, when pads are mapped to multiple sound sources with different velocity characteristics.
Other spiffy features include a built-in merger that passes incoming MIDI information to the MIDI output (for chaining multiple Octapads, in case 14 thonks is not enough) and the ability to send program changes to your sound boxes in response to pad strikes (while also depressing the appropriate footswitch).
There are, of course, some areas in which the Octapad is lacking. The most oft-cited one is that like the Casio DZ1, there are but four presets. There is no way whatsoever to back up this preset information - perhaps its function as a controller does not justify this need in some people's minds, but it could
have allowed off-line storage and retrieval of more than four presets.
What the Octapad ultimately does is fairly bare-bones - one pad strike translates to one MIDI note with variable velocity (that does not go down as far as it could, by the way), so some of the tricks from the more expensive boxes would really sweeten the pie. And in place of "mode" LEDs, the display blinks at different rates to state the mode, which can be annoying to some of us higher-strung types (doesn't a blinking light mean that something's about to blow up?).
But overall, a very useful and cost-effective unit.
Speaking of blinking displays, the first thing we noticed about the Yamaha PMC1
(£599) was that it had five large, red, seven-segment LED displays, as opposed to the common (and expected) Yamaha LCD. The second thing we noticed, right after we switched it on, was that the information in these displays blinked, alternating semi-leisurely between different sets of values. There - we have 80% of our complaints about the Yamaha out of the way.
Let's get another 5% or so out of the way next - the trigger in connectors on the back are all XLR. Fine for Yamaha's own pads (which even come with their own XLR cables) and early Simmons, but it means adaptors or expensive cabling for most other pads.
On to more positive things. The rest of the back panel connectors include one MIDI In and two MIDI Out jacks, footpedal and footswitch inputs, and a memory select footswitch input. The front panel of the 2U-high case includes eight trigger sensitivity controls (easier to grab and more conveniently spaced than any other interface we tested - ergonomic points) with eight large "select" buttons that have imbedded activity LEDs underneath. The five LED digits are split into two groups, each with its own inc/dec switches. The remaining controls include 16 function select switches (with their functions labelled on them) and a DX7 RAM cartridge slot (!) for downloading programs to an off-line FM brain. Indeed, the PMC1 came with a QX1 disk and a double-ROM cartridge of FM percussion voices which, while perhaps lacking a little in variation, literally blew us away with their bite and frequent realism. The voices in particular took advantage of the fixed-versus-pitched operators in the DX7, resulting in different keys transposing the "shell" sound while keeping the skin, snare, and other noise components of the sound constant across the keyboard - in other words, they transposed gracefully.
If you buy a piece of Yamaha gear, they expect you to own nothing but Yamaha gear in the rest of your system. The manual is written from this point of view, and something like the PMC1 is optimised to take advantage of such a setup. Given that, Yamaha take good care of you and present you with a well-integrated system that has a ton of power. Your mission is to fill in your own brand names (and MIDI implementations) in place of Yamaha's to take full advantage of the system - which ain't that hard.
For example, Yamaha assume that you are driving one of their drum machines and a DX/TX7. You're supposed to plug your (Yamaha) drum machine into MIDI Out B and your (Yamaha FM) synth into MIDI Out A. The MIDI In, which presumably connects to your (Yamaha) sequencer or master keyboard, is echoed to both Outs. Since your (Yamaha) drum machine maps each of its drum sounds to a selectable MIDI key, all MIDI Out B transmits is singular note messages (programmable per pad) on a selectable MIDI channel.
We hooked a sampler loaded with drum sounds to MIDI Out B and it worked just fine - and so would practically any MIDI'd drum machine. But since your (Yamaha FM) keyboard on MIDI Out A has considerably more flexibility in terms of transposition, chords, gate times and so on, this is where the fun really begins - and where we started to cross the line into using drums to create new "keyboard" lines.
Each trigger input may cause transmission of up to five selected MIDI notes. The gate times for these notes range from nothing up to about three seconds (fixed). The footswitch can be used to "hold" these notes on, and this feature may be programmed per pad (we found this very useful for drones or, say, phrasing a bassline from the bass drum pad).
Velocity may be used to decide how many notes of the programmed chord get sent - this proved to be a very useful "punctuation" feature, either for thickening up drum sounds (Yamaha's FM drum sounds respond very well to just whacking a fistful of keys at once) or bringing in notes and sub-octaves of a musical chord. The display shows these by alternating between the basic note and these "simul-notes". There can be a different MIDI channel assigned for each trigger in.
Now we start having real
fun. There's a feature called "dynamic shift" which allows you to have the velocity information determine the transposition of a single note (the other four are dropped) instead of determining how many notes of a chord get played. The transposition can be up or down, and selectable up to an octave. If you're driving percussion sounds, this means that different pitches or versions of a sound can exist under one pad. If you are driving a keyboard sound, you have just crossed into the realm of free music.
This mode provided an unexpected amount of enjoyment. As mentioned above, we had MIDI Out B driving "normal" percussion sounds on a sampler, and MIDI Out A driving a Prophet VS in split mode with the bass drum playing a monophonic bass patch and three tom pads (one in chord mode) playing a belllike sound. We have taped documentation of what this sounds like - and it's pretty intense. At times it even sounded like an ultra-tight jazz keyboardist with a drummer (circa early Frank Zappa).
As time went on, we realised that the PMC1 is really oriented towards keyboardists in a helpful way. (For further reference on the "keyboard potential" of drum playing, pick up a copy of David Torn's Cloud About Mercury
, or Earthworks' debut album discussed elsewhere this issue, both of which feature Bill Bruford doing similar things with his Simmons kit.)
Other features include a footpedal input for sending modulation to your slaved keyboard (it only goes out over MIDI Out A), program selects per trigger input (since each may be on a different MIDI channel) plus a "system" program number (for reverbs and the like), a reasonably generous memory of 32 programs, and a patch edit/recall memory that saved our skin at least once when we made changes we didn't intend to.
The remaining few percentage points of complaints include the wish that notes in dynamic shift mode were selectable, like the simul-notes are, so that scales and such could be programmed (it's hard to play against anyone who could pick any of the 12 semitones in an octave on the next beat). And the hold footswitch could have sent a MIDI hold pedal command - that would have meant we could program two different release times on keyboards that implement a "second release" function instead of just having everybody drone.
The lack of an LCD, weighed up against the wealth of features, meant constant references to the manual to learn the Yamaha, but fortunately the manual is well-written and, in general, we had a lot of fun with the PMC1.
Our modus operandi tends to be checking out equipment and making a lot of noise in general on Saturday nights, travelling into the local beach town (Santa Cruz) on Sunday mornings in search of the perfect brunch, and then tackling one more piece of equipment on Sunday afternoon. The relaxed state of mind this produces proved to be the best way to check out the Simmons MTM
(£599), because it had the most features - and hid them in the most devilish ways.
We don't want to give the impression right off the bat that the MTM was a bummer - like we said, it had the most features, and it proved to be an absolute gas once we figured it out. And if you follow the manual very carefully, you can
figure it out. As with the TMI, however, the manual left us wishing that the British would learn to speak plain English. (Funny; that's what I keep thinking the Americans would do - Ed.
) The front panel features an LCD - great - but it is sunken into the panel and not backlit, so you have to have a bright light behind you and the display has to be dead on eye level before you can read it. The LCD also seems to be an excuse for not including any status LEDs on the front panel, which makes it hard to figure out which pad you're thonking (even Simmons' lower-priced TMI has LEDs per trigger input), for example. The editing procedures also give the impression of climbing down a hole to reach a certain function, and needing to climb back out again before climbing down some other parameter's hole. In other words, it shouldn't be this
hard to use, Simmons people.
On a more positive note, the trigger flowchart is printed on the front panel, with each major section being printed over a membrane switch which helps select that function. The back panel includes eight sets of trigger in and trigger out jacks (both XLR and ¼"), along with MIDI In and MIDI Out jacks. In fact, as mentioned earlier, the MTM shares the rare quality with the Phi Tech Translator 2 of being bidirectional, in that MIDI In can be turned into triggers along with triggers being turned into MIDI. Trigger ins can also be routed to trigger outs, and some cross-patching is allowed.
Like most interfaces, the MTM also includes trigger sensitivity knobs on its front panel. In addition, it happens to have the most comprehensive trigger processing section of any converter we tested, allowing selection between 12 different velocity curves and control over threshold levels and trigger expansion/compression.
On to the fun. The MTM, too, has the ability to play chords from pads. As opposed to individually picking the notes, the user selects from a variety of "canned" chords and sets the root note for them - less open-ended than the Yamaha system, but quicker and also easier for those with little or no knowledge of music theory (ie. many drummers) to use. Velocity can be linked to how many notes in the chord (s) are played, or which single note of the chord is played - much more tonal and less random than the Yamaha's "dynamic shift" effect. (We disagreed on whose implementation we liked better.)
A feature unique to the MTM is its ability to generate "echoes" for each thonk. As with the thickness of the chords, the number of echoes can be derived from velocity. The delay between echoes can be set from 8 to 800 milliseconds, steppable in four-millisecond increments. The echoes can repeat up to 51 times, and successive echoes can even be transposed. Echo decay is set in a percentage - down to 1. These decays are stated as rates - on short decays, you cannot hear the repeats (we would have preferred that it took into account how many notes are in the echo, and calculated things so that the last one ended up at the set decay).
Another interesting feature is the ability to set up a sequence of notes for the MTM to transmit either on successive hits or as "repeats" of an echo. Simmons originally sold this as a "one-pad roll" feature, but (as stated in the manual) it can also be used for basslines and arpeggios. You can even program the phrasing (spacings) of these notes. There can be up to 16 notes in a sequence, and if more than 16 echo repeats are selected, the sequence repeats over and over again. We personally didn't find this feature as useful or fun as the chord features, but it does add another dimension.
The MTM's not done yet, either, as it allows still more fun to be had with velocity. For example, velocity may be routed to gate time and delay. In the former, harder hits hold notes on longer - the only converter we ran across that does this, and very useful for varying how much of a sample gets played back. The latter adds a smearing or delay effect to hits, which also proved to be a useful effect.
Of course, the MTM also has such expected features as the ability to select different MIDI channels per trigger input and the ability to transmit MIDI program changes for each of its 20 patches.
Next up is the Pearl SC40
, which retails for over $2000 in the States, and should be arriving in the UK soon. Why is it so expensive? Because along with being a trigger-to-MIDI translator, it is a sound playback "brain", complete with its own percussion voices. We're not here to make an extensive review of its sound capabilities, but it's worth mentioning that the sounds are an interesting combination of samples and analogue processing that range from snares to tightly looped cymbals to tubular bells to electronic toms. We started out really liking the sounds, many of which were truly unique, but about half of them grew tiring quickly. To each their own though, so listen for yourself if you're also looking for an electronic percussion sound module as opposed to a sampler.
The Pearl has several very nice features, such as an LED bargraph showing the level of the trigger along with an LED bar trigger input, and a mode where the eight front panel buttons played the currently selected sound at eight levels - very nice for working from the front panel. It also had sensitivity trimmers on the front panel - little round knobs with slots in the middle - a bit fiddly and harder to read than Yamaha or Simmons. Continuing downhill (unfortunately), the Pearl won both the Most Cryptic Front Panel and Most Indecipherable Manual awards quite easily. There was also one fatal editing flaw, in that for an edited MIDI keynumber to take effect, you must leave that mode - annoying, to say the least.
The SC40's MIDI implementation is fairly simple and straightforward. There are ultimately 32 kits, arranged in two banks of 16. The unit has three velocity scales that can be applied to both input and output (numbered 0, 1, 2 - how very computerese). It allows selecting if a pad triggers internal sounds, MIDI, or both. The Pearl can be played from MIDI, of course, and the MIDI key is selectable for each trigger.
The Pearl's one sleeper feature is the ability to expand from eight to 16 inputs. The drum kit we used for these tests included seven pads and a couple of bass drum pedals, and it becomes very tempting to have an octave of pads for playing scales and the like. In a live performance setup, you could quickly outgrow eight trigger inputs.
In general, though, unless you're a drummer needing a sound brain along with a trigger to MIDI converter and really dig the Pearl's sounds, this one is the least desirable of the converters we tested.
Several blown eardrums later...
In this part of our quest for The Better Beat, choice is not lacking. There are a few different approaches and a wide range of prices out there in the trigger-to-MIDI world.
A tight budget cries for the Phi Tech Translator. A little more buys the solid TMI, or the rather competent Casio DZ1. A Roland Octapad is perhaps the best buy, because it gives you eight pads and six additional trigger inputs along with some basic, useful functions at a reasonable price. And the Pearl is just plain expensive.
But the real jewels, to us, were the Yamaha PMC1 and Simmons MTM. They do the basic thonk-to-MIDI conversions of the less expensive units, and if you have the extra cash, they open up quite a wealth of alternative approaches to percussive and melodic playing techniques.
On MIDI Delays and Samplers
IN OUR WORK with triggering sounds from percussive instruments, delays become an unavoidable topic. Keyboardists are used to delays - there is always some delay between touching a piano or organ key and the point at which the hammer actually strikes the strings, the switch is tripped, or air pressure builds up in the pipes. With a guitar the delay is a bit more critical - there is always the noise of the pick first hitting and/or releasing the string as a cue that something's happening, even if the amp is 30ft away.
Percussionists, however, are the ones most used to instant feedback - their noisemaker is at most an arm and a stick away, and starts making sound the instant they hit it. Any perceptible delay between a strike and the resulting sound is going to be most alien to them. Therefore, when selecting a sound module for electronic drums or percussion, less delay means a more natural, more responsive instrument.
Also mixed in with that equation is the variance in delay. If one strike has X amount of delay and the second has Y, the groove can suffer from this unpredictability - particularly if the player is a deadly accurate drum machine or sequencer. The general rule-of-thumb is that a delay of 10-15 milliseconds between an action and a sound is perceptible to a musician, and that a variance as little as 2-5 milliseconds against another event or rhythm is enough to change the feel of a drum pattern.
As audio playback devices, samplers are much more complex than the average drum machine, and are expected to produce higher sound quality. As a consequence, in the process of outputting a sound, a sampler has a larger number of steps to go through, and may spend somewhat more time setting things up. Under most conditions, the resulting delays are not highly noticeable. But when an intricate, rapid drum pattern or a drummer playing MIDI pads is brought into the picture, these delays can present a real problem.
Until now there has been little hard information to go by (and no specifications are published by manufacturers), but impressions abound with respect to the relative time delays exhibited by different samplers upon playback. Sloppy sample editing (ie. leaving a bit of silence or noise on the front of a sound) can add further to the inherent delays. Of course, to avoid getting bitten by this, it is best to test a sampler in the exact configuration in which you intend to use it before
plonking the pound notes on the counter.
We set out to test the delays inherent in various sample playback units - the things we advocate in our Beating
series for percussion playback. The delays involved in converting a trigger to MIDI, and in transmitting the MIDI note itself, are in the order of a millisecond apiece - miniscule by themselves. And by now you're sophisticated enough to realise that such things as length of cables contribute mere microseconds which are certainly not audible - and they tend to be fixed.
Therefore, we built the adjoining table of times (Table 1) based upon how long it took 10 samplers to convert MIDI into a sound. To do this, we set up a repeating sequence of identical MIDI notes driving a sample that had been truncated down to a very short burst. We used an oscilloscope to display the MIDI stream on one line and the audio output on another. Then, we measured the time between the last bit of the MIDI note-on message and the start of the sound. What we found is ranked by average response time (fastest ones first).
A couple of the numbers in the table need additional explanation. The EII is commonly maligned as having a built-in 15-millisecond delay. Thus, we were surprised by the numbers we were actually measuring. However, careful study of the amplitude shape of the sound coming out of the EII proved interesting: even though the sound had started, it was obvious that the amplifier's envelope had not
opened up all the way yet, and actually needed another 5 milliseconds or so to reach full amplitude. This resulted in a rounding of the attack of the sound, and explains why many people (including E-mu themselves) append a little bit of silence on the front of their sounds to let the VCA open up all the way. Other ways around this are to leave the VCA release envelope out at its longest time (which means for rapid-fire notes, the VCA doesn't get a chance to close down, and therefore opens back up to full right away), or to try out the hardware modification E-mu recently announced to get around this problem.
On the other hand, the VCA on the Prophet seemed to open up about a millisecond before the sound actually started, meaning that nothing got rounded off, but an additional bit of delay got added. The two fastest machines - the S900 and SP12 - do not have a full complement of envelopes (the Akai lacking a filter envelope and the SP12 any at all), a fact which may help their performance. These times also validate their current popularity as percussion sound brains. We were frankly shocked at how fast and steady the DSS1 and Studio 440 were (true, the DPX1 was faster, but varied), and disappointed at the wild variance of the Rolands.
In short, these are some stats to bear in mind before purchasing your drum brain - important ones that you probably won't find in any glossy brochure.
Table 1 (all figures in milliseconds)
|Sampler ||Fastest Response ||Slowest Response ||Average |
|Akai S900 ||1.5 ||1.5 ||1.5 |
|E-Mu SP12 ||2 ||3 ||2.5 |
|Oberheim DPX1 ||2 ||6 ||4 |
|Korg DSS1 ||4 ||5 ||4.5 |
|Sequential Studio 440 ||5 ||5 ||5 |
|Ensoniq Mirage ||4 ||8 ||6 |
|Roland S50 ||2 ||11 ||6.5 |
|E-Mu Emax ||5 ||9 ||7 |
|Roland S10 ||4 ||10 ||7 |
|E-mu Emulator II ||7 ||10 ||8.5 |
|Sequential Prophet 2000 ||9 ||11 ||10 |