In the first part of an occasional series, we look at how the new MIDI Sample Dump Standard can be used in conjunction with the Prophet 2000, Oberheim Prommer, and Digidesign Sound Designer and SoftSynth programs.

A SCANT FEW YEARS ago, each synth or drum machine was an island unto itself. You played it. No other synth or drum machine was allowed access to what it was doing or how it was doing it.

As these machines became more microprocessor-based, a few manufacturers started allowing some of their own machines to speak to each other - Roland's DCB and the Oberheim System being notable in their field for this breakthrough.

And then came MIDI. Fast-forwarding through the first three-plus years of MIDI's life (yes, it really is that young/old, depending on your perspective), not only can machines tell each other what notes they're playing and with what program, they can now tell each other the time of day (MIDI Time Code) and exchange actual sound data itself (via the MIDI Sample Dump Standard).

For those who tuned in late, the MIDI Sample Dump Standard (SDS for short) is a universal way of transferring sample data between devices (samplers, computers, and so on) using MIDI System Exclusive messages. They were the first messages to use the new Universal System Exclusive area, reserved for all manufacturers as opposed to being manufacturer- and product-specific.

SDS was co-developed by Sequential and E-mu in 1985, the hope behind it being that it would be an easy way for users who own more than one piece of sampling-based equipment to exchange samples between them without any signal loss (resulting from going through an analogue conversion stage), and also that "generic" sample editing and synthesis packages would appear that could be used with any machine that supported the standard.

When the SDS first appeared on Sequential's Prophet 2000 a bit over a year ago, all of the above was mere fanciful speculation. Now that it's early 1987, none of the imagined processing packages have appeared, but several sampling-based instruments now support the SDS, and the time has come to evaluate what can be done with this aspect of the MIDI spec.

Now, I've had an Oberheim Prommer (see review elsewhere this issue) around the homestead for the past couple of weeks. A fellow writer and musician (Sequential's Stanley Jungleib) spotted it in my office one day at work, and it occurred to us both simultaneously that he had a Drumtraks drum machine that could benefit from my temporary possession of this device. Stanley and I both have Prophet 2000s, and fairly extensive libraries of percussion sounds on disk. I also have the Digidesign Sound Designer (a sample editing package) and SoftSynth (an additive synthesis package) for my 2000, and was interested to see how they could help me in creating sounds for the Prommer. So, without much further ado...

Prophet to Prommer

AN UNFORTUNATE FACT of many of today's samplers is that the audio input stage is one of their weaker links. To build a true brick-wall filter at the Nyquist frequency to kill any aliasing effects is expensive, tricky, and would introduce a large amount of phase shift at higher frequencies, anyway (audible as a slight smearing of the attack transient). To send this signal back out of a sampler, through an inevitable output filter and into yet another sampler's input stage with the additional phase shift and accumulated noise is not a prospect you or I should jump at considering. So, the ability to transfer sounds digitally with no generation loss via (you guessed it) the SDS immediately seems to be a more attractive proposition.

Stanley selected a set of seven sounds that he wanted burned into ROMs for his Drumtraks, and gave me the Prophet 2000 disks. I loaded these sounds one by one into my 2002, and made a working copy. Next, I referred to the back of the Prommer manual to see how long the sounds should end up in the Drumtraks, and truncated them on the 2002 until I was happy.

I then used the "Request MIDI" function of the Prommer to transfer the sounds over from the 2002, as opposed to sampling them. The 2002 uses a 12-bit linear format, as opposed to the Prommer's (and Drumtraks') eight-bit COMDAC format. Fortunately, the SDS takes this into account and forces the devices to convert to and from a common linear format.

In general, I was pleasantly surprised with how good the samples sounded in the Prommer, with the exception of the acoustic bass - it had taken on the hiss-like noise common to what eight-bit COMDAC inflicts upon sounds lacking an abundance of high-frequency components to mask the noise.

It was at this point that I spotted my first mistake - the Drumtraks uses a lower sample rate than the 31.25kHz that the 2002 sounds were created with. There is no way to do sample-rate conversion on the 2002 short of playing the sounds an octave higher, but fortunately, the Prommer has a "squash" feature that effectively halves the sample rate. So, I went back to the original samples on the 2002, made new truncated versions that were twice as long (since squashing was going to halve their size, too), and retransferred the sounds to the Prommer. I then "squashed" all of the samples on the Prommer.

The unfortunate side-effect of this is that it also halves the bandwidth, thus cutting some of the high end. However, they sounded cleaner than when I sampled at 15.625kHz on the 2002 (less clock and aliasing noise - the clock part of this I have to attribute to the Prommer's output filtering; the aliasing to the Prophet 2002's input filter at this frequency).

The Prommer also has a utility which envelopes the end of samples down to silence (dedicated drum machines generally do not have VCA and VCF filters to do the same thing) which I used to clean up the sounds further. (The 2000's VCA and VCF envelopes have no effect on sample data sent over MIDI.) When I occasionally messed this up (enveloping of this kind does irretrievably alter the sample data), I just retransferred the sound over from the 2002.

When I finally got round to burning ROMs for the Drumtraks, I ran into my first problem again - the sample rate still wasn't low enough for the Drumtraks, with the lowest tuning being just below the sample's original pitch. The playback rate of the Drumtraks varies from 2.3kHz to 25kHz (dependent on sound and channel), which means I probably wanted an effective sample rate of somewhere around 8-12kHz. I could squash the samples again, giving me an effective sample rate around 7.5-8kHz, but at this low a rate (and bandwidth) I would have preferred sampling the sounds into the Prommer at 12kHz with some form of pre-emphasis equalisation to compensate.

It's worth mentioning that I would have been better off if my eventual target had been an Oberheim DX/DMX (playback rates 12-32kHz) or a Linn (playback rates 24 or 32kHz), or if my 2002 sounds had been originally sampled at 41.67kHz (double squashing would have brought the effective sample rate down to just under 12kHz). As it was, I was thwarted, through no fault of any one piece of equipment, from doing everything digitally and still having the desired end result.

More of SDS

ANOTHER USE OF the Sample Dump Standard is to use features of one machine to process a sample you intend to play back on a different machine.

For example, there were several instances where the Prommer had editing functions essential to tailoring sounds for ROMs that my 2002 did not have. At first, I considered using the Prommer to edit sounds that I would then bounce back to the 2002 for use, but I quickly discarded this idea when I remembered that the Prommer was going to reduce everything to eight-bit COMDAC for its functions, and that I wasn't going to get my 12-bit linear fidelity back when I retransferred the sounds (some data formats "munch" or degrade the sample in interesting ways - unfortunately, I do not consider the shortcomings of eight-bit COMDAC to be very interesting, except for the possible brightening of a snare drum). However, with Sound Designer and SoftSynth for my Prophet 2002, my mind immediately went thinking of what I could do with those to augment the features of the Prommer.

In addition to the obvious advantage of visual editing, Sound Designer has some features that the Prommer does not - namely, the ability to equalise digitally and to crossfade from one sound to another. The digital equalisation would come in particularly handy for doing my "pre-emphasis" if I insisted on keeping everything in the digital domain, but considering that the target format is eight-bit COMDAC with a reduced sample rate and bandwidth, it became silly to bounce a sample up to Sound Designer, digitally equalise it (taking several tries, since I have to wait until the Mac's done number-crunching before I can hear what it's achieved), transfer it back down to my 2002, then transfer it over to the Prommer, where I would then have to squash it to make it match the Drumtraks. Sampling the 2002 through a graphic equaliser at the desired target sample rate would be much quicker, more intuitive, and worth the relatively small loss in sound quality.

Far more interesting is the use of SoftSynth in this system. The "Smart-Synth" function of this package creates several canned "dinks", "bumps" and "clicks" which are fun to graft on to the front of sounds. It's easy to create enveloped noise with SoftSynth for mixing with normal cymbals or toms on either the Prommer or with Sound Designer to create electronic (a la Simmons) drum sounds. And the Prommer has a feature that no other component of my sampling system can offer - a software ring modulator. This function multiplies two sounds together, which results in a sound that has the sum and difference of the harmonics of the two source samples. The usual result is some dangerous sound (unfortunately, more often obnoxious than musical), but this is because the two sounds quite often have clashing harmonics.

I was able to take a tom, transfer it to Sound Designer, calculate its fundamental frequency by looking at its wave, create a sine wave of this fundamental frequency with SoftSynth, then ring-modulate them together on the Prommer. The result was a more in-tune "android" tom than I would have created through random mixing.

Conclusions — So Far

PERHAPS THE ULTIMATE judgement of a feature is whether or not it is so important that it persuades you to buy an instrument. In this particular combination of using a Prophet 2000/2002 and the SDS to justify purchase of an Oberheim Prommer (or vice versa), I would have to say the answer is a (qualified) no.

The sample editing features the Prommer adds to the 2000 (enveloping, ring modulating) are offset by the loss in sound quality (since the Prommer is constrained by the eight-bit COMDAC format common to the drum machines it was designed to support). The improvement in sound quality by digitally transferring samples as opposed to resampling them is lost to the lower bandwidth of some drum machines, but perhaps worth it for those with higher sampling rates. If you already have Sound Designer or SoftSynth and a Macintosh computer, you'd spend less money buying Digidesign's Burner to create drum ROMs. (True, the Prommer is also a sample playback device, which Burner is not, but if you have this system, you already have a sampler, don't you?)

Of course, if you already have each of the components of this system (which I did, at least for a couple of weeks), there are some nice advantages to having the SDS on the Prophet 2000/2002 and the Prommer, which I was able to make good use of.

In future instalments, we hope to be taking a look at using the SDS among some similar machines, including the Prophet 2000/2002, Oberheim DPX1, E-mu Systems Emax, and Akai X7000, paying particular attention to what advantages the SDS brings in sound quality in swapping sounds between machines, along with some moral questions about doing this in the first place. Until then...