Sampling‎ > ‎

Sample swapping


Technology, Digital Sampling

Published by Keyboard/May 1988, posted Februari 2002

Peter Gotcher is the president of Digidesign, a music software company specializing in sampling.

  • HAS THE TIME FINALLY COME TO Upgrade your sampler? Do you have your eye on one of the newer models with higher fidelity, more features) and longer sampling time? The new crop of samplers is attractive, but there's one major problem with buying a new instrument: what happens to those sounds you have grown to love and depend on? Does replacing your sampler mean giving up your current sound library? The answer is both yes and no.

    Why not simply plug the output of the old sampler into the input of the new sampler and resample your favorite sounds? Well, in the first place, the fidelity of a sample degrades every time it is converted to an analog signal and then back to digital data. The degradation might not be very audible if you sample from an eight-bit sampler into a 16-bit sampler, but you are always better off avoiding unnecessary analog-to-digital conversions. Perhaps worse, though, is the notion of having to re-loop every single sample that you convert using this method. If you've spent hours getting just the right loops, you'll cringe at the thought of starting over. And when you're done looping, you can start remapping the samples to keyboard ranges and resetting all of the analog processing parameters (VCAs, filters, velocity effects, etc). Faced with the hassle of manually recreating a library of sampler disks, many musicians have decided that they don't really need a new sampler.

  • Some folks think that the answer to the sample swapping dilemma is provided by an addition to the MIDI specification called the sample dump standard (SDS). The SDS was a
    well-intentioned attempt to allow different make/model samplers to transfer sampled sound data back and forth digitally, including loop points. Alas, between the confusion among manufacturers about implementing SDS and the basic design differences among samplers, SDS has proven to be far from "standard." More often than not, alleged SDS-compatible samplers made by different manufacturers won't successfully transfer samples.

  • And even if the sound data transfers successfully, loop points may not. Some samplers have constraints on where loop points can be placed in asample. For example, the Akai S900 requires the loop end point to always be at the end of the entire sound file, and the Mirage (which doesn't use SDS anyway) requires loop end points to be placed at "page boundries," which occur only every 256 samples. If your SDS-compatible sampler has similar restrictions, the odds are one in 256 that your loop won't be changed. SDS allows two loops to be transferred (one sustain and one release loop), which is fine if you have a Sequential Prophet 2000, but what happens if you have a Casio FZ-1 sample that can contain up to eight loops? If by some miracle both the loop points and sound data transfer successfully, you'll still have to recreate the sample's mapping (the original pitch and range assignments on the keyboard). In most cases, you'll remap the sample only to find that it is out of tune. This happens because the sound is being played back at a different sample rate than the sample rate used when it was originally sampled. Most samplers offer only a few choices of sample rates, with no standard rates in use by the different manufacturers. For example, a sound sampled on an Emulator II has a sample rate of 27,777Hz, but a Korg DSS-1 can only play samples at 16,000,32,000, or 48,000 Hz. Hmm-time to exercise your perfect pitch and use the sampler's fine-tuning to adjust the sample's playback frequency.

    God willing, your sample is now digitally transferred, loop intact, and you've manually retuned it. Now all you have to do is recreate the analog processing parameters, which may or may not behave similarly to those on your old sampler, then repeat this process for every sample on the disk. As you can see, using SDS to transfer a library between samplers can be a huge undertaking.

  • Okay, time for plan B: Use sample editing software and a personal computer to digitally transfer sounds and loops between samplers. Some sample editing programs support several samplers, and are able to translate sound data and loop points between different instruments. If they are stored as a computer file in a standard format, they can even be transferred over phone lines using a modem (this requires patience and a willingness to run up a big phone biII~mpIes are generally very large files). This approach generally works much better than SDS because each sampler driver in the editing software is created specifically for a particular sampler, allowing the computer to compensate for the sampler's unusual characteristics. However, this method still suffers from many of the problems described above. Since only sound data and loop points are transferred, you still have to recreate the mapping and analog processing parameters.

  • Some sample editing programs and sampIers attempt to solve the retuning problem by providing resampling or sample rate conversion capabilities that convert the sample from its original sample rate to the desired sample rate. Many different methods can be used to achieve this effect, with varyingtrade
    offs. Some methods (such as linear interpolation) can convert the sound data fairly quickly, but seriously degrade the sound quality by introducing distortion. To test the quality of a sample rate conversion process, try converting a sample with lots of high-frequency content, such as a crash cymbal. If straight linear interpolation is used, it will sound like a distorted garbage can lid after conversion.

  • The format most commonly used by different manufacturers to store samples on computer disks is "Sound Designer" format, named after the sample editing program.
    However, Apple Computer has recently released a sound file format specification that is more flexible than the Sound Designer format; the latter is limited to a single channel (mono) sample and does not support more than two loop points. The Apple file format supports an unlimited number of channels, accommodating both today's stereo samplers and tomorrow's multi-track sample-to-disk systems. Unlimited loop point markers (or any other type of markers, such as edit points) are also provided. To keep options open for storing data with the sound, Sections of the header (data stored at the start of the file containing information such as name, sample rate, length, etc.) are reserved for use by individual programs. This reserved data space could be used to store mapping information, analog processing parameters, or the editing history of the sound file.

  • It remains to be seen whether software developers and sampler manufacturers will gravitate toward a single standard format for sampled sounds. Like MIDI, it would certainly be in the best interests of sampler users, allowing them to choose an instrument based on its features and fidelity rather than its available sound library. Wouldn't it be great if your sampler or editing program could be taught the characteristics of another sampler's analog processing, sample rates, mapping restrictions, etc.,and automatically convert entire disks while you kick back with a cup of java and the latest issue of Keyboard? It's certainly an appealing goal, but a few obstacles may stand in the way of a widespread standard.

For one, some manufacturers recognize the value (and competitive advantage) of having a large, quality sound library, and have spent a great deal of time and money developing their factory disks. Will they support a standard that might permit users to freely port their library to another brand of sampler, thus forfeiting their instrument's advantage? This may be one situation in which the dynamics of the free enterprise system will thwart a good idea.