REVIEW

Sound Designer

Digidesign
1360 Willow Road, Suite 101
Menlo Park, CA 94025
(415) 327-8811
$349, color or monochrome

Reviewed by Larry Herzberg

Originally written for the Macintosh, this powerful program has recently been made available to Atari ST owners.

A digital sampler analyzes any sound by measuring the amplitude of its waveform many thousands of times a second. Each discrete measurement is called a sample, and the entire set of samples used to reproduce a sound is a sound file. Since a sound file consists of nothing more than a long string of numbers, you might think that most samplers would offer a host of features for manipulating such simple data. If you own one, however, you've probably been frustrated on more than one occasion by its editing limitations. Most affordable samplers are unable to even graphically display waveforms, let alone "zoom in," redraw, cut, paste or copy segments of them. Mixing and merging sounds is usually well beyond their capabilities.

Sound Designer, a "universal" waveform editor and digital signal processor, does all of these things and much more. Originally written for the Macintosh, this powerful program has recently been made available to Atari ST owners.

Looking at sound

For those who have only a vague idea of what digital samplers do, Sound Designer can be a wonderfully educational tool. The manual, which covers the basics of sound and sampling, is clear and concise, but the real lessons are learned when you begin to examine the graphic displays of the sounds themselves.

Sound Designer puts GEM through its paces to make such examinations easy. Once a sound has been transferred from your sampler to the computer, the program writes all the data to a disk file and opens a window displaying as much of the waveform as it can. If the sound file is too long to fit into the display all at once, the window can be scrolled along its horizontal axis until the portion you wish to edit is shown. Up to three windows displaying different sound files can be opened on the screen at once, and no matter how "zoomed in" you might be on the waveform in the active window, an additional "overview window" is always available to display the entire sound. By moving the cursor around in the overview window, you can cause the active window to jump to whatever segment of the waveform you desire, including any portions that may not have been loaded yet from disk to RAM.

Once the proper waveform segment has been displayed, you can magnify any portion of it in one of two ways. By clicking on the horizontal or vertical arrows of the Scale Box, which resembles a small, empty window in the bottom-left corner of the editing screen, you can shrink or stretch either the horizontal (time) or the vertical (amplitude) axes until you've achieved the scaling you want. Then, if you've inadvertently magnified the view so much that the specific waveform segment you were concerned with is no longer showing, you can scroll around the window until you find it.

A faster way to zoom-in is by utilizing the Zoom Box. After clicking on the Zoom Box icon, the I-beam cursor used for basic editing functions is replaced with a cross hairs, and by clicking and holding down the mouse button, a rectangle can be "dragged" around any portion of the waveform. When the mouse button is released, this portion is redrawn to fill the entire window, and the time and amplitude axes are rescaled automatically.

If you wish to mark any position in a sound file for future reference, numbered icons can be dragged from the control panel into the active window. Clicking twice on such a marker displays its editable information box, allowing you to name it and catalog its purpose. To quickly locate any marker that has been placed in the active window, you simply hold down the alternate key and type the marker number. Ten markers are available for each window.

Redrawing waveforms

With each step of magnification, increasingly complex views of the sound are displayed. The differences between the waveforms produced by a flute and a saxophone, for instance, become obvious, as do anomalous noises that you might want to erase, or "draw out" of the sound.

In order to redraw any part of a waveform, you must be almost entirely zoomed-in on it, so that the scale of the time axis is reduced to a few milli- or microseconds. At that level, audible glitches spike out of the rest of the waveform like huge stalactites. To get rid of them, all you have to do is click on the Pencil icon; the cursor becomes a drawing tool that can be dragged across the base of the spike, effectively deleting it.

The loop window

Virtually all samplers allow the user to designate loop points in a sound file when a sustained sound is desired. Looping is necessary because RAM is limited on a sampler. If you use one byte to measure the amplitude of a wave at any given moment, and you sample at a rate of 40,000 samples per second, sampling a ten-second-long piano note will eat up 400,000 bytes!

The way around this excessive use of memory is to sample the note for, say, two seconds, find a segment with relatively stable amplitude and waveform, and, after the initial attack and decay, loop that segment over and over again while decreasing its volume with an independent ADSR (Attack, Decay, Sustain, Release) envelope on the sampler's amplifier circuit; this can approximate the sound's natural decay characteristics.

If you've ever tried to designate good loop points, especially within a complex sound like that of a saxophone, you know how difficult it can be. Unless the sound can be graphically represented in some detail, finding a segment with stable amplitude and waveform is not an easy matter, and that's just the beginning of the problem. If the slope of the waveform at the end of the loop does not match up exactly with the slope at the beginning, a click or pop will be heard every time the loop repeats.

Sound Designer takes the guesswork out of this process by providing a loop window that allows the user to fine-tune loop points. After special loop markers have been placed at the beginning and end of a decently stable portion of the waveform, the center line of the loop window shows precisely how the slope at the end of the loop matches up with the slope at the beginning. By clicking on the left and right arrows under each segment, either loop point can moved backward or forward one sample at a time until the optimal matchup is achieved. This is one of Sound Designer's most practical features.

Editing waveforms

All of the program's basic editing functions are as easy to use as the loop window. After a portion of the waveform has been selected by holding down the mouse button and dragging the I-beam cursor over it, that segment (its color scheme now inverted) can be reversed, creating strange backward effects, or zeroed-out, eliminating unwanted noise at the beginning or end of a sound. The selected segment can also be cut and, if desired, pasted elsewhere in the waveform, or simply copied to another location without being cut. A smoothing feature, when activated before a copy or paste operation, helps make edits seamless by creating a fast cross fade between divergent amplitudes around the edit point.

In addition, a portion cut out of one sound file can be pasted onto a different sound file altogether, as long as the destination file has been opened into one of the three available windows. Sound Designer allows this by always saving the waveform data most recently cut or copied to a clipboard, which becomes disk-based if it runs out of available RAM. Thanks to the clipboard, editing operations can be undone, as long as no additional editing operation has taken place since the one you want to undo.

The digital mixer

Mixing and merging, two of the five functions included in Sound Designer's Digital Mixer (which the user accesses by clicking on the blender icon), are operations that always involve two sound files. Mixing together instruments that play identical parts is one of the best ways to conserve RAM on your sampler. For example, multiple horns performing staccato unison lines can easily be mixed together, and since mixing simply adds waveforms together, sample by sample, several horns mixed into a section will only take up the amount of RAM utilized by the longest single horn sound. Waveforms can be mixed together in any proportion and with a temporal offset, so getting the balance right is just a matter of experimentation.

Merging is a slightly more complicated procedure, but can be useful in obtaining special effects. For instance, you might want to create a hybrid sound with the attack of a piano but the sustain of a guitar. Or you might like the breathy attack of one sampled flute, but the vibratoed sustain of another. To merge two sound files together, markers must first be placed in each; one designating where the merge should begin, the other where it should end. Then, after clicking on the blender icon and selecting the Merge option off the Digital Mixer menu, you choose which order to merge the sounds in, and identify the relevant marker in each sound file. A final option allows you to choose between two cross-fade shapes, linear or equal power. Since there are more variables to take into account with merges, they are a little harder to get right than mixes.

Three other features are included in the Digital Mixer: Gain Change, Cross-fade Looping and Digital Equalization. Gain Change allows you to alter the relative output volume of a sound, either in terms of decibels or by specifying a percentage of increase or decrease. When the program executes this function, it displays the total number of samples clipped (increased beyond the point where digital distortion occurs) as well as the number of continuous samples clipped. The second number is significant because small quantities of continuous samples clipped are often inaudible.

The Normalize option is perhaps the most useful feature of the Gain Change module. This function increases the gain of a sound file until the peak amplitude of its loudest sample reaches 100% full scale (meaning that it is represented by the largest number available in its particular data format). If you're uncertain whether or not a sound file needs normalization, its peak value can be determined by selecting the Peak Value button and clicking on Execute in the Gain Change dialog box.

Cross-fade Looping requires that loop markers already be placed in the sound file you wish to operate on. If you were unable to find "click free" loop points using the loop window, this should do the trick, as long as your loop markers have already been placed with some care. This function mixes data from the area around the loop end with data from the area around the loop start, sort of like a mini-merge. As in the Merge function, you can choose between two cross fade types, linear and equal power.

The final feature of the Digital Mixer is the Digital Equalization (Peak/Shelf EQ) module. Equalization is used to alter the tonal characteristics of a sound, to make it "brighter" or "darker" by increasing or decreasing the amplitudes of a specified range of frequencies. Three modes of Digital Equalization are available in Sound Designer. The Peaking/Notching mode resembles an analog parametric equalizer, with its three parameters of center frequency, band width and boost/cut amount. The two other modes, hi/lo shelving and hi/lo pass, affect all frequencies above or below a specified corner or cutoff frequency. Hi/lo shelving allows you to enter a boost/cut amount, while cutoff frequency is the only parameter used by the hi or lo pass filters.

Frequency analysis

To aid you in altering a sound's tonal characteristics, Sound Designer utilizes the Fast Fourier Transform (FFT) algorithm to create a three-dimensional representation of the waveform's frequency content as it changes over time. In this display, the horizontal axis is the frequency axis, calibrated in kilohertz from left to right; the vertical axis—curiously uncalibrated—represents amplitude, and the time axis extends "into" the screen.

Analysis always begins at the location in the sound file displayed at the left edge of the active window, and can be set to continue for up to 400 milliseconds along the waveform, although a considerable amount of detail is lost when more than 200 milliseconds are graphed.

Interesting as it is to view the tonal characteristics of sound this way, using the frequency analysis to guide you in fine tuning a sound's EQ is not an obvious matter; calibrations on the amplitude axis might help, as would a brief tutorial in the manual. However, just seeing where the bulk of a sound's frequencies lie might prevent you from wasting time trying to alter a band of frequencies that aren't even part of the waveform's structure.

Listening to the sound

Of course, once you've edited a sound file, you'll want to listen to it. Sound Designer offers several ways of doing this. First, by clicking on the Play Sound icon, you can hear the sound through your ST monitor's speaker. You'll probably be surprised by the fidelity of this quick method; it is quite sufficient to audition rough edits or loop points.

You can also connect the ST's audio output to an external D/A box, but the most accurate way of hearing the sound is to transfer it back to your sampler; after all, that's where you'll be playing it from. Transferring a sound to and from your sampler is as easy as clicking on the proper icon on the control panel, but it's not always necessary to send the whole sound file to your sampler. A Preview option allows you to transfer only a part of the sound currently residing in the ST's memory. This cuts down on disk access time, and eliminates a few other time-consuming protocols by always sending the sound to sample Location #1 in your sampler. By pressing the shift key while clicking on the Preview icon, only the loop points will be transferred, if your sampler supports this feature. And once the sound has been transferred, you can play it from the ST using a five-octave, mouse-driven MIDI keyboard, accessed from the drop-down Tools menu.

A few nits to pick

For all its impressive features, Sound Designer is not entirely free of problems, and most of them have to do with the program's disk-based design. As the manual is quick to point out, such a design does enable the program to handle sound files that are larger than available RAM, but it is also slow at times. The clipboard can be configured to operate on a RAMdisk, but this doesn't speed things up much, and if your RAMdisk is large enough to store sound files on, it will probably create more of a slowdown problem than it solves by limiting the amount of memory available for waveform display and playback. If you're operating with only one floppy drive, the disk-swaps required for some merges can drive you crazy. Sound Designer might be the best excuse you'll ever have to buy a hard disk.

In addition to straining your patience at times, the disk-based design does not adequately deal with the all-too-human tendency not to leave well enough alone. Since basic edits are written immediately to disk and other changes are saved when the sound file is closed, it is too easy to make permanent changes you might later regret. This is especially true if the only version of the sound you have is in Sound Designer format, rather than in your sampler's as well.

Two factors exacerbate this problem. First, only those editing operations that use the clipboard can be undone; waveform redrawings or loop-marker relocations cannot be. Second, the way the back-up file system has been implemented is a bit weird. If Back-up Files has been selected from the Options menu, Sound Designer will automatically attempt to create a backup file whenever a sound file is opened. This is fine, but if there is not enough room to create the backup on the same disk as the original, the program just throws up its hands in dismay and announces that all changes will be permanent! Why not simply request that the user insert another disk on which to write the backup file?

Furthermore, even if a backup file has been created, it is always deleted after the sound file is closed, whether or not you've chosen to save the changes you made. Wouldn't it make more sense to leave the backup intact, as an insurance policy against second thoughts? As it stands, you should always make a copy of any sound file you intend to open before running Sound Designer. If that's not possible, use the Save As option to save it under a different filename immediately after the program has opened it.

There are just a few other minor annoyances. For instance, you would think that such a disk-based program would include an option to format disks, but Sound Designer does not. Also, markers occasionally have problems. When one is placed at the very end of a sound file and the scale is magnified a bit, it sometimes remains out of view, even if the window is scrolled as far to the right as possible. If you then click on the bottom-right corner of the window (where you know the marker should be), it might pop up, but that corner of the window will not be redrawn properly.

A more significant problem involves the loop markers: occasionally there is a discrepancy between the time shown in a loop marker's information box and the time initially displayed on its side of the loop window. Whatever the source of this ambiguity, you can eliminate it by calibrating the horizontal axis of the active editing window in terms of consecutive samples rather than subdivisions of time; the program permits this kind of calibration in either decimal or hexadecimal form.

Supported samplers

Sound Designer supports many of the most popular samplers on the market, including the Akai S700/X7000, Akai S900, Casio FZ-1, E-mu Systems Emax, Ensoniq EPS, Ensoniq Mirage/Multisampler, Korg DSS-1, Korg DSM-1, Roland S-10/S-220/MKS-100, Roland S-50, Sequential P-2000, Yamaha TX16W, and any machine using the MIDI Sample Dump Standard. The manual is very helpful in describing the MIDI "personalities" of each of these samplers. If you own any of them and an Atari ST, check this program out.

After receiving a BA in philosophy from (the notorious) Reed College some 13 years ago, Larry Herzberg continued to pursue his interest in existentialism by becoming a freelance musician, writer and MIDI programmer in Los Angeles. Bucking centuries of ST-LOG tradition, he lives with no cats, but does share his apartment with a rather robust potted palm.