Digidesign Sound Tools
With a Macintosh computer and Digidesign's new CD-quality digital audio system, you can cut, splice, and create extended remixes to your heart's content. Paul D. Lehrman discovers what it is like to go tapeless in the studio.
The 'tapeless studio' is a phrase on a lot of people's lips these days. For some, it means a MIDI studio: a computer controlling a host of synthesizers, samplers, and processors. For others, it means a hard disk recording system, like that provided by the Fairlight or Synclavier or one of the newer 'digital audio workstations': sounds are recorded directly to a random-access hard disk, manipulated in various ways in the digital domain, and then spat out as finished productions. MIDI systems can start out cheap and get larger only as the user needs them to. Hard disk systems start out expensive, and get more so.
Sound Tools, from the American company Digidesign, is an attempt to bring the quality and flexibility of hard disk recording down to a more modest, one could say MIDI-sized, price bracket. It provides an environment for recording, editing, and playing back true 16-bit stereo PCM digital audio at a sampling rate of up to 48kHz, so it is appropriate for just about any audio application from film dialogue editing to CD mastering. Although it provides an enormous amount of power, it is surprisingly easy to use.
The system comes in the form of three components: AD IN, a stereo analogue-to-digital convertor in a stand-alone box; Sound Accelerator, a stereo digital-to-analogue convertor and digital signal processor (DSP) on a card that fits inside the host computer; and Sound Designer II, the latest version of the company's well-known sample editing software.
The fourth component required is a Macintosh SE or II computer with a hard disk. (In the case of an SE, the sampling rate is restricted to 32kHz and certain real-time functions aren't available. A version for the new Mac SE/30, which will have the same capabilities as the Mac II version, will be available in a couple of months.) The storage capacity of the system is directly proportional to the hard disk space available: approximately 10 megabytes is required for each minute of stereo recorded sound. The system is deliberately designed to be used with off-the-shelf components, and so as hard disk storage capacities continue to rise and prices to fall, the user will be able to increase the size of his system easily and increasingly cheaply.
In large measure, Sound Tools is an extension of Digidesign's original Sound Designer software, and most of its features will be familiar to those who have used Sound Designer to edit samples. But because the system now includes dedicated hardware, and because it is now highly expandable beyond the storage restrictions of other manufacturer's hardware samplers, it is much faster and more flexible, and can deal with sounds that are considerably longer than what we normally think of as 'samples'. With a 20 megabyte hard disk, it can easily produce a 30-second commercial. With a 500 megabyte disk, it can put an entire CD on line.
Installing the hardware is easy: the Sound Accelerator card fits into a slot in the computer (on the Mac II, any slot will do), and the AD IN then plugs into the Sound Accelerator card by means of a thick cable with DB-25 connectors at each end. Audio input to the AD IN is via two unbalanced high-level (-10 to +8dBm) quarter-inch jacks. The output of the Sound Accelerator card appears at a single stereo (headphone-style) quarter-inch jack, providing a +4dBm signal on each channel. Output trim pots are on the card for use by studios that like to use lower levels, and a splitter cable is included that provides unbalanced quarter-inch (mono) jacks for each output channel.
Recording a signal into the system is simplicity itself. Two front panel knobs on the AD IN box adjust the input level, with green '-20dB' and red 'Clip' indicators on each channel to help out. From the Sound Designer software, a 'tape recorder' window is opened up, which features transport-like controls and LED-like level meters.
The software then presents you with various options to choose from, but if you don't want to get involved with learning what they mean, the default settings are usually adequate. The options let you specify how much of a disk buffer you want to set aside for incoming data; whether you want the disk space used for recording the audio to be contiguous (contiguous recording is less prone to errors, but it usually means you have less space available); whether you want the audio inputs of the AD IN to be echoed at the outputs of the Sound Accelerator card; and the sample rate, from 8kHz to 48kHz, with 44.1 kHz as the default.
Now if you click on the 'Record' button, the system starts to record the sound coming into the AD IN. A running display tells you how much time has elapsed, and how many samples have been recorded. Clicking again stops the recording. You can hear what you've recorded using other buttons for playback, rewind and fast forward (non-audible), and return-to-zero. You can stop recording at any point and start again at the same point, and you can even record over existing data, and punch in and out on-the-fly. If you want to record sound coming off a tape that also has SMPTE timecode on it (for example, a line of dialogue from a video tape that you want to edit), you can instruct the software to start recording when it receives a specific SMPTE frame number (via MIDI Time Code - you'll need a SMPTE-to-MIDI Time Code convertor if you want to use this feature), and stop at another frame. (Punching, however, cannot be automated to SMPTE.)
Once you've recorded a piece of audio - now known as a 'soundfile' - you can view it in a large graphic window, which shows the sound as amplitude versus time. If the file is stereo, there are two windows, one for each channel. The horizontal (time) axis can be scaled from about three seconds full-screen to about 500 microseconds, at which point individual samples become clearly visible. A smaller window always shows the overall length of the soundfile, with a marker to show you what part of the file the larger window is showing. Within the main window, you can select a region (consisting of either channel or both) and perform various kinds of editing functions on it. The program provides two types of editing operations, 'destructive' and 'non-destructive'. Destructive editing actually changes the nature of the soundfile (although you can always preserve the original by saving the altered version with a different name), while non-destructive functions change various parameters associated with the file, but leave the file itself intact.
You can listen to sections of soundfiles with a 'scrub-wheel' function, in which moving the mouse over the screen causes the sound to play forwards or backwards at varying speed, similar to a jog wheel on a video or audio tape recorder. When you've found the points you want to edit at, you go to a 'selection' mode and click on the beginning and end of the region, thereby selecting it.
Among the destructive functions are the standard Mac editing operations like Cut, Copy, Paste (which moves old data aside to make room for new), Replace (which writes over old data), and Clear, as well as audio-specific functions such as reverse, silence, trim (eliminate everything outside the region), fade in and fade out, change the gain up or down, normalise (that is, increase the overall amplitude of the region so that its loudest point is at the maximum recordable level), and phase invert.
There are also a host of 'DSP' functions, including a parametric equaliser, which is configurable as peaking, shelving, high-pass or low-pass filter, and can cut or boost up to 24dB over a bandwidth as small as 10Hz. The window that opens when you select this function has a 'Preview' button, which lets you listen to the effect of the equaliser on the selected region in real time as you adjust it. Once you arrive at the setting you want, click on 'Process' and the region is recalculated with the new equalisation. Although the equaliser only acts on a single frequency band at a time, you can use it on the same piece of sound as many times as you want. A volume slider is included to keep the level from getting too high or too low as the equaliser modifies it.
If you design an EQ setting you particularly like, you can save it as part of the soundfile and recall it at any time. In fact, an EQ setting can be saved as part of the soundfile without actually having been used to process the file - and it can then be imposed on it the next time the file is played back. This lets you take advantage of the equaliser in a non-destructive way.
There is also a seven-band graphic equaliser function, with adjustable centre frequencies and bandwidths for each band, and with the same preview and storage features.
Another DSP function is Merging. This allows you to splice two soundfiles together, with a programmable crossfade time that can be several seconds long (it depends on how much RAM is available). Yet another is Mixing: up to four soundfiles (mono or stereo) can be combined into one, with adjustable level, pan position, and starting delay (in milliseconds) for each of the original files. Both of these functions are Undo-able, and both automatically create a new soundfile, thereby preserving the original components.
Finally, the system allows soundfiles to be stretched or shrunk in time without changing their apparent pitch. This means, for example, that a line of dialogue can be sped up to fit a scene that has been shortened, or a tyre squeal from a sound effects library can be drawn out to fill a long shot of a car careering down a mountain road. The software shows you the length of a chosen segment and lets you specify either a new length in seconds or a ratio of expansion or compression. The current version of the software will only perform this function on mono files, and Digidesign says that the algorithm in use is optimised only for speech, but it doesn't sound too bad on music, and algorithms specifically designed for music will probably be available in a software update before long. Like the Mix and Merge functions, the time-changing function automatically creates a new soundfile, leaving the original intact.
Destructive editing of sound has one major drawback: if you want to keep your original versions (or various intermediate ones) around, you're going to find yourself using up an awful lot of hard disk space. Non-destructive editing allows you to maintain just one copy of the soundfile, along with several different sets of parameters for dealing with it, which take up much less space - while soundfiles take up tens of megabytes, parameter sets only take up a couple of kilobytes. Non-destructive editing, because it involves moving far less data around, is also considerably faster.
For example, one way of manipulating the various parts of a file so that they play at different times and in a different order (for example, assembling a piece of dance music out of several short, repeating components) would be to Cut and Paste the sections. But this is destructive. Sound Tools provides an alternative: select the various sections, give each one a name, and then set up a 'Playlist' for playing them back in any order you like.
As with the editing functions, a section is selected and auditioned from the graphic screen, and is called a 'region'. Regions can overlap, and one region can even be a subset of another. After you've defined and named all of the regions you want (and there is no limit to how many you can have), you open a Playlist window, and all the names appear. You then construct the playlist by dragging the name of each region down to a lower window in the order you want them to play. The software automatically calculates and displays the start time, length, and stop time of each region as you assemble the list. Re-ordering the list is merely a question of dragging the names around until they're where you want them. You can then hear your entire list by clicking on the 'Play' icon. You can also listen to the playlist starting in the middle, or just a single region.
Initially, when you place two regions together, they are 'butt-spliced', but if the transition is too abrupt you can specify a crossfade of any length from one millisecond up to several seconds. The crossfade can either be linear or 'equal power', which will sound smoother with some material, and it can be set to occur during the transition between the two regions (ie. it affects both regions equally) or before the transition (ie. it finishes before the second region starts useful when the second region starts with a percussive downbeat). Each entry on the playlist can have its own crossfade, so you can mix up long and short ones to your heart's desire, and even put a nice long fade on the end of the last region (if you have enough RAM).
The playlist can be set to play back locked to incoming SMPTE (actually MIDI) timecode. You can specify a SMPTE start time for the first region, or alternatively you can assign a specific time to any region on the list, and the software will calculate the start time for the whole list. Any number of playlists, each with its own name, can be associated with a soundfile. This means that with a given amount of raw material (a soundfile), you can create an enormous variety of music or sound, using only the amount of disk space it takes to store a few lists of parameters.
Fooling around with sounds on a hard disk is only part of what the Sound Tools system is capable of. Like earlier versions of Sound Designer, the Sound Tools package is a very powerful sample editor and manipulator. Any sound recorded with the system can be transferred to another sampler (as long as the sampler has enough memory for it) via MIDI or, if the sampler supports it, the faster RS422 or SCSI protocols. MIDI System Exclusive dumps are supported for most major samplers, and the MMA Sample Dump Standard (in both 16-bit and 12-bit versions) is also available.
Of course, samples can also be loaded from these hardware samplers into Sound Tools for editing. With the Sound Accelerator card, any modifications you make to a sample can be auditioned immediately, without having to wait for it to be transferred back to the sampler. The AD IN box allows you to record new samples for samplers that have no recording capabilities of their own, like the Oberheim DPX1.
An important part of sample editing is designing loops, and the software has some very advanced capabilities in this regard that make finding good splice points absurdly easy. An unlimited number of loops can be specified, with no restrictions on length or overlap. Loop points are edited in a special Loop window, which allows very fine visually-aided adjustment of the splice point, and an 'Auto' function is available that automatically finds points in the waveform for splicing that match in both value and slope. As a last resort, if none of these methods work, a crossfade function is available. Again, the Sound Accelerator card allows immediate auditioning of the edited loop.
The software will also handle sample rate conversions, so if you need to output a soundfile to a sampler with a fixed playback rate (for example, the AKG ADR 68K only plays samples at 32kHz), you can do so easily. If you want to hear the sample polyphonically before you output it, you can turn on a 'MIDI Preview' function, which allows you to play up to eight voices of the sample (with velocity sensing!) from an external MIDI keyboard. (You can also play soundfiles from a keyboard without having to boot the Sound Designer software using a small utility program provided with the system called, not surprisingly, 'MIDI Preview'.)
Files can also be saved on disk as 'snd' or 'AIFF' files. 'Snd' files are 'resources' that allow 8-bit digitised sounds to be inserted into various other Macintosh applications and played back by those applications, using the Mac's internal audio hardware and speaker. AIFF, which stands for Audio Interchange File Format, is a more complex (16-bit, multichannel) format designed by Apple for exchanging sounds among applications. In addition, a utility program is provided for installing a 'snth' resource into the Macintosh system software, which will allow playback of Sound Designer files from within HyperCard (and probably other programs in the future), using either the Mac's electronics or the Sound Accelerator card.
Also provided as part of the package is Digidesign's Softsynth software [reviewed SOS July 89], a wonderful program that lets you design sounds from scratch using additive and FM synthesis, and either play them directly with the Sound Accelerator card or transfer them to an external sampler.
Finally, the software provides an impressive three-dimensional plotting function, showing the sound graphically as amplitude versus frequency versus time. A variety of scales, perspectives, densities, and directions are available. I'm not sure how much use this all is, but it is a lot of fun to look at.
No doubt about it, Sound Tools is brilliant. It makes great use of the Macintosh, both its user interface and its computing power, to provide a system that is easy to use, eminently practical in a wide variety of situations, and sounds terrific. But that's not to say there is no room for improvement.
In some ways, the software is a little clumsy. The relationship between the various windows, and how the tools behave in them, is fuzzy and occasionally seems inconsistent. Finding and selecting editing points requires more mouse movements and a better visual memory than it should: when you've found an edit point with the scrub function, you have to remember where it is by eye, get the selection tool, and select it. Having an Option key combination to automatically invoke the selection cursor while you're using the scrub function would help a lot. Echo loop editing is also a little slow, and the procedure for listening to loops could be streamlined.
Individual items on a playlist cannot be adjusted slightly in time without going back to the editing screen and redefining the region's length, and then saving it under a new name and re-inserting it into the playlist. Placing silence into a playlist can only be done by selecting a 'silent' (all zeroes) region, adjusting its length carefully, and then inserting it.
Some operations can be quite slow (for example, changing the sample rate on a 25-second mono recording took about 6½ minutes on a Mac IIcx), which in itself is understandable (that's a lot of data to crunch), and the software is kind enough to give you a running progress indicator so you know when it's time to make a cup of tea, but it would be better if it warned you beforehand how long the operation is going to take and/or give you a chance to change your mind.
The Sound Tools hardware is solid and essentially idiot-proof, but the fact that it has only unbalanced inputs and outputs gives it a less-than-professional look and feel. If the system is part of a complex studio setup, and you're not careful about signal routing, the AD IN-to-Sound Accelerator echoing function can cause the system to go into instant digital feedback. This wouldn't be much of a problem, except that the tape recorder window in the software defaults with this function turned on, so it can catch you by surprise.
Lastly, although the software is very robust, and it has an admirable habit of creating backup files at the drop of a hat to ensure you don't lose too much if it does crash, I did find a couple of bugs. Nothing too serious, and Digidesign assures me they're all going to be fixed forthwith, but annoying nonetheless.
But the most serious problem with Sound Tools has nothing to do with Digidesign. It's a problem that everyone doing hard disk audio has to face: storage. Digital sounds take up a lot of room - that 25-second mono recording I mentioned above takes up 2.15 megabytes at 44.1 kHz. In stereo, or at 48kHz, it would take up even more. Having a 500 megabyte hard disk is all well and good for dealing with a single project, but if you want to start a new project, that hard disk is no better than a tape recorder that you can't take the reels off. So what do you do with the sound when you're done playing with it?
In the traditional studio, analogue tape is the storage medium of choice, both for the short term and the long term - it serves just as well for this week's sessions as for last year's masters. But one medium might not be enough to solve both the immediate and long term storage needs of the digital studio.
One possible solution is to download every soundfile onto digital audio tape (DAT). This method certainly preserves the audio quality, and the transfer can be done more-or-less in real time. It will become an even more attractive option when Digidesign finishes the DAT I0 (pronounced, and I'm not responsible for this, 'daddy-o') system later this year, which will allow the hardware to interface directly with equipment using the AES/EBU Consumer or S/P DIF digital formats. However, any file stored this way would lose all of its editing information: EQ files, playlists, loop points.
Recordable optical disks might be an answer, once they become available, assuming the cost per megabyte is low enough, although write-once disks would only be appropriate for archival storage. With removable hard disk cartridges, data transfer takes no time at all (the cartridge is the hard disk), so they're ideal for active storage, but they are currently very expensive. Keep in mind that analogue 2-track audio tape costs in the neighbourhood of 45 pence per minute (depending on the speed), and any medium that is substantially more expensive than that will not gain much acceptance.
At the present time, the major disadvantages that Sound Tools has when compared with a high-end multitrack digital workstation is that: (a) it only allows two channels of audio, either as 'samples' or as 'tracks' (the Synclavier treats them separately); and (b) it is self-contained and cannot communicate with the outside world, except in terms of syncing itself to incoming timecode.
Its major advantages are: (a) its ease of use; (b) its cost - a fully configured system, including computer, is of the order of one-tenth of what you'd pay for a similarly-powered Synclavier; (c) when you want to upgrade it, you aren't locked into expensive hardware modules only available from the manufacturer - the system will be happy to work with just about any off-the-shelf RAM or storage medium you want, as long as it's compatible with the Macintosh. (Be careful with hard disks, however - some of them are too slow to access the data properly. Digidesign provides a list of which common disks are suitable and which are not.)
The two-channel issue may not be solved soon. You can install more than one Sound Accelerator card into a Mac II chassis, but since the software already is using most of the available CPU speed to get the audio from the disk directly to just one card, there is no way that it can address more than one. This may change when Apple releases its next line of Macintosh IIs, which promise faster processors. In the meantime, it is theoretically possible to have two cards making sounds, but only one of them can be running off the Sound Designer software, while the other would have to be driven by some other program. Getting the system to run at the same time as a sequencer, event-list editor, or other MIDI program will become a reality when Apple releases its MIDI Manager, a system software module in development for some time now that will allow multiple MIDI applications to be run simultaneously under MultiFinder, locked to a common clock like MIDI Time Code.
But Digidesign is not waiting for that to happen (or for other developers to make their software MIDI Manager-compatible) before they expand Sound Tools' real-world capabilities. The company plans to release soon a new version of its Q-Sheet event-list editor that will allow a Sound Tools playlist to run in the background. Since Q-Sheet includes sequencer-like functions as well as MIDI File playback capability, this will allow the Mac to handle sequence playback, sample triggering (including, of course, effects created or edited in Sound Designer and then off-loaded into hardware samplers), automated mixing, processor automation, and two-channel digital audio playback - all at the same time, and all linked to external timecode.
Another major advantage of running a system like this on a widely available computer is that third-party developers will be encouraged to write software that will use it. Once the hardware is in place, there is an awful lot of digital signal processing that can be done in software. Sound Designer is only the beginning - I can envision delay and reverb modules, noise and impulse filters, compressors and expanders, and processing that no-one has thought of yet, working both on recorded files and in real time.
Sound Tools is not a sampler, and it will not replace a good sampler in a MIDI studio. Nor is it really a master recorder, unless you have access to huge amounts of storage. It is very much what its name says: a bunch of highly useful and innovative tools. If you work extensively with samplers, with sound effects, with dialogue, or with music for film, you will find Sound Tools has many uses and will solve many otherwise-unsolvable problems. Even if you're strictly in the record business, its ability to tweak and shape sound with perfect fidelity and exquisite precision will prove invaluable. Its designers have done a superb job of making it understandable and easy to learn, and to make it work well. Considering what it can do already, and even more what it will do in the future, Sound Tools is a wonderful investment for anyone serious about music production and with an interest in owning a Macintosh.
Paul D. Lehrman is a composer, author, and consultant based in Boston, USA. He wrote a sample editing program six years ago that sold for $80. It wasn't as good as this one.
The Sound Tools system, which includes the Sound Accelerator card, AD IN, and software, depending on configuration, costs around £3140 inc VAT.
Sound Technology plc, (Contact Details).
Gear in this article:
Review by Paul D. Lehrman
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!