Michael Pounds
I am just finishing a yet untitled composition that began in Studio A using the Buchla and Roland analog synthesizers as my only sources of sound. I chose to do this partly as an exercise in the fundamentals of synthesis, and partly to take advantage of some historical equipment while it still lingers around. I did it just in time, too: by the fall of 1997 the newly renovated Studio A will reopen with no analog synthesizers. The course of my composition that began with analog equipment in Studio A passed through some digital processing, and ended with some intensive digital editing in Studio D.
ANALOG
I have a love/hate relationship with analog synthesizers. I find something aesthetically pleasing about their simplicity, their sound (the same sound that is at a crest of a wave of retro-mania in the commercial music world), and just plugging in all those cables and twisting those knobs. On the other hand, I spend much of my effort trying to make sounds that do not actually sound like stereotypical analog synthesizer sounds.
The initial premise for my composition involved creating increasingly dense layers of sound that eventually attain the identity of a solid mass of sound rather than discreet events. To realize this, a great number of source sounds were required. At first I found that the sounds that were least identifiable as typical analog sounds had rather short envelopes and often were percussive in nature. Most of the percussive sounds employed a combination of additive synthesis and frequency modulation, and occasionally some noise thrown into the mix. Some of the less percussive sounds involved heavy amounts of frequency modulation, and these were the ones that tended to sound the most trite. I found that these sounds were most convincing when their durations were kept fairly short.
With all of these short sounds, I quickly realized that it would be difficult to attain a high degree of density without recording hundreds of sounds. One solution was to create gestures of many short sounds. Analog modular synthesizers are well equipped for this, with voltage-controllable sequencer clocks that can trigger individual envelopes. I used envelopes to control the rate of occurrence of events, but even then the resulting gestures can sound predictable. To solve this problem, I employed my favorite Buchla module, the random signal generator. This wonderful module can add spice to just about any sound or gesture. Not only did I use it to make the time between events less predictable, but I also used it in conjunction with a patch created on the Roland that automatically pans sound in the stereo field according to a control voltage. The gestures were then able to move spatially in a random way.
Another problem that resulted from the short durations of the sounds was that the composition lacked continuity. It became necessary to create some longer sustaining sounds. Many of the longer sound used additive synthesis techniques to achieve a somewhat vocal character. To make them sound more natural, two tricks were employed. The first was, of course, the Buchla random signal generator. If more than one random signal, perhaps with the addition of some LFOs, are used to just slightly modulate the frequency of multiple oscillators so that each oscillator is affected in a different way, the result is a much thicker, more natural sound. This can also work if all of the oscillators are tuned to approximately the same frequency. One of my sounds uses this technique in combination with some slow envelopes affecting the pitch of each oscillator in a slightly different way. Another sound uses several oscillators tuned to the harmonic series, with the random signals and LFOs affecting the amplitude of each. The result is a shimmering sound with the harmonic content continuously changing.
The other trick to make the sounds more natural was to add liberal quantities of reverb. It can help sustained sounds that change even slightly over time to sound much more full and complex. In at least one case, the final sound included only the output of the effects processor, with none of the ³dry² signal present. This was especially useful for a certain patch that combined low frequencies sine waves with noise-modulated sine waves. The sound was heavily filtered, but still the resulting sound had a very unnatural and synthetic quality. Next, the signal processor was used. When the original signal was eliminated from the mix, leaving only the ³wet ³ signal from a long reverb, the somewhat embarrassing slurpy sounds from the synthesizer were magically transformed into menacing snarls which mixed well with the low-frequency roar underneath. To add further interest to the sound, I manually adjusted the filter parameters in real time as I was recording the sound.
All of the techniques described above were specifically used for a synthesis system that will no longer be available by the time this article is published. However, most of them can be applied using a good digital synthesizer, like the Kurzweil K2000/2500, or to a lesser degree on older systems like the Yamaha FM synthesizers (although I will miss that Buchla random signal generator).
DIGITAL
To realize the dense layers of sound that were planned for the piece, the Pro Tools system in Studio D became an invaluable tool. All of the source sounds were recorded onto digital audio tape (DAT). In addition, much of the first section of the composition was constructed directly on an Alesis ADAT eight-track digital recorder. This was a laborious process that involved recording each individual sound on a specific track (or on a stereo pair of tracks). When the timing of a single event needed to be adjusted the only solution was to re-record the sound using a ³punch-in² technique that required great care in order not to record over other material on the tape. In addition, the only way to adjust the relative amplitude of the sounds was to manually move the fader for each track during the mix. If many such adjustments are to be made, the mix-down process quickly becomes a very complicated job that requires a lot of rehearsal.
While all of this has historically been standard procedure for multitrack recording systems, digital audio workstations like the Digidesign Pro Tools system in Studio D change everything. When I began to work on the composition in Studio D, I first digitally transferred all of the tracks from my ADAT tape directly to the Pro Tools system, which stores the tracks on a hard drive. Since the sounds never left the digital domain, there was no deterioration of the signal during the transfer. Working in Pro Tools, I was not only able to completely automate the mix of the tracks, but I was able to cut and paste each sound to adjust the timing, or even move sounds between tracks. I was also able to further shape the amplitude envelopes of each sound. One limitation of the Buchla synthesizer is that the envelopes are very simple and basic. In Pro Tools I was able to specify amplitude changes during each sound or gesture that were as complex as I wanted. These envelopes were automated and were very easily edited.
Once I developed a mix that I liked, I ³bounced² the tracks down digitally to two tracks on an ADAT tape. Again, there was no loss due to digital-to-analog conversion. Now I was able to create eight more tracks in Pro Tools and mix those to two more tracks on the ADAT recorder. Our system allows the Pro Tools software and the ADAT recorder to play in perfect synchronization, so there was never a problem getting the later tracks to align with the earlier tracks. If this mixing and layering process is performed four times, the resulting ADAT tape will contain four stereo mixes from 32 original tracks, with all of the mixing and transferring remaining in the digital domain. With this system it was very easy to create the dense layers of sound that I wanted for my composition.
Pro Tools was also indispensable in the creation of individual sound ³events² that were actually masses of individual sounds that were layered and mixed together. This transformation from individual sounds to masses of sound was an important part of my original compositional premise. I was able to accomplish this fairly efficiently in Pro Tools. First I created several Sound Designer files that each contained many of my original analog sounds. Then I imported these files into Pro Tools, where they each became separate tracks (Pro Tools allows a great many tracks, but only plays back eight at once on our system). I was able to quickly separate each track into individual, short ³regions² of sound. Pro Tools allows the user to easily construct new tracks from these regions by simply dragging the regions from a list onto the graphic representation of the track. These new tracks do not actually take up any disk space, because Pro Tools always plays back from the original sound files on the disk. I was able to quickly create eight layers of these regions, and I then mixed them all to one monophonic file directly on the hard disk. This monophonic file became a new track, which I was able to slice up into a number of new regions. Each of these new regions became a separate sound mass event that I used in my composition. When these sound mass events were in turn layered together along with my longer sustained sounds and mixed to ADAT, I was able to achieve some extremely dense layers of sound.
Looking back at the course of this composition, I see that I have enjoyed and learned much from both the older analog equipment and the newer technology. In working with the analog synthesis, I became more skilled in basic concepts of sound synthesis, developed an appreciation for the historical composition work done using these synthesizers, and had some fun twisting knobs. I have also developed a real respect for the power of the newer technology. I have become aware of some of the capabilities of the tools currently available, and I am more skilled in their use. I am looking forward to my next composition using the modern technology and our newly renovated studios.