The Toccata and Fugue was realized in 1970 using a music software system called NOTRAN. NOTRAN, which is a derivative of the words NOte TRANslation, was developed as a semester project for a music appreciation course I took while an undergraduate student at North Carolina State University. The computer utilized was an Ambilog 200 which is a rather obscure 30 bit word minicomputer equipped with a 14 bit digital-to-analog converter and two IBM style tape drives. Except for the DAC, there was no other sound generation hardware.
Unlike other systems represented on this record, the NOTRAN system did not produce sound in real-time. Thus the program “ran” for a much longer time than the duration of the piece. During this time, about 3.5 hours, a series of over 15 million 12-bit numbers was computed. Each number represented the composite sound waveform for a period of about 30 microseconds which means that 32,000 such numbers are needed for each second of sound. As the numbers were computed, they were written onto the tape drive. When the piece was complete, a different “playback” program read the tape at the rate of 8000 numbers per second and sent them out through the digital-to-analog converter. The resulting audio signal, which was still too slow by a factor of 4, was recorded at 15 inches per second on a laboratory tape recorder. Finally the tape was replayed at 60 inches per second and recorded on a standard audio tape recorder.
The primary advantage of such a system is that the sound complexity and detail possible is virtually unlimited since the synthesis program is allowed to run at its own rate. The primary disadvantage, of course, is a substantial waiting period before results are heard. Although many thousands of dollars worth of university owned equipment was used in realizing this piece, any hobbyist system with dual floppy disks (preferably double density) and a digital-to-analog coverter could do the same although somewhat more slowly (the Ambilog had hardware multiply/divide).
Several features of the system should be apparent while listening to the piece. Since a pipe organ was being simulated, different voicing was used for the great, swell, and pedal keyboard parts. Each part in turn could have several notes sounding simultaneously. In particular, when the decay of one note overlaps the attack of the next, both will be sounding simultaneously for a time which is a requirement for realism. The voices themselves were an ad-hoc mixture of sawtooth, triangle, rectangle, and sine waves at 1, 2, and 4 times the fundamental pitch. In addition the attack, decay, and relative loudness of each voice could be specified. Near the middle of the piece the voicing is changed several times. Even the reverberation is computed by the NOTRAN system by simulating a tapped audio delay line in software.