Rossum (Continued from p.2)

Rossum started out his collegiate career as a biology student, but switched to studying electronics after he encountered his first Moog synthesizer. Soon after graduating, he started E-mu Systems in 1972 with other likeminded engineers interested in creating electronic equipment for musicians. Their first product, the E-mu Modular System, operated like a telephone switchboard so as to patch differ- ent oscillators and processing modules together. This system con- tained no sampling technology, and was mostly utilized by universi- ties.

E-mu discovered two unrelated products at the 1980 AES conven- tion which sparked their imaginations: first was the Fairlight CMI 6502 which was designed for reproducing an arbitrary, hand-drawn waveform (sampling, as in feeding in a signal with a microphone for reproduction later, was an afterthought), second was the Publison Digital Echo Unit, which could capture two different sounds in memory. They conceived of a keyboard product that could incorpo- rate these technologies so as to reproduce stored samples at differ- ent pitches. In July of 1981, E-mu started shipping their Emulator product. At an initial cost of $10,000, it included a floppy disc drive mounted in a steel frame case. Samples could be loaded, played, and basic manipulations could be performed so as to tailor the sound output. In January of 1982, the price was lowered to $8000, with sequencer capability and a sound library added. One other change, thought minor at the time, was to make samples end when the key was released. This small functional change was welcomed by musi- cians, who liked the added ability to control the sound output.

Sampling involves breaking an analog signal down into discrete com- ponents, or samples, each of which can be stored as a numeric word of a certain bit length. This stored data can either be manipulated in the time domain (multiplying the samples by a train of periodic impulses) or in the frequency domain (convolving the samples with a train of periodic impulses). The sampling rate governs the highest- frequency that can be accurately represented in the data, being half that of the sampling rate (the Nyquist frequency). So if you are tak-

(Continues p.4)