Electronic Music Wiki

(1926-2011) Noted music and audio researcher, who spent most of his professional career at Bell Laboratories. Mathews studied electrical engineering at both the California Institute of Technology and the Massachusetts Institute of Technology, receiving a Doctor of Science degree from the latter in 1954. Mathews then went to work for Bell Labs at its Murray Hill, New Jersey location, where the Labs were engaged in several research programs concerning the electronic generation of speech and sound. As an amateur violinist, he was keen to launch a project to see if comprehensible music could be generated by computer. He wrote a "programming language" for music generation called MUSIC (now retroactively labeled MUSIC-I), which first ran in 1957. This made it probably the second system anywhere in the world to produce computer-generated music, after the CSIRAC project in Australia a few years prior. MUSIC-I was monophonic and was capable of generating only one waveform, a triangle wave. Nonetheless, it was the first of its type. To actually run the program, Mathews had to take code to IBM's headquarters in New York City where he could run the program on an IBM 704 mainframe computer. This produced the digital output on magnetic tape, which Mathews then took back to Murray Hill and ran on a machine which read the tape and passed the digital data through a D/A converter so that it could be heard. By Mathews' telling, his first composition with the program was 17 seconds long, and took all night to compute on the 704.

Murray Hill soon acquired its own IBM 7094, and Mathews set about improving MUSIC. The next version MUSIC-II, was polyphonic and capable of producing up to 16 voices, and had several choices of waveforms. MUSIC-III, first fielded in 1963, went in a different direction. It allowed the code to manipulate the digital music at the word level. Waveforms could be pre-computed and stored in internal wavetables, so the choice of waveform was now unlimited. And a concept called the "unit instrument" allowed the code to divided into blocks of individual "instrument" code, which could be invoked any number of times, so polyphony and the length of compositions was now only limited by the composer's patience (or his/her budget for CPU time). MUSIC-IV contained a number of optimizations for taking advantage of vector processing on computers having the necessary hardware, with large time savings in computing.

Mathews' final version of MUSIC, MUSIC-V, was motivated by the impending retirement of the now 10-year-old 7094 in the late 1960s. Most of the program was re-written in FORTRAN, which made it possible (not easy, but possible) to port the program to other computers. From here, other parties took the code and branched out to write a number of variants; included among these was an effort at MIT to re-write the entire program in the C programming language, with the result becoming Csound. At this point, Mathews turned his attention to other projects, including a system for controlling an analog synth to execute a score in real time as directed by the performer on the fly, called GROOVE. Additionally, Mathews began developing a variety of electric violins that were coupled with digital processing to vastly expand the number of timbres available, using some early concepts in resynthesis Mathews also did work on a "front end" for MUSIC which would allow a score to be composed by drawing shapes to control aspects of the notes to be played.

Mathews left Bell Labs in 1974 and moved to IRCAM in France, where he became a scientific advisor to the institution, returning to the U.S. in 1980. He then became a professor of electronic music at Stanford University, a position he held until his death.