CHARLES DODGE – SPEECH SONGS (1973)

NOTE BY CHARLED DODGE

The works on this disk are called computer music because the sounds were created numerically in a computer before reaching magnetic tape and loudspeakers. The computer did not composer the music, however. It was only used as a sound synthesis and analysis medium. In order to hear the sounds which are represented in the computer as successions of numbers it was necessary to attach a digital-to-analogue converter to the computer. The converter transformed the succession of numbers into a fluctuating voltage which was recorded directly onto audio tape.

The technique used to create the synthetic voices is called speech synthesis-by-analysis. For this method of speech synthesis, only those words, phrases and sentences which have been spoken into the computer previously (via an analogue-to-digital converter) may be synthesized. First the digitally-recorded speech is analyzed by programs which extract the attributes (parameters) of the speech for short time segments (.01 sec.). From these parameters the speech may be recreated in a form which resembles the original recording very closely. But for musical purposes the parameters are most often altered before synthesis.

It is possible, for example, to change the natural pitch contour of the speech into a melodic line or to change the speech speed without altering the natural pitch level. The variety of musical patters which may be created from the analytic parameters is limited only by the composer’s imagination.

. . . The system employed to realize SPEECH SONGS was created at the Bell Telephone Laboratories for research in synthetic speech and speech communication. Since the system was not designed for musical purposes, its limitations were severe and its musical use a great (and often rewarding) challenge.

(from CRI SD 348 liner notes, 1976)