Digital Sound Synthesis I

How do we hear ?.

Sound sources (voices, noises, singing, etc.) set in motion air waves producing changes in the air pressure: air molecules are brought closer together and pulled apart. They hit our ear drums whose vibrations are transmitted through the Middle Ear to the Cochlea the internal ear. There we find the basilar membrane on which little hair cells are connected to nerve endings. Different regions of the Basilar membrane respond to different frequencies and the response is less pronounced toward its end. That explains the Fletcher-Munson curves derived experimentally in the 1930s.

Digital recording and synthesis.

A microphone transforms the air pressure wave into changes in an electric current which is transmitted to an Analog-to-Digital converter (ADC). The ADC samples ie. takes measurements of the wave at a steady rate producing a series of numbers. The CD standard is 44,100 samples/sec. which is a little higher than the double of the limit of human hearing (20,000 Hz). In order to catch all the variations in the analogue wave, the sampling rate needs to be at least twice the frequency of the sound we want to hear (Nyquist frequency). These numbers are stored in a computer and printed on the CD.

When a CD is played back, a laser beam reads such numbers and sends them to a Digital-to-Analog converter (DAC). There they are transformed back into an analogue wave of changes in the electric current and send to a speaker. The speaker cones vibrate setting in motion a air pressure wave that we can hear.

First attempts to synthesize sounds digitally, "from scratch" took place at the Bell Labs in Murray Hill, NJ. in 1956, about the same time Lejaren Hiller was performing his experiments leading to the creation of the Illiac Suite. Max Mathews developed a computer program MUSIC I followed by an entire family of such programs, part of the MUSIC N paradigm. Influenced by the 19th. century orchestra scheme, MUSIC N involves an orchestra of digital instruments, a score made out of I cards and a "conductor", MUSIC N itself. The result is a sequence of numbers (44,100 for one second of sound).


[ back to Music 408E | back to Lecture notes | to Reserve list | to Syllabus | back to Courses | back to Sever Tipei's home page | Computer Music Project ]