Unlike text, sound is generated outside the computer. Because sound
is w
really caused by pressure differences in the air, it can be converted
to
voltage levels.
This means that the sound wave (think of an ocean wave) can be converted
to a sequence of bits. The sound wave is
an analog signal (continuous, like the ocean wave) and this signal is sampled
(measured) periodically with a 0 or 1 stored depending on the magnitude of
the signal at the times it is measured.
This conversion is done by audio cards. Less expensive 8-bit cards sample the
wave less often than more expensive 16-bit cards. These 8-bit cards are
considered adequate for speech of the
quality on a telephone or answering machine, while 16-bits is best for
CD quality.
These cards also convert the sound back from its digital bit representation to the analog signal recognized by speakers.
For 8-bit cards, 256 values (00000000, 00000001, etc.) of the sound wave can
be stored. The signal is (usually) sampled 8000 times a second (called 8 kilo Hertz -
8 kHz). Since 8 bits are stored at the rate of 8000 times a second, a sequence
of 64,000 bits per second are transmitted.
In general, you can play back less than half the range that is sampled.
For CD-quality music, the wave is sampled 44,100 times a second (44.1 kHz)
and because 16 bits are used to code each sample, a sequence of
16 * 44,100 (705, 600) bits per second are generated.
Other sampling rates and varying storage sizes store signals with values
between these two extremes. For example, AM quality broadcasts are stored
in 4 bits with a sampling frequency of 18.9 kHz.
Improvements are made by not storing the entire wave during silences or by
coding differently. For example, the difference in 2 sampled values
can be
stored rather than the actual values. This takes less space because
the wave may not change much between samples.
This encoding of a signal by not storing every value is called
compression.
The Motion Picture Experts Group has developed standards for representing
combined audio-visual communication. For audio,
MPEG uses
psychacoustic
principles , that is, it takes into account how the brain perceives
sound quality, and what frequency ranges ears are most sensitive to.
Because people wanted to hitch up musical instruments to a computer, a
standard for encoding and decoding such music has been developed called
MIDI - Musical Instrument Digital Interface.
Prof Ed Fox on Audio
Send questions and comments to: Karen Lemone