Digital Sound & Music: Concepts, Applications, & Science, Chapter 5, last updated 6/25/2013
59
phenomenon called masking can occur. Masking results when two frequencies are received by a
critical band at about the same moment in time, one of the frequencies being significantly louder
than the first such that it makes the first inaudible. The loud frequency is called the masking
tone, and the quiet one is the masked frequency. Masking causes the threshold of hearing to be
raised within a critical band in the presence of a masking tone. The new threshold of hearing is
called the masking threshold, as depicted in Figure 5.46. Note that this graph represents the
masking tone in one critical band in a narrow window of time. Audio compression algorithms
look for places where masking occurs in all of the critical bands over each consecutive window
of time for the duration of the audio signal.
Figure 5.46 Threshold of hearing altered in presence of masking tone
Another type of masking, called temporal masking, occurs with regard to transients.
Transients are sudden bursts of sounds like drum beats, hand claps, finger snaps, and consonant
sounds. When two transients occur close together in time, the louder one can mask the weaker
one. Identifying instances of temporal masking is another important part of perceptual encoding.
5.3.8.3 MP3 and AAC Compression
Two of the best known compression methods that
use perceptual encoding are MP3 and AAC. MP3
is actually a shortened name for MPEG-1 Audio
Layer III. It was developed through a
collaboration of scientists at Fraunhofer IIS,
AT&T Bell Labs, Thomason-Brandt, CCETT,
and others and was approved by ISO/IEC as part
of the MPEG standard in 1991. MP3 was a
landmark compression method in the way that it
popularized the exchange of music by means of the web. Although MP3 is a lossy compression
Aside: MPEG is an acronym for Motion
Picture Experts Group and refers to a family
of compression methods for digital audio
and video. MPEG standards were developed
in phases and layers beginning in 1988, as
outlined in Table 5.3. The phases are
indicated by Arabic numerals and the layers
by Roman numerals.
Previous Page Next Page