Digital Signal Processing Reference
In-Depth Information
The compression rates are determined above all by the setting of the quantisation
thresholds in each partial band. These are determined by the instantaneous masking
characteristics of the 32 channels, which determine which tones (frequencies) must be
transmitted. A digital PCM audio signal of CD quality and for mono requires 768 kbit/s.
By means of the MPEG audio encoding the signal can be compressed to under 100 kbit/s.
The MP3 process (or more precisely: MPEG 1 Layer 3) is particularly popular for audio
compression. By means of numerous technical refinements - see Illustration 249 - the
compression rate can be increased to less than 10% of the PCM audio signal. Hence, at
least 10 normal audio CDs go on to one MP3 CD.
Encoding and physics
Encoding is described here as one of the most important possibilities of modern commu-
nications technology. It is striking that with the exception of the compression of audio
data - the fundamental physical phenomena of the first chapters - FOURIER, Uncertainty
and Symmetry Principle - seem to have little to do with this. They were scarcely used in
this chapter.
This has to do with the historical development of communications theory. All the physi-
cally based information theories associated with names such as HARTLEY, GABOR,
CLAVIER, and KÜPFMÜLLER were overshadowed by the grandiose success of the in-
formation theory of Claude SHANNON (1948) which was based purely on mathematics
(statistics and the calculus of probability). This theory made modern applications of signal
processing such as satellite and mobile telephony possible.
It is still difficult to get a clear grip on the concept of information from a physical point
of view. There is still a yawning gap between SHANNON's theory and physics, to which
all technology is subject.
SHANNON's theory will be examined in more detail at the end of the next chapter.
