Digital Signal Processing Reference
In-Depth Information
How speech, tones and sounds are generated and perceived
There are two reasons why I would like to make a short excursion into human anatomy
and, in the same context, into the physics of waves and oscillations:
• Nature is the greatest preceptor of natural sciences such as communications enginee-
ring. In the course of evolution nature has developed mechanisms of perception which
have enabled the various different species to survive. Human sense organs are highly
sensitive and are precise sensors which far surpass technological imitations. A good
example is the human ear: even the weakest acoustic signals which are just slightly
above the noise level caused by the bombardment of the eardrum with air molecules
are perceived.
• These physiological correlations are the basis of acoustics and of any form of spoken
communication.
Engineers for some reason often shy away from physics and above all from biophysics
although engineering is nothing more than the sensible and responsible application of the
laws of natural sciences. The theorists in natural sciences, however, have so far failed to
produce a mathematical model of sensory perception such as hearing, and anything which
cannot be defined mathematically is widely ignored in textbooks on the theory of commu-
nications engineering.
My objective is to demonstrate how sounds and language are generated in the speech
organs and how the acoustic perception in the human ear works. As a highlight we are
going to develop a case study of a simple computer-aided voice recognition system.
We will have to content ourselves with a very simple model of the human ear because the
acoustic perception and above all the processing of signals by the human brain are so
complex - as I have already pointed out - that so far relatively little is known about these
processes.
In any case, it has all got to do with the physics of oscillation and waves of which a little
knowledge is necessary to be able to understand these processes. Please remember: there
are no processes in nature or in engineering which would contradict the laws of natural
science - e.g. the laws of physics. So let us begin with these laws.
If we want to do research in the field of the generation of tones, sounds and speech we will
have to ask ourselves what is required to generate them. What we basically need are
oscillators and cavity resonators as well as energy to trigger the process.
Definition: a mechanical oscillator is a vibrating system which - once it has been
given a mechanical pulse - oscillates in its characteristic natural frequency. The
oscillation can be maintained if the relevant energy is supplied. The oscillation can
be particularly effectively maintained if the spectrum of the energy supplied
contains in its signal the natural frequency of the oscillator. An example of a
mechanical oscillator is a tuning fork, another example is the reed in the mouthpiece
of a clarinet which is kept vibrating by means of an airstream.
