Speech Signal Analysis and Modelling - Digital Speech: Coding for Low Bit Rate Communication Systems - page 56

Voiced Speech

Unvoiced Speech

0.0 50.0 100.0 150.0 200.0 250.0 300.0

Time (Samples)

Figure 4.1 Voiced and unvoiced speech waveforms (unvoiced amplified by 5)

useful definition of the time-dependent Fourier transform is [1],

∞

S k (e jω ) n)s(n)e − jωn = w(k − (4.1)

n =−∞

where w(k n) is a real window sequence used to isolate the portion of

the input signal that will be analysed at a particular time index, k . During
the analysis of speech signals, the shape and length of the window can
affect the frequency representation of speech (or any other signal). Various
types of window have been studied by researchers, producing window
shapes and characteristics suitable for various applications. In the following,
a brief description of windowing and its effects on the short-time Fourier
representation are given.

−

4.2.1 RoleofWindows
The window, w(n) , determines the portion of the speech signal that is to be
processed by zeroing out the signal outside the region of interest. The ideal
window frequency response has a very narrow main lobe which increases
the resolution and no side lobes (or frequency leakage). Since such a window
is not possible in practice, a compromise is usually selected for each specific
application. There are many possible windows (e.g. Rectangular, Bartlett,