Digital Signal Processing Reference
In-Depth Information
(a)
(b)
0.0
55.0
110.0
165.0
220.0
0.0
1.0
2.0
3.0
Time (samples)
Frequency (kHz)
(c)
(d)
0.0
55.0
110.0
165.0
220.0
0.0
1.0
2.0
3.0
Figure 4.4
Effects of window types on voiced speech with a 220 sample window
length: (a) and (b) show the time and frequency plots of speech using a rectangular
window, and (c) and (d) show the time and frequency plots of speech using a
Hamming window
relative to that produced by the Hamming window. However the high
frequency leakage produced by the larger side lobes makes rectangular
windowed speech look more noisy. This undesirable high frequency leakage
between adjacent harmonics tends to offset the benefits of the flat time domain
response (greater frequency resolution) of the rectangular window. As a
result, rectangular windows are not usually used in speech spectral analysis.
The effect of windowing unvoiced speech is shown in Figure 4.5. Again
the spectra are slowly varying with a series of sharp peaks and valleys.
The noisy appearance of the spectrum (for both windows) however, is due
to the random nature of unvoiced speech. Although the signal itself is
random, again the Hamming window produces a smoother spectrum than
the rectangular window.
Search WWH ::
Custom Search