Digital Signal Processing Reference
In-Depth Information
" up "
FFT 00
Amplit. spec
" do wn "
FFT 01
" sto p "
FFT 02
" left "
FFT 03
Power spectr
" ri g h t "
FFT 04
Frequency domain
Power spectrum
0,07
0,06
0,05
0,04
0,03
0,02
0,01
0,00
0,125
0,100
0,075
0,050
0,025
0,000
0,050
0,045
0,040
0,035
0,030
0,025
0,020
0,015
0,010
0,005
0,000
0,125
0,100
0,075
0,050
0,025
0,000
0,125
0,100
0,075
0,050
0,025
0,000
0,00125
0,00100
0,00075
0,00050
0,00025
0,00000
0,0040
0,0035
0,0030
0,0025
0,0020
0,0015
0,0010
0,0005
0,0000
0,0007
0,0006
0,0005
0,0004
0,0003
0,0002
0,0001
0,0000
0,0040
0,0035
0,0030
0,0025
0,0020
0,0015
0,0010
0,0005
0,0000
0,0040
0,0035
0,0030
0,0025
0,0020
0,0015
0,0010
0,0005
0,0000
"up"
"up"
"down"
"down"
"stop "
"stop"
"left"
"left"
"right"
"right"
250
750
1250
1750
2250
2750
3250
3750
250
750
1250
1750
2250
2750
3250
3750
Hz
Hz
Illustration 82:
Amplitude spectrum and power spectrum
In the left hand column you see the amplitude spectra and in the right hand column the power spectra for
the same signal. You obtain a power spectrum by calculating the square of the amplitudes. What is the
sense in this?
In the amplitude spectrum you recognise the "lines" of the characteristic frequencies which come from the
vowels and which rise above the other less typical frequencies. By squaring the amplitudes of this
spectrum the characteristic frequencies are given even more weight, the less important ones are so down-
graded as to be negligible (see right hand side).
You should now examine whether it would not be better for this reason to carry out voice recognition by
means of the power spectra. The power spectrum is of great theoretical importance (WIENER-
KHINTCHINE theorem) in the field of pattern recognition.
