Digital Signal Processing Reference
In-Depth Information
ADC Input
Data Trigger
Write AS
Relay
Ti me domai n
FFT
Freq.domai n
Blockaverage
A verage
Write PS
10, 0
7, 5
5, 0
2, 5
0, 0
-2, 5
-5, 0
-7, 5
- 10, 0
Time domain
„up“
0. 00
0. 25
0. 50
0. 75
1. 00
Y/t Chart 0
s
0, 35
0, 30
0, 25
0, 20
0, 15
0, 10
0, 05
0, 00
0, 0275
0, 0250
0, 0225
0, 0200
0, 0175
0, 0150
0, 0125
0, 0100
0, 0075
0, 0050
0, 0025
0, 0000
Single Amplitude Spectrum
Single Power Spectrum
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
3250
3500
3750
Y /t Chart 0
Y /t Chart 1
Hz
0, 175
0, 150
0, 125
0, 100
0, 075
0, 050
0, 025
0, 000
0, 009
0, 008
0, 007
0, 006
0, 005
0, 004
0, 003
0, 002
0, 001
0, 000
Averaged Amplitude Spectrum
Averaged Power Spectrum
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
3250
3500
3750
Y /t Chart 0
Y /t Chart 1
Hz
„down“
10, 0
7, 5
5, 0
2, 5
0, 0
-2, 5
-5, 0
-7, 5
- 10, 0
Time domain
0. 00
0. 25
0. 50
0. 75
1. 00
Y /t Chart 0
s
0, 40
0, 35
0, 30
0, 25
0, 20
0, 15
0, 10
0, 05
0, 00
0, 035
0, 030
0, 025
0, 020
0, 015
0, 010
0, 005
0, 000
Single Amplitude Spectrum
Single Power Spectrum
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
3250
3500
3750
Y /t Chart 0
Y /t Chart 1
Hz
0, 250
0, 225
0, 200
0, 175
0, 150
0, 125
0, 100
0, 075
0, 050
0, 025
0, 000
0, 0200
0, 0175
0, 0150
0, 0125
0, 0100
0, 0075
0, 0050
0, 0025
0, 0000
Averaged Amplitude Spectrum
Averaged Power Spectrum
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
3250
3500
3750
Y /t Chart 0
Y /t Chart 1
Hz
Illustration 83:
Tolerant method for speech recognition
The method according to Illustration 81 has some disadvantages, in terms of frequency shall specifically
analyze to "arm " that very small changes in vocal pitch, which manifest themselves as slight change in
frequency affect significantly the correlation factor and yield false results quickly.
In search of "stable" method, offers the idea by averaging several reference spectra of the same word to
achieve more tolerance in pattern recognition. It is hoped thereby to crystallize what is typical of the refer-
ence spectrum better.
Above, the circuit can be shown, with the help of which averaged reference spectra for amplitude and
power density create and save. In this and Illustration 84 are recorded for each of the five words, all sig-
nals for time and frequency domain.
A closer look at these spectra shows that the averaged spectra well the general tendency of all spectra bet-
ter than the spectrum represents a single sample, which always include an atypical-between "gaffe".
You should try to achieve with your own voice samples improved word recognition!
