| Stationarising speech signals [message #14969] |
Thu, 08 April 1999 00:00 |
Mark McGillion
Messages: 14
Registered: December 1997
|
Junior Member |
|
|
Hi,
I am working with speech signals (acoustic and impedance signals
captured via an Electrolaryngograph) that suffer from drifting DC.
To remove this drift, I am stationarising the signal by forward
differencing as follows:
FOR i=0,N-2
x[i]=x[i+1]-x[i] ;x is a vector containing the signal
ENDFOR
This works fine for most signals and removes a great deal of the power
in the lower frequency ranges, particularly where there is strong drift.
However, where the drift is weak, forward differencing simply kills the
fundamental frequency and distorts the harmonics.
A possible solution is to high-pass filter the signal, say from 50Hz to
Nyquist (10kHz in this case), removing most of the power in the DC
drift.
Does anyone have experience of this and/or any suggestions, e.g. if
filtering is the solution, what type/order of filter do you recommend?
Regards,
Mark
please also respond to me at mm@fs1.co.umist.ac.uk.nospam (removing the
.nospam at the end).
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
