From: <HIDDEN>
Newsgroups: comp.soft-sys.matlab
Subject: psychoacoustical Masking.
Date: Fri, 26 Mar 2010 13:42:04 +0000 (UTC)
Organization: Concordia
Lines: 10
Message-ID: <hoidjc$t5q$>
Reply-To: <HIDDEN>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
X-Trace: 1269610924 29882 (26 Mar 2010 13:42:04 GMT)
NNTP-Posting-Date: Fri, 26 Mar 2010 13:42:04 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 1057560
Xref: comp.soft-sys.matlab:620649

I am trying to make those frequencies of my speech signal combined by white gaussian noise inaudible .First,I mapped from frequency to Bark scale by  z=13*atan((.76*T)/1000)+3.5*atan((T/7500).^2); second, defined energy per critical band E.then obtained the spread masking across CBR which is defined as  C=conv(E,B); here B is basiilar membrane spreading function B(dB)=B=15.81+7.5*(z+.474)-17.5*sqrt(1+(z+.474).^2); finally,masking threshold function is calculated by MASK=((10.^(P-Q))./N); 

where :

My problem is reffered to as Matrix mistmatch between P and Q in matlab.that sounds reasonable because if you assume E and B by size L.their convolution C should be at most 2*L-1 and at least L.This is due to the Convolution property,considering Ofset=delta*(14.5+z)+(1-delta)*5.5;  Q seems to be a matrix by size L.I really need your help to find out how these matixes fit together dimensionally to calculate MASK.