annotate Code/Descriptors/Matlab/MPEG7/FromWeb/VoiceSauce/func_GetH1H2_H2H4.m @ 4:92ca03a8fa99 tip

Update to ICASSP 2013 benchmark
author Dawn Black
date Wed, 13 Feb 2013 11:02:39 +0000
parents
children
rev   line source
Dawn@4 1 function [H1H2, H2H4] = func_GetH1H2_H2H4(H1, H2, H4, Fs, F0, F1, F2, B1, B2)
Dawn@4 2 % [H1H2, H2H4] = func_GetH1H2_H2H4(H1, H2, H4, Fs, F0, F1, F2, B1, B2)
Dawn@4 3 % Input: H1, H2, H4, vectors
Dawn@4 4 % Fs - sampling frequency
Dawn@4 5 % F0 - vector of fundamental frequencies
Dawn@4 6 % Fx, Bx - vectors of formant frequencies and bandwidths
Dawn@4 7 % Output: H1A1, H1A2, H1A3 vectors
Dawn@4 8 % Notes: Function produces the corrected versions of the parameters. They
Dawn@4 9 % are stored as HxHx for compatibility reasons. Use func_buildMData.m to
Dawn@4 10 % recreate the mat data with the proper variable names.
Dawn@4 11 % Also note that the bandwidths from the formant trackers are not currently
Dawn@4 12 % used due to the variability of those measurements.
Dawn@4 13 %
Dawn@4 14 % Author: Yen-Liang Shue, Speech Processing and Auditory Perception Laboratory, UCLA
Dawn@4 15 % Copyright UCLA SPAPL 2009
Dawn@4 16
Dawn@4 17
Dawn@4 18 if (nargin == 7)
Dawn@4 19 B1 = func_getBWfromFMT(F1, F0, 'hm');
Dawn@4 20 B2 = func_getBWfromFMT(F2, F0, 'hm');
Dawn@4 21 end
Dawn@4 22
Dawn@4 23
Dawn@4 24 H1_corr = H1 - func_correct_iseli_z(F0, F1, B1, Fs);
Dawn@4 25 H1_corr = H1_corr - func_correct_iseli_z(F0, F2, B2, Fs);
Dawn@4 26 H2_corr = H2 - func_correct_iseli_z(2*F0, F1, B1, Fs);
Dawn@4 27 H2_corr = H2_corr - func_correct_iseli_z(2*F0, F2, B2, Fs);
Dawn@4 28 H4_corr = H4 - func_correct_iseli_z(4*F0, F1, B1, Fs);
Dawn@4 29 H4_corr = H4_corr - func_correct_iseli_z(4*F0, F2, B2, Fs);
Dawn@4 30
Dawn@4 31 H1H2 = H1_corr - H2_corr;
Dawn@4 32 H2H4 = H2_corr - H4_corr;
Dawn@4 33
Dawn@4 34
Dawn@4 35