annotate Code/Test/checkVUV.m @ 4:92ca03a8fa99 tip

Update to ICASSP 2013 benchmark

function [] = checkVUV()

sampleName = '01_a_male01_pos';
[x, fs, frameLength, noOfFrames] = openFile( [sampleName '.wav'] );
y = buffer( x, frameLength );

 %---------------- GET THE SILENT FRAME VALUES -------------------
    
% only wish to consider pitch values from voiced frames.
% silent and unvoiced frames will produce pitch values that
% are random and therefore will bias our results
segmentFrames = getSilenceDecision( sampleName );
[ silentFrames ] = removeSilentData( segmentFrames, noOfFrames );
    

% % open original annotation file
% fileName = [sampleName '.txt'];
% % read metrics from file
% fileID = fopen( fileName );
% data = fscanf( fileID, '%d', inf );
% annotatedPitch = data(3:4:end);
% fclose( fileID );

% open HNR file
fileName = [sampleName '_HNR.txt'];
% read metrics from file
fileID = fopen( fileName );
data = fscanf( fileID, '%f', inf );
harmonic2noise = data(2:2:end);
fclose( fileID );

largestH2NValue = max( [max(harmonic2noise) abs(min(harmonic2noise))] );
%normalise
harmonic2noise = harmonic2noise/largestH2NValue;
%round to 1dp
harmonic2noise= (round(harmonic2noise*10))/10;

% open audio power file
fileName = [sampleName '_AP.txt'];
% read metrics from file
fileID = fopen( fileName );
data = fscanf( fileID, '%f', inf );
audioPower = data(2:2:end);
fclose( fileID );
maxPower = max(audioPower);


minFreq = ceil(fs/getVariables('getMinFreq')); %default minimum fundatmental frequency
maxFreq = ceil(fs/getVariables('getMaxFreq')); %default maximum fundamental frequency
    
% open pitch file
fileName = [sampleName '_YIN_pitch.txt'];
% read metrics from file
fileID = fopen( fileName );
data = fscanf( fileID, '%f', inf );
pitch = data(2:2:end);
fclose( fileID );

% open band power file
% fileName = [sampleName '_ASBP_norm.txt'];
% % read metrics from file
% fileID = fopen( fileName );
% data = fscanf( fileID, '%f', inf );
% bandPowerRatio = data(2:2:end);
% fclose( fileID );
% bandPowerRatioRMS =   sqrt( mean(( bandPowerRatio/max(bandPowerRatio) ).^2 ));
% 


HNRThresh = 0;
hold off; 
minLength = min([length(harmonic2noise) length(audioPower) length(pitch)]);
plotUnits = 1/frameLength;
for( i=1:minLength )


     if( silentFrames(i) == 0 ) % harmonic2noise(i) == 0 )
         %silent
%         plot( ((i-1)*frameLength)+1:(i*frameLength), y(:,i), 'y' );hold on; 
        plot( i-0.5 + plotUnits : plotUnits : i+0.5, y(:,i), 'y' );hold on; 
        vuv(i) = 0;
     elseif( ((harmonic2noise(i) <= HNRThresh))... %|| (bandPowerRatio(i)/max(bandPowerRatio) < 0.001)) ...
            || (pitch(i) > minFreq) ...
            || (pitch(i) < maxFreq)) %...
%             || (audioPower(i)/maxPower < 0.0005))
        % unvoiced
%         plot( ((i-1)*frameLength)+1:(i*frameLength), y(:,i), 'r' );hold on;
        plot( i-0.5 + plotUnits : plotUnits : i+0.5, y(:,i), 'r' );hold on;
        vuv(i) = 2;
     else
         %voiced
%         plot( ((i-1)*frameLength)+1:(i*frameLength), y(:,i), 'g' );hold on;
        plot( i-0.5 + plotUnits : plotUnits : i+0.5, y(:,i), 'g' );hold on;
        vuv(i) = 1;
    end
end

    
% plot( 1 : 1 : length(harmonic2noise), harmonic2noise, 'b' );hold on;
% plot( 1 : 1 : length(pitch), pitch/max(pitch), 'w' );hold on;
% % pitch thresholds
% L=line([0 noOfFrames], [maxFreq/max(pitch) maxFreq/max(pitch)]);
% set(L,'color',[1 0 0]);
% L=line([0 noOfFrames], [minFreq/max(pitch) minFreq/max(pitch)]);
% set(L,'color',[1 0 0]);
% 
% %HNR threshold
% L=line([0 noOfFrames], [HNRThresh HNRThresh]);
% set(L, 'color', [0 0 1] );