function process_audio()

load session_info.mat;
% load chunk_info.mat
path = 'features/';
wav_dir = 'wav/';
wav_ext = '.wav';
d = 0;

for g=1:4
    for s =1:12
        files = [];
        person1 = [wav_dir session{1,s,g} wav_ext];
        person2 = [wav_dir session{2,s,g} wav_ext];
        person3 = [wav_dir session{3,s,g} wav_ext];
        person4 = [wav_dir session{4,s,g} wav_ext];
        count = 1;
        if (exist(person1)) 
            files{count} = person1;
            count = count + 1;
        end
        if (exist(person2)) 
            files{count} = person2;
            count = count + 1;
        end
        if (exist(person3)) 
            files{count} = person3;
            count = count + 1;
        end
        if (exist(person4)) 
            files{count} = person4;
            count = count + 1;
        end
        if (~isempty(files))        
            d = d+1
            names{d} = files;
            sessions{d} = ['group_' num2str(g) '_session_' num2str(s)];
            %orignal code to create features
%             features = speech_features_separate_files(files);
%             eval(['features_' num2str(g) '_' num2str(s) ' = features;']);
%             save([path 'features_' num2str(g) '_' num2str(s)],['features_' num2str(g) '_' num2str(s)]);
            load([path 'features_' num2str(g) '_' num2str(s)]);
            eval(['features = features_' num2str(g) '_' num2str(s) ';']);
            clear(['features_' num2str(g) '_' num2str(s)]);            
            [states_voiced, states_speaking, voiced_all{d}] = voicing_speaking_sip(features, 'threshold', 0.10);
            all_states_voiced{d} = states_voiced;
            all_states_speaking{d} = states_speaking;
            [one_min_means{d}, one_min_stds{d}, one_min_others{d}] = chunk_features(features, all_states_voiced{d}, all_states_speaking{d},1);    
            one_min_alphas{d} = chunk_influence(states_speaking,1);
            [total_means{d}, total_stds{d}, total_others{d}] = chunk_features(features, all_states_voiced{d}, all_states_speaking{d},inf);    
            total_alphas{d} = chunk_influence(states_speaking,inf);
            save('audio_features','one_min*','total*','all_states_speaking','all_states_voiced','names','sessions','voiced_all');        
        end
    end
end