speech recognition using knn

See more:

C

MatLab

speech

recognition

hii,
I have used mfcc for feature extraction of speech samples and then normalized them using min_max algorithm.Now I want to take 70% of them for training and 30% for sampling or testing. I am unable to do that using the code which I have.I am writing the code below .I ma thankful to the person for helping me in advance.

clc;
training = min_max_norm(n700train,-1,1);% training sample or matrix
sample = min_max_norm(n300test,-1,1);%testing samples or matrix

no_obj = 2;
no_sample = 350;
u=no_obj*(no_sample);
group=repmat(1:no_obj,[no_sample,1]);
group=group(:);

CLASS = knnclassify(sample,training,group,1,'euclidean');
res=sum((CLASS==group))/u;
fprintf( 'The result of classification is %f%% \n\n',res*100/2);

%NEW code%
for j = 1: no_obj

       rs = reshape(CLASS',250,no_obj);
       sm = sum(rs(j,:)==j);
       if (sm > 3)
            fprintf('\nSpeaker %d is matched with Speaker %d',j,j);
        else
            fprintf('\nSpeaker %d is not matched ',j);
        end
   end
    fprintf('\n\n%d samples matched out of %d\n',res*u,u);
    fprintf( 'The result of classification is %f%% \n\n',res*100);

it shows following error

??? Error using ==> eq
Matrix dimensions must agree.

Error in ==> test_one at 12
res=sum((CLASS==group))/u;

please help me .my mail_id is [email id removed - once someone asnswers, automatically an email will be sent to registered id.]