Prev: convert m-file to simulink
Next: MEX Fortran
From: Pedro Nunes on 14 Sep 2006 05:43
Hi I'm working with ASR using Neural networks.
I get this error when:

net=newrbe(train_vector,target_vector,spread_value);

Warning: Rank deficient, rank = 0, tol = 1.#QNBe+000.
> In newrbe>designrbe at 120
In newrbe at 103
>>

when I do:

Y=sim(net,test_vector);

I get the same result for each output --> "NAN"

Here goes my code:

t=2;
fs=22050;
centroids=16;

s=wavrecord(t*fs,fs);
codebook=mfcc_vquant_codebook(s,fs,centroids)
train_vector=[train_vector codebook(:)] % puts codebook(39*16) ->
(624*1)

my target_vector is a matrix with the same number of rows as words .
the columns match the number of samples for each train vector.

example 3 words with 3 sample for word.
target_vector=
[ 1 1 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 1 1 1 0 0 0
0 0 0 1 1 1 0 0 0
0 0 0 1 1 1 0 0 0
0 0 0 0 0 0 1 1 1
0 0 0 0 0 0 1 1 1
0 0 0 0 0 0 1 1 1 ]

the test_vector build the same way as the target_vector.

Can anyone help me

Thank You
Pedro
 | 
Pages: 1
Prev: convert m-file to simulink
Next: MEX Fortran