Menu ▾ ▴

Training AM : aborted k-means

Help
Carl Geißner
2012-04-02
2012-09-22

  • Carl Geißner

    Carl Geißner - 2012-04-02 
    ...
INFO: main.c(613):  -> Aborting k-means, bad initialization
INFO: kmeans.c(153): km iter [0] 1.000000e+00 ...
WARNING: "kmeans.c", line 431: Empty cluster 29
WARNING: "kmeans.c", line 431: Empty cluster 39
WARNING: "kmeans.c", line 431: Empty cluster 43
WARNING: "kmeans.c", line 431: Empty cluster 45
WARNING: "kmeans.c", line 431: Empty cluster 55
WARNING: "kmeans.c", line 431: Empty cluster 60
WARNING: "kmeans.c", line 431: Empty cluster 62
WARNING: "kmeans.c", line 431: Empty cluster 77
WARNING: "kmeans.c", line 431: Empty cluster 84
WARNING: "kmeans.c", line 431: Empty cluster 88
WARNING: "kmeans.c", line 431: Empty cluster 104
WARNING: "kmeans.c", line 431: Empty cluster 110
WARNING: "kmeans.c", line 431: Empty cluster 113
WARNING: "kmeans.c", line 431: Empty cluster 116
WARNING: "kmeans.c", line 431: Empty cluster 124
WARNING: "kmeans.c", line 431: Empty cluster 134
WARNING: "kmeans.c", line 431: Empty cluster 137
WARNING: "kmeans.c", line 431: Empty cluster 139
WARNING: "kmeans.c", line 431: Empty cluster 142
WARNING: "kmeans.c", line 431: Empty cluster 144
WARNING: "kmeans.c", line 431: Empty cluster 147
WARNING: "kmeans.c", line 431: Empty cluster 153
WARNING: "kmeans.c", line 431: Empty cluster 154
WARNING: "kmeans.c", line 431: Empty cluster 155
WARNING: "kmeans.c", line 431: Empty cluster 157
WARNING: "kmeans.c", line 431: Empty cluster 159
WARNING: "kmeans.c", line 431: Empty cluster 169
WARNING: "kmeans.c", line 431: Empty cluster 170
WARNING: "kmeans.c", line 431: Empty cluster 172
WARNING: "kmeans.c", line 431: Empty cluster 176
WARNING: "kmeans.c", line 431: Empty cluster 178
WARNING: "kmeans.c", line 431: Empty cluster 181
WARNING: "kmeans.c", line 431: Empty cluster 182
WARNING: "kmeans.c", line 431: Empty cluster 184
WARNING: "kmeans.c", line 431: Empty cluster 186
WARNING: "kmeans.c", line 431: Empty cluster 187
WARNING: "kmeans.c", line 431: Empty cluster 189
WARNING: "kmeans.c", line 431: Empty cluster 190
WARNING: "kmeans.c", line 431: Empty cluster 191
WARNING: "kmeans.c", line 431: Empty cluster 196
WARNING: "kmeans.c", line 431: Empty cluster 198
WARNING: "kmeans.c", line 431: Empty cluster 199
WARNING: "kmeans.c", line 431: Empty cluster 200
WARNING: "kmeans.c", line 431: Empty cluster 201
WARNING: "kmeans.c", line 431: Empty cluster 202
WARNING: "kmeans.c", line 431: Empty cluster 205
WARNING: "kmeans.c", line 431: Empty cluster 206
WARNING: "kmeans.c", line 431: Empty cluster 207
WARNING: "kmeans.c", line 431: Empty cluster 208
WARNING: "kmeans.c", line 431: Empty cluster 209
WARNING: "kmeans.c", line 431: Empty cluster 211
WARNING: "kmeans.c", line 431: Empty cluster 213
WARNING: "kmeans.c", line 431: Empty cluster 219
WARNING: "kmeans.c", line 431: Empty cluster 220
WARNING: "kmeans.c", line 431: Empty cluster 222
WARNING: "kmeans.c", line 431: Empty cluster 223
WARNING: "kmeans.c", line 431: Empty cluster 226
WARNING: "kmeans.c", line 431: Empty cluster 229
WARNING: "kmeans.c", line 431: Empty cluster 231
WARNING: "kmeans.c", line 431: Empty cluster 233
WARNING: "kmeans.c", line 431: Empty cluster 234
WARNING: "kmeans.c", line 431: Empty cluster 239
WARNING: "kmeans.c", line 431: Empty cluster 240
WARNING: "kmeans.c", line 431: Empty cluster 241
WARNING: "kmeans.c", line 431: Empty cluster 247
WARNING: "kmeans.c", line 431: Empty cluster 250
WARNING: "kmeans.c", line 431: Empty cluster 253
INFO: main.c(613):  -> Aborting k-means, bad initialization
INFO: main.c(622):  best-so-far sqerr = -1.000000e+00
ERROR: "main.c", line 841: Too few observations for kmeans
ERROR: "main.c", line 1408: Unable to do k-means for state 0; skipping...
INFO: s3gau_io.c(226): Wrote /home/user/Test/model_parameters/test.ci_semi_flatinitial/means [1x4x256 array]
INFO: s3gau_io.c(226): Wrote /home/user//Test/model_parameters/test.ci_semi_flatinitial/variances [1x4x256 array]
INFO: main.c(1509): No mixing weight file given; none written
INFO: main.c(1669): TOTALS: km 0.024x 9.989e-01 var 0.000x 0.000e+00 em 0.000x 0.000e+00 all 0.024x 1.002e+00
Mon Apr  2 12:01:08 2012

    Hey there,

    I tried to train a new model for the German language. After I got errors which
    I suppose were UTF8-Decoding problems (ä, ö, ü weren't recognized as such in
    the dictionary) I minimized everything to one single transcription-sentence
    and the corresponding words in the dictionary. What I got after running : 

    ./scripts_pl/make_feats.pl -ctl etc/test_train.fileids
./scripts_pl/make_feats.pl -ctl etc/test_test.fileids
./scripts_pl/RunAll.pl

    was the error on the top. I suppose it is because I gave not enough
    observation material ?

     

  • Carl Geißner

    Carl Geißner - 2012-04-03

    Okay, I found the answer. The search engine is your friend ;)

    Can somebody give me hints on how to train an existing language model with new
    audio data (acoustic model). In my instance I want to train an existing 16khz
    model with 8khz audio data, but I don't see an advantage over creating a
    totally new model despite I can use the existing language model.

     

  • Nickolay V. Shmyrev

    Nickolay V. Shmyrev - 2012-04-07

    how to train an existing language model with new audio data (acoustic
    model). In my instance I want to train an existing 16khz model with 8khz audio
    data

    It's easier to train 8khz model from scratch.

     

  • Saumya Vij

    Saumya Vij - 2012-07-02

    Hi mikadomoses!

    I am trainging the TIdigits data and I am getting the same error. I have
    sufficient data for training if I compare it with the an4 corpus.
    How did you solve this problem?

     

  • Saumya Vij

    Saumya Vij - 2012-07-02

    Sorry i forgot to add, how did you solve the first error that you have posted.

     

Log in to post a comment.