Error when trying to train

[JRMeyer]

>>> sawantilak
[December 14, 2017, 2:49pm]

Hi, slash
I was following the instructions to train the model, but I getting the
below error. Can some one please help me identify why I am getting this
error and how can I get this fixed?

(deepspeech-venv) sawan slash Sawan-Office: slash ~/pfiles/ds_home/DeepSpeech slash $
./DeepSpeech.py slash --train_files
.../.../.../data/deepvoice/cv_corpus_v1/cv-valid-train.csv,.../.../.../data/deepvoice/cv_corpus_v1/cv-other-train.csv
slash --dev_files .../.../.../data/deepvoice/cv_corpus_v1/cv-valid-dev.csv
slash --test_files .../.../.../data/deepvoice/cv_corpus_v1/cv-valid-test.csv slash
Traceback (most recent call last): slash
File './DeepSpeech.py', line 1828, in slash
tf.app.run() slash
File
'/home/sawan/venv/deepspeech-venv/local/lib/python2.7/site-packages/tensorflow/python/platform/app.py',
line 48, in run slash
slash _sys.exit(main( slash _sys.argv slash [:1 slash ] + flags_passthrough)) slash
File './DeepSpeech.py', line 1785, in main slash
train() slash
File './DeepSpeech.py', line 1465, in train slash
next_index=lambda i: COORD.get_next_index('train')) slash
File '/home/sawan/pfiles/ds_home/DeepSpeech/util/feeding.py', line 95,
in init slash
self.files = self.files.sort_values(by='wav_filesize',
ascending=ascending) slash
File
'/home/sawan/venv/deepspeech-venv/local/lib/python2.7/site-packages/pandas/core/frame.py',
line 3619, in sort_values slash
k = self.xs(by, axis=other_axis).values slash
File
'/home/sawan/venv/deepspeech-venv/local/lib/python2.7/site-packages/pandas/core/generic.py',
line 2335, in xs slash
return self slash [key slash ] slash
File
'/home/sawan/venv/deepspeech-venv/local/lib/python2.7/site-packages/pandas/core/frame.py',
line 2139, in getitem slash
return self. slash _getitem_column(key) slash
File
'/home/sawan/venv/deepspeech-venv/local/lib/python2.7/site-packages/pandas/core/frame.py',
line 2146, in slash _getitem_column slash
return self. slash _get_item_cache(key) slash
File
'/home/sawan/venv/deepspeech-venv/local/lib/python2.7/site-packages/pandas/core/generic.py',
line 1842, in slash _get_item_cache slash
values = self. slash _data.get(item) slash
File
'/home/sawan/venv/deepspeech-venv/local/lib/python2.7/site-packages/pandas/core/internals.py',
line 3843, in get slash
loc = self.items.get_loc(item) slash
File
'/home/sawan/venv/deepspeech-venv/local/lib/python2.7/site-packages/pandas/core/indexes/base.py',
line 2527, in get_loc slash
return self. slash _engine.get_loc(self. slash _maybe_cast_indexer(key)) slash
File 'pandas/ slash _libs/index.pyx', line 117, in
pandas. slash _libs.index.IndexEngine.get_loc slash
File 'pandas/ slash _libs/index.pyx', line 139, in
pandas. slash _libs.index.IndexEngine.get_loc slash
File 'pandas/ slash _libs/hashtable_class_helper.pxi', line 1265, in
pandas. slash _libs.hashtable.PyObjectHashTable.get_item slash
File 'pandas/ slash _libs/hashtable_class_helper.pxi', line 1273, in
pandas. slash _libs.hashtable.PyObjectHashTable.get_item slash
KeyError: 'wav_filesize'

[This is an archived TTS discussion thread from discourse.mozilla.org/t/error-when-trying-to-train]

[JRMeyer]

>>> lissyx
[December 14, 2017, 3:21pm]

Can you make sure you post the stacktrace using backquotes? Otherwhise
some characters are missing and it's harder to read
![:slight_smile:](

like that

[Archived Post]

[JRMeyer]

>>> lissyx
[December 14, 2017, 3:22pm]

The error mentions wav_filesize, can you make sur your CSV files does
have that column?

[Archived Post]

[JRMeyer]

>>> sawantilak
[December 15, 2017, 6:02am]

I checked and the header 'wav_filesize' exists in all the files. I am
trying to attach the csv files here, but this site is not allowing me to
upload as I am a new user.

Would appreciate anyone who can help me out with this issue. Any
pointers on how I can troubleshoot this one?

[Archived Post]

[JRMeyer]

>>> sawantilak
[December 15, 2017, 6:31am]

Thanks lissyx for taking out the time to reply and trying to help me. I
found out what was wrong, had to run DeepSpeech.py in debugger mode to
find out.

Turns out there are 2 sets of slash *.csv files getting generated and I have
to pick the csv files in the root folder to train. I was picking the
wrong files which had the below headers. I pointed to the right files,
problem solved

filename text up_votes down_votes age gender accent duration

Thanks

[Archived Post]

[JRMeyer]

>>> phanthanhlong7695
[January 30, 2018, 9:45am]

what can i do to fix that ?

[Archived Post]

[JRMeyer]

>>> sawantilak
[January 30, 2018, 2:05pm]

You are probably using the wrong csv file. See if there are csv files in
the parent folder with 3 columns. Use those csv files.

Sawan.

[Archived Post]

[JRMeyer]

>>> lissyx
[January 30, 2018, 2:25pm]

I'm taking the opportunity, since you found your issue, would you mind
updating the title with a better wording to make it easier for others to
find your specific problem? Thanks!

[Archived Post]