logging.info("%s written." % fn)
fn = '%s/local/lm/wordlist.txt' % data_dir
with open(fn, 'w') as f:
for token in sorted(utt_dict):
f.write((u'%s\n' % token).encode('utf8'))
logging.info("%s written." % fn)
#
# copy scripts and config files
#
misc.copy_file('data/src/speech/kaldi-run-lm.sh', '%s/run-lm.sh' % work_dir)
# misc.copy_file ('data/src/speech/kaldi-run-am.sh', '%s/run-am.sh' % work_dir)
# misc.copy_file ('data/src/speech/kaldi-run-nnet3.sh', '%s/run-nnet3.sh' % work_dir)
misc.copy_file('data/src/speech/kaldi-run-chain.sh',
'%s/run-chain.sh' % work_dir)
misc.copy_file('data/src/speech/kaldi-cmd.sh', '%s/cmd.sh' % work_dir)
misc.copy_file('data/src/speech/kaldi-path.sh', '%s/path.sh' % work_dir)
misc.mkdirs('%s/conf' % work_dir)
misc.copy_file('data/src/speech/kaldi-mfcc.conf',
'%s/conf/mfcc.conf' % work_dir)
misc.copy_file('data/src/speech/kaldi-mfcc-hires.conf',
'%s/conf/mfcc_hires.conf' % work_dir)
misc.copy_file('data/src/speech/kaldi-online-cmvn.conf',
'%s/conf/online_cmvn.conf' % work_dir)
misc.mkdirs('%s/local' % work_dir)
misc.copy_file('data/src/speech/kaldi-score.sh',
misc.mkdirs('%s/valid' % data_dir)
misc.mkdirs('%s/train' % data_dir)
#
# load dict
#
logging.info("loading lexicon...")
lex = Lexicon(file_name=dictionary)
logging.info("loading lexicon...done.")
#
# language model
#
misc.copy_file('%s/lm.arpa' % language_model_dir, '%s/lm.arpa' % data_dir)
#
# scripts
#
misc.render_template('data/src/speech/w2l_run_train.sh.template', '%s/run_train.sh' % work_dir, w2l_env_activate=w2l_env_activate, w2l_train=w2l_train)
misc.mkdirs('%s/config/conv_glu' % work_dir)
misc.render_template('data/src/speech/w2l_config_conv_glu_train.cfg.template', '%s/config/conv_glu/train.cfg' % work_dir, runname=model_name)
misc.copy_file('data/src/speech/w2l_config_conv_glu_network.arch', '%s/config/conv_glu/network.arch' % work_dir)
#
# export audio
#
