Commit 731779d4 authored by Dan Povey's avatar Dan Povey
Browse files

trunk: add wsj joint training in top-level run.sh (commented out).

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@4761 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8
parent bc2cf518
#!/bin/bash
# note: see the newer, better script run_nnet2_wsj_joint.sh
# This script assumes you have previously run the WSJ example script including
# the optional part local/online/run_online_decoding_nnet2.sh. It builds a
# neural net for online decoding on top of the network we previously trained on
......
......@@ -177,8 +177,21 @@ steps/decode.sh --config conf/decode.config --nj 20 --cmd "$decode_cmd" \
# which calls local/online/run_gmm_pitch.sh
#
# local/run_nnet2.sh
# local/online/run_nnet2_baseline.sh
# local/online/run_nnet2_multisplice.sh
# local/online/run_nnet2_multisplice_disc.sh
# ##some older scripts:
# # local/run_nnet2.sh
# # local/online/run_nnet2_baseline.sh
# ## if you have a WSJ setup, you can use the following script to do joint
# ## RM/WSJ training; this doesn't require that the phone set be the same, it's
# ## a demonstration of a multilingual script.
# local/online/run_nnet2_wsj_joint.sh
# ## The following is an older way to do multilingual training, from an
# ## already-trained system.
# #local/online/run_nnet2_wsj.sh
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment