update my github

.gitignore

@@ -1,7 +1,8 @@
 # Compiled extensionless executable files in /src/*/
 # This stanza must precede wildcard patterns below!
 /src/*/*
-!/src/*/*.*
+!/src/lm/test_data/
+!/src/*/?*.*
 !/src/doc/*
 !/src/*/Makefile
 !/src/*/README
@@ -72,22 +73,26 @@ GSYMS
 /src/kaldi.mk.bak
 
 # /egs/
-/egs/*/s*/mfcc
-/egs/*/s*/plp
-/egs/*/s*/exp
-/egs/*/s*/data
+/egs/*/*/mfcc
+/egs/*/*/plp
+/egs/*/*/exp
+/egs/*/*/data
 
 # /tools/
+/tools/pocolm/
 /tools/ATLAS/
 /tools/atlas3.8.3.tar.gz
 /tools/irstlm/
+/tools/mitlm/
 /tools/openfst
 /tools/openfst-1.3.2.tar.gz
 /tools/openfst-1.3.2/
 /tools/openfst-1.3.4.tar.gz
 /tools/openfst-1.3.4/
 /tools/openfst-1.4.1.tar.gz
 /tools/openfst-1.4.1/
+/tools/openfst-1.5.0.tar.gz
+/tools/openfst-1.5.0/
 /tools/openfst-1.5.4.tar.gz
 /tools/openfst-1.5.4/
 /tools/openfst-1.6.0.tar.gz
@@ -96,6 +101,13 @@ GSYMS
 /tools/openfst-1.6.1/
 /tools/openfst-1.6.2.tar.gz
 /tools/openfst-1.6.2/
+/tools/openfst-1.6.5.tar.gz
+/tools/openfst-1.6.5/
+/tools/openfst-1.6.7.tar.gz
+/tools/openfst-1.6.7/
+/tools/BeamformIt/
+/tools/libsndfile-1.0.25.tar.gz
+/tools/libsndfile-1.0.25/
 /tools/pa_stable_v19_20111121.tgz
 /tools/portaudio/
 /tools/sctk-2.4.0-20091110-0958.tar.bz2
@@ -126,11 +138,13 @@ GSYMS
 /tools/srilm.tgz
 /tools/liblbfgs-1.10.tar.gz
 /tools/liblbfgs-1.10/
-/tools/openfst-1.5.0.tar.gz
-/tools/openfst-1.5.0/
 /tools/srilm-1.7.2-beta.tar.gz
 /tools/liblbfgs/
 /tools/sequitur-g2p/
-/tools/phonetisaurus-g2p
 /tools/phonetisaurus-g2p/
+/tools/mmseg-1.3.0.tar.gz
+/tools/mmseg-1.3.0/
 /kaldiwin_vs*
+/tools/cub-1.8.0.zip
+/tools/cub-1.8.0/
+/tools/cub

.travis.yml

@@ -21,6 +21,7 @@ addons:
       - gfortran-4.9
       - liblapack-dev
       - clang-3.8
+      - sox
 
 branches:
   only:
@@ -47,8 +48,8 @@ script:
   # http://peter.eisentraut.org/blog/2014/12/01/ccache-and-clang-part-3/
   # for the explanation why extra switches needed for clang with ccache.
   - CXX="ccache clang++-3.8 -Qunused-arguments -fcolor-diagnostics -Wno-tautological-compare"
-    CFLAGS="-march=native"
-    LDFLAGS="-llapack"
+    CFLAGS=""
+    LDFLAGS="-llapack -Wl,-fuse-ld=gold"
     INCDIRS="$XROOT/usr/include"
     LIBDIRS="$XROOT/usr/lib"
       tools/extras/travis_script.sh

COPYING

@@ -56,32 +56,74 @@ contributors and original source material as well as the full text of the Apache
 License v 2.0 are set forth below.
 
 Individual Contributors (in alphabetical order)
-
-      Mohit Agarwal      
+
+      Mohit Agarwal
+      Tanel Alumae
       Gilles Boulianne
       Lukas Burget
+      Dogan Can
+      Guoguo Chen
+      Gaofeng Cheng
       Cisco Corporation
-      Ondrej Glembek
+      Pavel Denisov
+      Ilya Edrenkin
+      Ewald Enzinger
+      Joachim Fainberg
+      Daniel Galvez
+      Pegah Ghahremani
       Arnab Ghoshal
+      Ondrej Glembek
       Go Vivace Inc.
+      Allen Guo
+      Hossein Hadian
+      Lv Hang
       Mirko Hannemann
+      Hendy Irawan
       Navdeep Jaitly
       Johns Hopkins University
+      Shiyin Kang
+      Kirill Katsnelson
+      Tom Ko
+      Danijel Korzinek
+      Gaurav Kumar
+      Ke Li
+      Matthew Maciejewski
+      Vimal Manohar
       Yajie Miao
       Microsoft Corporation
       Petr Motlicek
+      Xingyu Na
+      Vincent Nguyen
+      Lucas Ondel
       Vassil Panayotov
+      Vijayaditya Peddinti
+      Phonexia s.r.o.
+      Ondrej Platek
+      Daniel Povey
+      Yanmin Qian
       Ariya Rastrow
       Saarland University
-      Petr Schwarz      
-      Georg Stemmer
+      Omid Sadjadi
+      Petr Schwarz
+      Yiwen Shao
+      Nickolay V. Shmyrev
       Jan Silovsky
-      Phonexia s.r.o.
-      Yanmin Qian
-      Lucas Ondel
+      Eduardo Silva
+      Peter Smit
+      David Snyder
+      Alexander Solovets
+      Georg Stemmer
+      Pawel Swietojanski
+      Jan "Yenda" Trmal
+      Albert Vernon
       Karel Vesely
+      Yiming Wang
+      Shinji Watanabe
+      Minhua Wu
       Haihua Xu
-
+      Hainan Xu
+      Xiaohui Zhang
+
 Other Source Material
 
     This project includes a port and modification of materials from JAMA: A Java
@@ -94,9 +136,9 @@ Other Source Material
   "Signal processing with lapped transforms," Artech House, Inc., 1992.  The
   current copyright holder, Henrique S. Malvar, has given his permission for the
   release of this modified version under the Apache License 2.0.
-  
-  This project includes material from the OpenFST Library v1.2.7 available at 
-  http://www.openfst.org and released under the Apache License v. 2.0.   
+
+  This project includes material from the OpenFST Library v1.2.7 available at
+  http://www.openfst.org and released under the Apache License v. 2.0.
 
   [OpenFst COPYING file begins here]
 
@@ -105,7 +147,7 @@ Other Source Material
     You may obtain a copy of the License at
 
        http://www.apache.org/licenses/LICENSE-2.0
- 
+
     Unless required by applicable law or agreed to in writing, software
     distributed under the License is distributed on an "AS IS" BASIS,
     WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.

README.md

@@ -1,5 +1,4 @@
-[![Build Status](https://travis-ci.org/kaldi-asr/kaldi.svg?branch=master)](https://travis-ci.org/kaldi-asr/kaldi)
-
+[![Build Status](https://travis-ci.com/kaldi-asr/kaldi.svg?branch=master)](https://travis-ci.com/kaldi-asr/kaldi)
 Kaldi Speech Recognition Toolkit
 ================================
 

egs/aishell/README.txt

@@ -5,5 +5,9 @@ Aishell is an open Chinese Mandarin speech database published by Beijing Shell S
 The database can be downloaded from openslr:
 http://www.openslr.org/33/
 
+This folder contains two subfolders:
+s5: a speech recognition recipe
+v1: a speaker recognition recipe
+
 For more details, please visit:
 http://www.aishelltech.com/kysjcp

egs/aishell/s5/RESULTS

@@ -1,8 +1,18 @@
-%WER 33.82 [ 35432 / 104765, 743 ins, 3991 del, 30698 sub ] exp/mono/decode_test/cer_12_0.0
-%WER 19.39 [ 20310 / 104765, 903 ins, 1452 del, 17955 sub ] exp/tri1/decode_test/cer_13_0.5
-%WER 19.23 [ 20147 / 104765, 910 ins, 1287 del, 17950 sub ] exp/tri2/decode_test/cer_14_0.5
-%WER 17.14 [ 17961 / 104765, 812 ins, 1024 del, 16125 sub ] exp/tri3a/decode_test/cer_14_0.0
-%WER 13.64 [ 14294 / 104765, 669 ins, 736 del, 12889 sub ] exp/tri4a/decode_test/cer_14_0.5
-%WER 12.23 [ 12809 / 104765, 656 ins, 580 del, 11573 sub ] exp/tri5a/decode_test/cer_13_1.0
-%WER 8.45 [ 8849 / 104765, 312 ins, 538 del, 7999 sub ] exp/nnet3/tdnn_sp/decode_test/cer_13_1.0
-%WER 7.46 [ 7813 / 104765, 287 ins, 472 del, 7054 sub ] exp/chain/tdnn_1a_sp/decode_test/cer_10_1.0
+%WER 36.41 [ 38146 / 104765, 837 ins, 3114 del, 34195 sub ] exp/mono/decode_test/cer_10_0.0
+%WER 18.76 [ 19654 / 104765, 949 ins, 1152 del, 17553 sub ] exp/tri1/decode_test/cer_13_0.5
+%WER 18.64 [ 19531 / 104765, 941 ins, 1159 del, 17431 sub ] exp/tri2/decode_test/cer_14_0.5
+%WER 17.04 [ 17849 / 104765, 810 ins, 1021 del, 16018 sub ] exp/tri3a/decode_test/cer_14_0.5
+%WER 13.82 [ 14482 / 104765, 764 ins, 670 del, 13048 sub ] exp/tri4a/decode_test/cer_13_0.5
+%WER 12.12 [ 12694 / 104765, 751 ins, 523 del, 11420 sub ] exp/tri5a/decode_test/cer_13_0.5
+%WER 8.65 [ 9064 / 104765, 367 ins, 455 del, 8242 sub ] exp/nnet3/tdnn_sp/decode_test/cer_14_0.5
+%WER 7.48 [ 7839 / 104765, 285 ins, 454 del, 7100 sub ] exp/chain/tdnn_1a_sp/decode_test/cer_10_1.0
+
+# nnet3 tdnn with online pitch, local/nnet3/tuning/tun_tdnn_2a.sh
+%WER 8.64 [ 9050 / 104765, 349 ins, 521 del, 8180 sub ] exp/nnet3/tdnn_sp/decode_test/cer_15_0.5
+%WER 8.72 [ 9135 / 104765, 367 ins, 422 del, 8346 sub ] exp/nnet3/tdnn_sp_online/decode_test/cer_12_1.0
+%WER 9.36 [ 9807 / 104765, 386 ins, 441 del, 8980 sub ] exp/nnet3/tdnn_sp_online/decode_test_per_utt/cer_13_1.0
+
+# chain with online pitch, local/chain/tuning/run_tdnn_2a.sh
+%WER 7.45 [ 7807 / 104765, 340 ins, 497 del, 6970 sub ] exp/chain/tdnn_2a_sp/decode_test/cer_11_0.5
+%WER 7.43 [ 7780 / 104765, 341 ins, 469 del, 6970 sub ] exp/chain/tdnn_2a_sp_online/decode_test/cer_11_0.5
+%WER 7.92 [ 8296 / 104765, 384 ins, 472 del, 7440 sub ] exp/chain/tdnn_2a_sp_online/decode_test_per_utt/cer_11_0.5

egs/aishell/s5/conf/online_cmvn.conf

@@ -1 +1 @@
-# configuration file for apply-cmvn-online, used in the script ../local/run_online_decoding.sh
+# configuration file for apply-cmvn-online, used when invoking online2-wav-nnet3-latgen-faster.

egs/aishell/s5/conf/online_pitch.conf

@@ -0,0 +1,4 @@
+--sample-frequency=16000
+--simulate-first-pass-online=true
+--normalization-right-context=25
+--frames-per-chunk=10

egs/aishell/s5/local/aishell_prepare_dict.sh

@@ -15,21 +15,9 @@ mkdir -p $dict_dir
 cp $res_dir/lexicon.txt $dict_dir
 
 cat $dict_dir/lexicon.txt | awk '{ for(n=2;n<=NF;n++){ phones[$n] = 1; }} END{for (p in phones) print p;}'| \
-  sort -u |\
-  perl -e '
-  my %ph_cl;
-  while (<STDIN>) {
-    $phone = $_;
-    chomp($phone);
-    chomp($_);
-    $phone = $_;
-    next if ($phone eq "sil");
-    if (exists $ph_cl{$phone}) { push(@{$ph_cl{$phone}}, $_)  }
-    else { $ph_cl{$phone} = [$_]; }
-  }
-  foreach $key ( keys %ph_cl ) {
-     print "@{ $ph_cl{$key} }\n"
-  }
+  perl -e 'while(<>){ chomp($_); $phone = $_; next if ($phone eq "sil");
+    m:^([^\d]+)(\d*)$: || die "Bad phone $_"; $q{$1} .= "$phone "; }
+    foreach $l (values %q) {print "$l\n";}
   ' | sort -k1 > $dict_dir/nonsilence_phones.txt  || exit 1;
 
 echo sil > $dict_dir/silence_phones.txt

egs/aishell/s5/local/aishell_train_lms.sh

@@ -23,7 +23,7 @@ kaldi_lm=`which train_lm.sh`
 if [ -z $kaldi_lm ]; then
   echo "$0: train_lm.sh is not found. That might mean it's not installed"
   echo "$0: or it is not added to PATH"
-  echo "$0: Use the script tools/extra/install_kaldi_lm.sh to install it"
+  echo "$0: Use the script tools/extras/install_kaldi_lm.sh to install it"
   exit 1
 fi