[egs] swbd/s5c, added 5 layer (b)lstm recipes (#1759)
[scripts] Fix bug in segment_long_utterances.sh (#1758)
[src] Fix indexing error in nnet1::Convolutional2DComponent (#1755)
[src] Fix usage message of program (thanks:jubang0219@gmail.com)
[egs] some small updates to scripts (installing beamformit; segmentation example)
[egs] Small fix to ami/s5b/local/chain/compare_wer_general.sh (#1751)
[build] Add configuration check for incompatible g++ compilers when CUDA is enabled. (#1749)
[egs] Update Librispeech nnet3 TDNN recipe (old one did not run) (#1727)
[egs] APIAI example: model downloads links changed (#1747)
[src] remove remap-leaves.h (thanks: @kkm000)
[egs] Add updated TDNN+BLSTM scripts for swbd/s5c, with dropout etc. (#1730)
[src] Make sure softmax-related functions can work in-place. (#1729)
[src] Bug-fix in model-collapsing code (thanks: aarora8)
[scripts] bug-fix in nnet3 scripts: change type of max_lda_jobs to int (#1728)
[scripts,egs] simplify nnet3 scripts by removing unused feature types (LDA, delta); add sMBR recipe for mini-librispeech (#1711)
[src] Fix bug in lattice string-pushing, affecting lattice-push (#1724)
note from Dan: probably this bug did not affect the behavior of the tool.
note from Dan: probably this bug did not affect the behavior of the tool.
[src] Fix for threaded nnet2 decoding: check if threads are joinable before calling join(), to avoid multiple calls to join() (#1725)
[scripts] run.pl: Default concurrent jobs to number of GPUs (#1723)
When --gpu 1 is specified on command line, and no explicit --max-jobs-run
is provided, then the default is set to the number of GPUs in the system,
as enumerated by `nvidia-smi -L`.
Closes: #1720
When --gpu 1 is specified on command line, and no explicit --max-jobs-run
is provided, then the default is set to the number of GPUs in the system,
as enumerated by `nvidia-smi -L`.
Closes: #1720
[scripts] Remove bogus note on CUDA non-use from compute_average_posterior() (#1722)
[scripts] Quote '{' in perl regexp (#1721)
Fix Perl warninig "Unescaped left brace in regex is deprecated." This use
has been deprecated in Perl 5.22, and would become an erorr in 5.26.
http://search.cpan.org/dist/perl-5.22.0/pod/perldelta.pod#A_literal_%22{%22_should_now_be_escaped_in_a_pattern
https://unix.stackexchange.com/a/238708/103076
The use of [{] vs. \{ is probably the most backward-compatible.
Fix Perl warninig "Unescaped left brace in regex is deprecated." This use
has been deprecated in Perl 5.22, and would become an erorr in 5.26.
http://search.cpan.org/dist/perl-5.22.0/pod/perldelta.pod#A_literal_%22{%22_should_now_be_escaped_in_a_pattern
https://unix.stackexchange.com/a/238708/103076
The use of [{] vs. \{ is probably the most backward-compatible.
[build] update tools/extras/install_speex.sh to address #1718 (#1719)
[egs] improve TDNN model in tedlium example (fewer jobs, proportinal-shrink 20) (#1715)
[build] IRSTLM build: resolve problems with compilers by patching configure.ac (#1713)
[egs,scripts]: replace non-portable read-link -f with utils/make_absolute.sh (#1694)
* remove links
* added steps link
* REplacement script for the gnu tool readlink with the -f option
* changed sh to bash
* replaced readlink -e with utils/make_absolute.sh
* added comments and down cased variables
* remove links
* added steps link
* REplacement script for the gnu tool readlink with the -f option
* changed sh to bash
* replaced readlink -e with utils/make_absolute.sh
* added comments and down cased variables
[build] Update README.md / fix ci badge (#1709)
[egs] Adding hub4-ne broadcast spanish recipe (#1665)
[egs] small update to librispeech recipe, RE const-FST.
[src] Fix bug in fstrmymbols RE recent const-fst changes (thanks: Jon Nichols); other cosmetic changes.
[egs] fix problems in multilingual BABEL setup (#1691)
[src] nnet3: fix assertion that shouldn't have been there. Thanks: @vimalmanohar
[src] Fix compiler warnings and work around bug on Windows (#1698)
[src] Adding options to MBR/confidence code (#1696)
[scripts] in subsegment_data_dir.sh, warn if utt2num_frames missing, etc. (#1702)
[egs] babel recipe: check if icu4c is installed (#1697)
[src] nnet3 model-collapsing code, for slight decoding speedup (#1671)
[egs] Rename files with Windows-incompatible names (#1690)
[src] Fix to multiple-fst case of latgen-faster-mapped-parallel (memory bug) (#1688)
[egs] Fix failure in multilingual BABEL recipe (regenerate cmvn.scp) (#1686)
[src,scripts,egs] Backstitch code+scripts, and one experiment, will add more later. (#1605)
See http://www.danielpovey.com/files/2017_nips_backstitch.pdf for details.
See http://www.danielpovey.com/files/2017_nips_backstitch.pdf for details.
[egs] CNN+TDNN+LSTM experiments on AMI (#1685)
[egs,scripts,src] Tune image recognition examples; minor small changes. (#1682)
[src] Fix bug in looped computation (#1673)
[build] when installing sequitur and mmseg, look for lib64 as well (thanks: @akshayc11) (#1677)
[src] fix to gst-plugin/Makefile (remove -lkaldi-thread) (#1680)
[src] Cosmetic fixes to usage messages
[egs] Fix to some --proportional-shrink related example scripts (#1674)
[build] Fix small bug in configure
[scripts] Fix small bug in utils/gen_topo.pl.
[scripts] Add python script to convert nnet2 to nnet3 models (#1611)
[doc] Fix typo (#1669)
[src] nnet3: fix small bug in checking code. Thanks: @maddin2000.
[src] Add #include missing from previous commit
[src] Fix bug in online2-nnet3 decoding RE dropout+batch-norm (thanks: Wonkyum Lee)
[scripts] make errors getting report non-fatal (thx: Miguel Jette); add comment RE dropout proportion
[src,scripts] Use ConstFst or decoding (half the memory; slightly faster). (#1661)
[src] keyword search tools: fix Minimize() call, necessary due to OpenFst upgrade (#1663)
[scripts] do not fail if the ivector extractor belongs to different user (#1662)
[build,scripts] Update scripts that make version info; remove no-op option from script.
[src] minor bugfix in convolutional component (doesn't affect experiments)
[scripts] nnet3 script cleanups; add --proportional-shrink in more places. (#1659)
[scripts] Fix bug in PR #1646 (#1658)
Merge pull request #1547 from kaldi-asr/kaldi_52
This makes kaldi 5.2 the current main-line version.
This makes kaldi 5.2 the current main-line version.
[build] Upgrade .version (this is official start of kaldi 5.2)
Merge pull request #1656 from danpovey/kaldi_52_merge_master
This PR merges recent changes from master.
This PR merges recent changes from master.
Merge remote-tracking branch 'upstream/master' into kaldi_52
[egs] adding proportional-shrink scripts to AMI (#1654)
[scripts] Getting egs, limit max open filehandles to 512 (thanks: gaoxinglong9999)
[src] Fix bug in newly refactored threading code
[src] keyword search: fix invalid assumption about the end states (#1651)
[scripts] Lexicon expansion script -- fix for LM-probs, make it work for non-ASCII langs or langs w. large grapheme set (#1650)
[scripts,egs] minor script fix; fixes in various recipes (#1649)
[egs] updated the LDC web address for wsj0-train-spkrinfo.txt (#1648)
[egs] Ported Fisher spanish recipe to use new LDC dir structure. Other small fixes (#1647)
[scripts] python3 compatibility: decode the output of get_command_stdout if not str (#1646)
[scripts] Fix bugs in automatic report generation for nnet3 training
[src] Use STL thread support library instead of pthread. (#1350)
[src] Add extra diagnostic in nnet3-show-progress
[scripts] Make more informative error in validate_lang.pl when path.sh prints something
[build] Change check_dependencies.sh to not look for yum if apt-get present.
[build] Check python version is 2.7*, not just 2.*.
[src,scripts,egs] Merge master into kaldi_52 (#1628)
* [scripts] nnet1: minor update i-vector and mpe scripts (#1607)
- mpe: backward compatibility is provided
- ivec: the ivectors get stored in binary format (saves space)
* [src] cosmetic change to const-arpa-lm-building code; remove too-general template. (#1610)
* [src,scripts,egs] Segmenting long erroneous recordings (#1167)
This is a solution for creating ASR training data from long recordings with transcription but without segmentation information.
* [egs] thchs30 cmd and stage bug fix (#1619)
* [src] Change to GPU synchronization, for speed (disables GPU stats by default) (#1617)
* [src] Fix template instantiation bug causing failure if DOUBLEPRECISION=1
* [egs,scripts] Updates to BUT-specific cmd.sh settings (affects only Brno team); changes RE verbose level in nnet1 scripts.
* [src] fix a small bug: logging cuda elapsed time (#1623)
* [src,scripts,egs] Add capability for multilingual training with nnet3; babel_multilang example.
* [scripts] Fix some merge problems I noticed on github review.
* [src] fix problem in test code.
* fixed some issues to merge kaldi_52 into master.
* removed add_lda parameter and its dependency.
* [scripts] nnet1: minor update i-vector and mpe scripts (#1607)
- mpe: backward compatibility is provided
- ivec: the ivectors get stored in binary format (saves space)
* [src] cosmetic change to const-arpa-lm-building code; remove too-general template. (#1610)
* [src,scripts,egs] Segmenting long erroneous recordings (#1167)
This is a solution for creating ASR training data from long recordings with transcription but without segmentation information.
* [egs] thchs30 cmd and stage bug fix (#1619)
* [src] Change to GPU synchronization, for speed (disables GPU stats by default) (#1617)
* [src] Fix template instantiation bug causing failure if DOUBLEPRECISION=1
* [egs,scripts] Updates to BUT-specific cmd.sh settings (affects only Brno team); changes RE verbose level in nnet1 scripts.
* [src] fix a small bug: logging cuda elapsed time (#1623)
* [src,scripts,egs] Add capability for multilingual training with nnet3; babel_multilang example.
* [scripts] Fix some merge problems I noticed on github review.
* [src] fix problem in test code.
* fixed some issues to merge kaldi_52 into master.
* removed add_lda parameter and its dependency.
[scripts] fix bugs in align_basis_fmllr.sh [thanks: Filip Jurcicek]
[scripts,egs] Fixes to long-recording segmentation (#1639)
[scripts] Fix steps/cleanup/make_biased_lm_graphs.sh to actually add the top-n-words into the lms (#1637)
[scripts, egs]: fix to egs/lre07/v2 (test was trained on); other updates to LRE scripts.
[src] fix regarding first/last chunk's right-context in chain models (#1632)
This bug-fix only affects BLSTMs that are trained with 'newer' scripts and use the --extra-right-context-final option. (Before the fix, the option was being used in test but not in training, leading to a mismatch).
This bug-fix only affects BLSTMs that are trained with 'newer' scripts and use the --extra-right-context-final option. (Before the fix, the option was being used in test but not in training, leading to a mismatch).
[src] Make parsing error-msg more informative (thanks: Stefan-Adrian Toma)
[egs] Further tuning of --proportional-shrink in WSJ
[scripts] Fix to long-utterance segmentation script (#1631)
[egs] Adding --proportional-shrink example for WSJ.
[doc] small fix RE queue configuration.
[src,egs,scripts] Add SVHN example; fix asymmetry in image-augmentation; minor script changes. (#1630)
[src,scripts,egs] Add capability for multilingual training with nnet3; babel_multilang example.
Changing proportional-shrink from 120 to 150 in mini-librispeech example.
[egs,scripts] Add, and use the --proportional-shrink option (approximates l2 regularization). (#1627)
[src] fix a small bug: logging cuda elapsed time (#1623)
[egs,scripts] Updates to BUT-specific cmd.sh settings (affects only Brno team); changes RE verbose level in nnet1 scripts.
Merging master into kaldi_52 (#1621)
* [scripts] nnet1: minor update i-vector and mpe scripts (#1607)
- mpe: backward compatibility is provided
- ivec: the ivectors get stored in binary format (saves space)
* [src] cosmetic change to const-arpa-lm-building code; remove too-general template. (#1610)
* [src,scripts,egs] Segmenting long erroneous recordings (#1167)
This is a solution for creating ASR training data from long recordings with transcription but without segmentation information.
* [egs] thchs30 cmd and stage bug fix (#1619)
* [src] Change to GPU synchronization, for speed (disables GPU stats by default) (#1617)
* [src] Fix template instantiation bug causing failure if DOUBLEPRECISION=1
* [scripts] nnet1: minor update i-vector and mpe scripts (#1607)
- mpe: backward compatibility is provided
- ivec: the ivectors get stored in binary format (saves space)
* [src] cosmetic change to const-arpa-lm-building code; remove too-general template. (#1610)
* [src,scripts,egs] Segmenting long erroneous recordings (#1167)
This is a solution for creating ASR training data from long recordings with transcription but without segmentation information.
* [egs] thchs30 cmd and stage bug fix (#1619)
* [src] Change to GPU synchronization, for speed (disables GPU stats by default) (#1617)
* [src] Fix template instantiation bug causing failure if DOUBLEPRECISION=1
[src] Fix template instantiation bug causing failure if DOUBLEPRECISION=1
[src] Change to GPU synchronization, for speed (disables GPU stats by default) (#1617)