[WIP] Iam ocr#14

Open

ChunChiehChang wants to merge 225 commits into

hhadian:ocr2from

ChunChiehChang:iam_ocr

Commits on Jul 21, 2017

Commits on Jul 24, 2017

Merge remote-tracking branch 'aarora8/IAM_eng' into iam_ocr
Chun-Chieh Chang
committed
added code to perform splits
Chun-Chieh Chang
committed

Commits on Jul 25, 2017

adding code for creating train,validation and test sets and also create spk2utt file
aarora8
committed
changed so train, test, valid examples are separated
Chun-Chieh Chang
committed
merged aarora8/IAM_eng and fixed some minor things
Chun-Chieh Chang
committed
accidently split spk2utt incorrectly
Chun-Chieh Chang
committed
adding changes for setting desired variance floor value
aarora8
committed

Commits on Jul 26, 2017

Commits on Jul 27, 2017

WIP added scripts to prepare dict and lexicon
Chun-Chieh Chang
committed
fixed bug where due to rounding errors the images won't be resized exactly
Chun-Chieh Chang
committed
fixing bug in initializing variance floor vector
aarora8
committed

Commits on Jul 28, 2017

Commits on Jul 31, 2017

added decoding to run.sh and the necessary files for decoding into local
Chun-Chieh Chang
committed

Commits on Aug 1, 2017

Commits on Aug 10, 2017

added scripts to use characters instead of words for models. Also renamed run.sh to run{1-6}.sh
Chun-Chieh Chang
committed
added in comments the ability to use individual words in training/testing instead of lines of text
Chun-Chieh Chang
committed

Commits on Aug 11, 2017

changes to creating lexicon
Chun-Chieh Chang
committed

Commits on Aug 15, 2017

added scripts to ensure lm decodes one word without uniform distribution.
Chun-Chieh Chang
committed
modified script to run on individual words instead of lines of text
Chun-Chieh Chang
committed
enabled silence models to have two states
Chun-Chieh Chang
committed

Commits on Sep 9, 2017

code for handwriting word recognition for database IAM
aarora8
committed

Commits on Sep 11, 2017

fixing reset
aarora8
committed

Commits on Sep 13, 2017

Commits on Sep 21, 2017

adding modification for getting vocab from corpus, removing comments from compare_wer, removing affix from run.sh script
aarora8
committed

Commits on Oct 2, 2017

Commits on Oct 3, 2017

Commits on Oct 7, 2017

adding modifications for including validation data in training
aarora8
committed
bug fix thanks Yiwen Shao
aarora8
committed
removed variance floor as it did not seem to help
Chun-Chieh Chang
committed
bug fix
aarora8
committed
removed unecessary run file
Chun-Chieh Chang
committed
modified score to use steps/scoring/scoring_kaldi_wer.sh
Chun-Chieh Chang
committed
updated scripts to most recent version. Finished the run.sh
Chun-Chieh Chang
committed
Merge remote-tracking branch 'aarora8/iam' into iam_ocr
Chun-Chieh Chang
committed
bug fix
aarora8
committed
removing .txt
aarora8
committed
Merge remote-tracking branch 'aarora8/iam' into iam_ocr
Chun-Chieh Chang
committed

Commits on Oct 30, 2017

Merge remote-tracking branch 'upstream/master' into iam_ocr
Chun-Chieh Chang
committed
OCR: Add IAM corpus with unk decoding support (#3)

aarora8
authored and
hhadian
committed

Commits on Nov 2, 2017

Add a new English OCR database 'UW3'

ChunChiehChang
authored and
hhadian
committed

Commits on Nov 15, 2017

Some minor fixes re IAM corpus
hhadian
committed

Commits on Nov 20, 2017

Fix an issue in IAM chain recipes + add a new recipe (#6)

aarora8
authored and
hhadian
committed

Commits on Nov 29, 2017

removed part to prepare data for validation sets and added CER to score
Chun-Chieh Chang
committed
merged branch from hhadian/ocr and resolved some conflicts
Chun-Chieh Chang
committed

Commits on Dec 22, 2017

Some fixes based on the pull request review

aarora8
authored and
hhadian
committed
Various fixes + cleaning on IAM
hhadian
committed

Commits on Dec 31, 2017

Add UW3 OCR database

ChunChiehChang
authored and
hhadian
committed

Commits on Jan 1, 2018

merged from hhadian ocr
Chun-Chieh Chang
committed
forgot to remove last of variance floor option
Chun-Chieh Chang
committed

Commits on Jan 10, 2018

different number of states for punctuations
Chun-Chieh Chang
committed

Commits on Jan 12, 2018

removed commented out code
Chun-Chieh Chang
committed
adding updated results
Chun-Chieh Chang
committed
removed unnecessary folder
Chun-Chieh Chang
committed
removed some changes from variance floor option that is now removed
Chun-Chieh Chang
committed

Commits on Jan 19, 2018

moved s5 to v1
Chun-Chieh Chang
committed
merge master branch to get hhadian commits
Chun-Chieh Chang
committed
moved s5 to wrong location
Chun-Chieh Chang
committed
removed unused files
Chun-Chieh Chang
committed
changed run to use local prepare_lang.sh
Chun-Chieh Chang
committed

Commits on Feb 23, 2018

[egs] Fix script bug (in error-checking statement) (#2239)

mpuels
authored and
danpovey
committed
[scripts] Fix typo (#2237)

mpuels
authored and
danpovey
committed
[scripts] Fix typo in comment (#2238)

mpuels
authored and
danpovey
committed

Commits on Feb 24, 2018

[scripts] Bug-fix in analyze_phone_length_stats.py (thanks: @stanleyguan)
danpovey
committed

Commits on Feb 26, 2018

Merge remote-tracking branch 'origin/master' into iam_ocr
Chun-Chieh Chang
committed

Commits on Feb 27, 2018

Commits on Feb 28, 2018

Commits on Mar 1, 2018

Commits on Mar 2, 2018

[src] Fix to cu-rand.cc RE large matrix sizes. (#2249)
danpovey
authored
added unk and added wellington corpus. removed LOB because IAM and LOB overlap
Chun-Chieh Chang
committed
forgot to add some stuff for unk and also for different topo for punctuations
Chun-Chieh Chang
committed

Commits on Mar 3, 2018

[build]: Dockerfile for kaldi installation (#2246)

sih4sing5hong5
authored and
danpovey
committed
[src] Fix macro redefinition problems in tensorflow rnnlm. (#2250)

dogancan
authored and
danpovey
committed
[src,scripts,egs] Enable backstich training for RNNLM; add examples (#2096)

freewym
authored and
danpovey
committed

Commits on Mar 4, 2018

[src] Fixed bug in GeneralDropoutComponent when block-dim set (#2252)

freewym
authored and
danpovey
committed
[egs] Port factorized TDNN script to librispeech (#2233)

davidavdav
authored and
danpovey
committed

Commits on Mar 5, 2018

[scripts] python3 compatibility: change / to // (#2253)

xiaohui-zhang
authored and
danpovey
committed

Commits on Mar 6, 2018

[src] Fix CPU swap methods of CuVector and CuPackedMatrix (#2254)

dogancan
authored and
danpovey
committed
[scripts] Add example of accessing nnet3 model from python (#2255)
danpovey
authored
[src,scripts,egs] End-to-end chain training (#2072)

hhadian
authored and
danpovey
committed

Commits on Mar 8, 2018

Commits on Mar 9, 2018

Commits on Mar 12, 2018

[src] bug-fixes for end2end chain code (#2270)

hhadian
authored and
danpovey
committed
[src] Fix subsampling factor in nnet3 egs (#2265)

vimalmanohar
authored and
danpovey
committed

Commits on Mar 13, 2018

[egs] Add chime5 baseline recipe (#2262)

jtrmal
authored and
danpovey
committed
[scripts] Add delta_opts for train_mono.sh (#2221)

svecjan
authored and
danpovey
committed

Commits on Mar 14, 2018

[egs] fix generate_plots.py to plot deriv-avg and oderiv-rms (#2274)

qbetterk
authored and
danpovey
committed
[tools] fix beamformit installation (#2276)

sw005320
authored and
jtrmal
committed
[scripts] Bug-fix in convert_data_dir_to_whole.sh (thx: @olix20) (#2277)

vimalmanohar
authored and
danpovey
committed

Commits on Mar 15, 2018

Commits on Mar 16, 2018

Commits on Mar 17, 2018

[src] Add a nnet3 optimization that tries to replace commands ending in Multi with other commands. (#2229)
danpovey
authored

Commits on Mar 18, 2018

[scripts] Fixes to segmentation scripts (#2273)

vimalmanohar
authored and
danpovey
committed
[scripts] Add support for --use-gpu=wait to nnet3 scripts (#2289)

hhadian
authored and
danpovey
committed
[egs] Add improved TDNN-F example for Switchboard. (#2291)
danpovey
authored

Commits on Mar 19, 2018

Commits on Mar 20, 2018

Merge remote-tracking branch 'hhadian/e2e_ocr' into iam_ocr
Chun-Chieh Chang
committed
[src] Enable reading wavs produced on iPhones with extra RIFF tags. (#2293)

KarelVesely84
authored and
danpovey
committed

Commits on Mar 21, 2018

[egs] fixing LM issue in CHiME5 (#2298)

jtrmal
authored and
danpovey
committed

Commits on Mar 23, 2018

[src] Speedup and code simplification for chain supervision merging (etc.) (#2302)

chenzhehuai
authored and
danpovey
committed

Commits on Mar 24, 2018

Commits on Mar 25, 2018

[scripts,egs] Fix to perturb_speed_to_allowed_lengths.py; egs fix (thanks: @calderma) (#2307)

hhadian
authored and
danpovey
committed

Commits on Mar 26, 2018

[windows] fix for compiling on Windows VS2017 (15.5.2) (#2310)

saillabs1
authored and
jtrmal
committed

Commits on Mar 27, 2018

Commits on Mar 28, 2018

Commits on Mar 30, 2018

[build] Updates to Sequitur installation script (for compatibility with new setuptools) (#2323)

jtrmal
authored and
danpovey
committed

Commits on Apr 4, 2018

Commits on Apr 5, 2018

Commits on Apr 6, 2018

[scripts] Make nnet3 logging output look more like chain logging output (#2338)

Ore-an
authored and
danpovey
committed
[scripts] Fix bug in chain/get_egs.sh affecting semi-supervised training (#2339)

vimalmanohar
authored and
danpovey
committed

Commits on Apr 7, 2018

[egs] IAM script fixes using the right lang for training/lattice generation + cleaning (#2340)

hhadian
authored and
danpovey
committed

Commits on Apr 8, 2018

[egs] IAM: add chain recipe starting with e2e alignments and tree (#2343)

hhadian
authored and
danpovey
committed

Commits on Apr 9, 2018

[egs] Fix scoring issue in IAM (#2346)

hhadian
authored and
danpovey
committed

Commits on Apr 10, 2018

[src] Revert KALDI_MEMALIGN->aligned_alloc change from #2304, fixes #2347 (#2348)
danpovey
authored

Commits on Apr 11, 2018

Commits on Apr 12, 2018

Commits on Apr 13, 2018

added decode_gmm option. Removed unecessary file
Chun-Chieh Chang
committed
[egs] Adding Wellington corpus for LM in IAM OCR (#2354)

ChunChiehChang
authored and
danpovey
committed
[scripts] Make beam configurable, +default smaller, in pruned RNNLM rescoring (#2325)

hainan-xv
authored and
danpovey
committed

Commits on Apr 14, 2018

[src] small fix to comment (#2357)

wan guanglu
authored and
danpovey
committed
[egs] A fix to second-stage chain recipes in IAM and lang fixes in UW3 (#2358)

hhadian
authored and
danpovey
committed

Commits on Apr 16, 2018

[egs] Fix script bug in IAM setup

adelra
authored and
danpovey
committed
[src] Cosmetic code fixes in mel-computations.cc (#2359)

wan guanglu
authored and
danpovey
committed

Commits on Apr 18, 2018

added option to include/not include wellington corpus
Chun-Chieh Chang
committed
fixed bug in train_lm
Chun-Chieh Chang
committed

Commits on Apr 22, 2018

[src] Comment fix to transition-model.h (#2378)

wan guanglu
authored and
danpovey
committed

Commits on Apr 23, 2018

cleaned up and fixed some regex
Chun-Chieh Chang
committed
Merge branch 'iam_ocr' of github.com:ChunChiehChang/kaldi into iam_ocr
Chun-Chieh Chang
committed

Commits on Apr 24, 2018

Commits on Apr 25, 2018

Merge remote-tracking branch 'origin/master' into iam_ocr
Chun-Chieh Chang
committed
[src] Add a cast to prevent errors from some compilers (#2384)

danijel3
authored and
danpovey
committed
[egs] Semisupervised training recipes: bug fix regarding context info (#2385)

vimalmanohar
authored and
danpovey
committed
[egs] Remove accidentally committed "exit 0" in WSJ run.sh (#2386)

Allan Peng
authored and
danpovey
committed

Commits on Apr 26, 2018

[src,scripts] Cosmetic fixes.
danpovey
committed
[scripts] add wordlists mismatch test for rnnlm rescoring (#2387)

hainan-xv
authored and
danpovey
committed

Commits on Apr 27, 2018

[scripts] Clarify usage of diarization make_rttm.py (#2390)

mmaciej2
authored and
danpovey
committed
[egs] Add speaker verification recipe for the VoxCeleb2 corpus (#2380)

entn-at
authored and
danpovey
committed

Commits on Apr 29, 2018

[src] Remove un-needed code. Thanks: anand@sayint.ai. (#2393)

huangruizhe
authored and
danpovey
committed
[egs] Update nnet3/chain scripts for HKUST example (#2368)

Xuechen Liu
authored and
danpovey
committed

Commits on Apr 30, 2018

updating results on top of scripts
Chun-Chieh Chang
committed

Commits on May 1, 2018

[scripts] Fix in steps/nnet3/get_egs.sh to avoid failure due to broken pipe (#2397)

freewym
authored and
danpovey
committed
[egs] Cleanup and update results in IAM OCR scripts (#2395)

ChunChiehChang
authored and
danpovey
committed

Commits on May 2, 2018

[scripts] rnnlm scripts: ignore first iteration while looking for the best model (#2399)

nshmyrev
authored and
danpovey
committed

Commits on May 3, 2018

[scripts] Fix path.sh bug in vad_to_segments.sh (#2402)

nshmyrev
authored and
danpovey
committed
[src] Remove unnecessary file chain/context-dep-topology.h (#2401)

huangruizhe
authored and
danpovey
committed

Commits on May 4, 2018

[src,scripts,egs] x-vectors for diarization (#2391)

david-ryan-snyder
authored and
danpovey
committed

Commits on May 5, 2018

[egs,scripts,src] Add IFN/ENIT dataset Handwriting Recognition Example (#2257)

babakrkb
authored and
danpovey
committed

Commits on May 6, 2018

[src] Some small cosmetic changes (#2404)
danpovey
authored

Commits on May 7, 2018

[egs] Fix python error by moving __future__ import to top (#2405)

entn-at
authored and
danpovey
committed
[src] make e2e/"unconstrained" numerator computation faster (#2392)

jtrmal
authored and
danpovey
committed

Commits on May 8, 2018

[src] Speed up VectorBase<Real>::Sum() by using BLAS (#2394)

galv
authored and
danpovey
committed

Commits on May 9, 2018

[build] update dockerfile for Ubuntu and cuda (#2408)

sih4sing5hong5
authored and
danpovey
committed

Commits on May 10, 2018

Merge remote-tracking branch 'origin/master' into iam_ocr
Chun-Chieh Chang
committed