Skip to content
Snippets Groups Projects
Select Git revision
  • master default protected
1 result
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.015Oct1413121110875432130Sep29282726252322212019181716151413141310876542131Aug30292728272826242322212019181716171614121110111098754343231Jul302928272827262524232221201920191716151413108643129Jun28262526252322212019171413121110898764343131May3029282726252423201918191819181716151615161513121112111097654545432130Apr292827262524222120212019181716151413Fixing deletion errors in text matrix parsing code (#1121)Remove orphaned script reduce_data_dir_by_reclist.sh'"--extra-word-disambig-syms" for prepare_lang.sh (#1113)revise the chain lstm scripts in a new branch #1100 (#1101)resolves #1108 (#1116)Fix to assertion in online-endpoint.cc to stop spurious assert failure. Thanks to Ognjen Todic for reporting the bug.Fix to steps/decode_combine.sh to remove a command that was added in the tombstone branch and somehow merged into master.. probably never was helpful even therecircumventing compilation issues on macos (#1114)Unk model (#1058)Fixing --zero-if-disjoint -> --drop-frames in a few scripts where it had not been changed; a couple of other minor fixes.fix out-of-bound global read bug found by cuda-memcheck in cu-vector-test (#1103)raw_python_script: Changing module importsResample the RIRs and noises to suit data of different sampling rate (#1099)Clarify a couple of usage messages; fix a nnet3 Properties() flag that was causing test failure on mac.raw_python_script: Missed variable renamesraw_python_script: Addressed comments and made changesaspire : refactoring the decode scripts, and adding examples for online decoding (#1075)Various SID Fixes (#1093)change before sort in wsj_train_rnnlms.sh to avoid filling up /tmp/ (#1095)Merging from masterretain only valid options for rnnlm and faster-rnnlm in wsj_train_rnnlms.sh (#1094)added the latest cleanup code and left-biphone models to the librispeech recipe (#991)New augmentation recipe for reverberant speech (#1076)Nnet1 dropout ivec (#1090)Added DropoutComponent in nnet3 (#1032)Merge pull request #1079 from jtrmal/srilmfixtrying to fix some issues under macosMerge pull request #1038 from jtrmal/openfst_patches2updating the fst patchesnnet3 reporting : Added code to summarize the timing info from nnet3*train --computation.debug=true outputs (#1088)nnet3: added command execution time computation code (#1087)Fix recently introduced bug in tedlium/s5_r2/local/ted_train_lm.shWIP fixing the cantab lm links (#1086)fix bug in ted_train_lm.sh introduced by previous commitspecify lm_dir for train_lm.py in ted_train_lm.sh (#1085)raw_python_script: Refactoring RNN and DNN scriptscosmetic fix: removed unused --chain.ngram-order option in chain/train.py (#1084)minor bugfix nnet3 python training scripts. Does not affect results (#1083)In clean_and_segment_data.sh use provided $cmd, not $decode_cmd (#1082)Remove filter_cmd from the ctm generation part (#1081)