Evernote web clips, 10/1/2020
Old links
How to get timestamps of an audio file using pre-trained model with Kaldi
kaldi/create_segments_from_ctm.pl
Create shared versions of get_ctm_conf.sh, add get_ctm_conf…
Added a new lexicon learning (adaptation) recipe for tedlium
Output-Gate Projected Gated Recurrent Unit for Speech Recognition
Improvements to multi_en tdnn-opgru/lstm recipes #2824
Fix a typo in steps/dict/learn_lexicon_bayesian.sh #3288
Backstitch: Counteracting Finite-Sample Bias via Negative Steps
Acoustic data-driven lexicon learning based on a greedy pronunciation selection framework
Evaluating Natural Language Understanding Services for Conversational Question Answering Systems
[ASR Independent Hybrid Recurrent Neural Network Based Error Correction for Dialog System Applications | Request PDF](https://www.researchgate.net/publication/302498352_ASR_Independent_Hybrid_Recurrent_Neural_Network_Based_Error_Correction_for_Dialog_System_Applications) |
Improving Performance of End-to-End ASR on Numeric Sequences
REFORMER: THE EFFICIENT TRANSFORMER
Reformer: The Efficient Transformer - Pastebin.com
Fix issue in copy_lat_dir.sh affecting combine_lat_dirs.sh
kaldi/compute-gop.cc at b0f6bcb76824fc7ce0dbf8ecaba3467273bba1c7
Make compute-gop work with missing alignments
Add layer for attention with bypass by danpovey
Handwriting recognition and language modeling with MXNet Gluon