Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Discophone #6

Merged
merged 30 commits into from
Oct 13, 2020
Merged

Discophone #6

merged 30 commits into from
Oct 13, 2020

Conversation

syfengcuhk
Copy link
Owner

No description provided.

@pzelasko
Add scripts for training phone-level LMs
6c5ed4b
@pzelasko
Some auto reformatting
89da150
@pzelasko
Support for phone/phone-token level transcript preparation for easier…
5b7df13 
… PER scoring; some documentation
@pzelasko
Fix a bug (missing --phones option) that caused most of the training …
569609e 
…text to be converted to <unk>
@pzelasko
Fix LM ppl computation and set default ngram order to 2 (better ppls)
2aee7e6
@pzelasko
Add a scoring script
0ff08a4
@pzelasko
Make scripts executable
2d9ceaf
@pzelasko
Remove comments and add initial script for running decoding with all …
6988f31 
…the GMM models
@pzelasko
Fix the GMM decoding script paths
8e4f998
@pzelasko
Fix a bug - wrong LM used in scoring
8410e44
@pzelasko
Discophone multilingual training from GMM-HMM to TDNNF
77fb52d 
Discophone multilingual training from GMM-HMM to TDNNF
@pzelasko
Merge branch 'discophone' into feature/phone-lms
b47b9bd
@pzelasko
Initially adapted Siyuan's multilingual script to phone LM changes (w…
7f302ec 
…on't work yet)
@pzelasko
Remove setup_languages_multilang.sh which is not needed anymore
45ba8c1
@pzelasko
Add missing copyright
15a8e04
@pzelasko
Making Siyuan's multilingual recipe work with phone-level transcripts…
4687795 
… and LMs
@pzelasko
Best-effort at decoding multilingual GMMs (without having the models …
251172c 
…available yet)
@pzelasko
Make directory combination a separate step
aaaf98e
@pzelasko
Increase the number of jobs used by default in multilingual discophone
26fa32b
@pzelasko
Some fixes when running TDNN recipe
36d6ab6
@pzelasko
Auto-reformatting by PyCharm
949d892
@pzelasko
Training multilingual phone-level LMs
1d93a30 
Training multilingual phone-level LMs
@pzelasko
Initial (untested) implementation of decoding for the DNN multilingua…
def64ba 
…l model
@pzelasko
Add feats + ivector extration for eval data and GPU decoding
04ea0d9
@pzelasko
Fine tuning decoding settings
2723d60
@pzelasko
Decoding for the DNN multilingual model
ef2ca9e 
Decoding for the DNN multilingual model
@pzelasko
Fix decoding number of jobs to the number of speakers to avoid issues
d626057
@pzelasko
Fix decoding number of jobs to the number of speakers to avoid issues
c596b67 
Fix decoding number of jobs to the number of speakers to avoid issues
@pzelasko
Change LM weight range for scoring (more adequate for PER)
12f1c84
@pzelasko
Change LM weight range for scoring (more adequate for PER)
4135d47 
Change LM weight range for scoring (more adequate for PER)
syfengcuhk
syfengcuhk commented Oct 13, 2020
View reviewed changes
Copy link
Owner Author

@syfengcuhk syfengcuhk left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done from piotr's discophone to my discophone_multilang2

@syfengcuhk syfengcuhk merged commit dd579d7 into syfengcuhk:discophone_multilang2 Oct 13, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@syfengcuhk @pzelasko