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.

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

Alignments generated by force_align is different from the original alignment. #60

Open
pluiez opened this issue Apr 14, 2021 · 2 comments

Comments

@pluiez
Copy link

pluiez commented Apr 14, 2021

Hi,

I followed the steps in issue 33 and saved fwd_params, fwd_err, rev_params and rev_err after running fast_align on the training corpus. Later I compared the alignment results by running fast_align from scratch and running force_align.py with the saved params on that training corpus.

I was expecting fast_align produces exactly the same alignments either from scratch or by force_align, but I got different alignments.

I'm sure I followed the correct steps to run these comparisons, so I'm wondering why the alignments are differnt.

@yjlong
Copy link

yjlong commented Jul 9, 2021

hi,
I ran into the same problem. After reading the code, I found that when the ExportToFile method in ttables.h persists parameters, parameters are filtered by threshold.
It may be that some parameters are filtered out and the results are inconsistent, but I don't know if that's the root cause.

@anderleich
Copy link

Any updates on this?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants