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

[runtime] context biasing with ac automaton #1937

Merged
merged 1 commit into from
Aug 11, 2023
Merged

[runtime] context biasing with ac automaton #1937

merged 1 commit into from
Aug 11, 2023

Conversation

pengzhendong
Copy link
Member

No description provided.

@pengzhendong pengzhendong marked this pull request as ready for review August 11, 2023 02:09
@pengzhendong
Copy link
Member Author

pengzhendong commented Aug 11, 2023 •
edited
Loading

librispeech:
old context graph: 8.14%(wer) 5.63%(u-wer) 30.19(b-wer)
ac automaton context graph: 8.13%(wer) 5.61%(u-wer) 30.26(b-wer)

aishell:
old context graph: 12.86%(含热词) 5.42%(不含热词)
ac automaton context graph: 12.99%(含热词) 5.42%(不含热词)

@pengzhendong pengzhendong merged commit 886f88f into main Aug 11, 2023
6 checks passed
@pengzhendong pengzhendong deleted the zhendong-hotword branch August 11, 2023 02:28
@yangyyt
Copy link

yangyyt commented Aug 29, 2023

librispeech: old context graph: 8.14%(wer) 5.63%(u-wer) 30.19(b-wer) ac automaton context graph: 8.13%(wer) 5.61%(u-wer) 30.26(b-wer)

aishell: old context graph: 12.86%(含热词) 5.42%(不含热词) ac automaton context graph: 12.99%(含热词) 5.42%(不含热词)

Can you provide some examples for us to try and learn from?

@kaixunhuang0
Copy link
Collaborator

librispeech: old context graph: 8.14%(wer) 5.63%(u-wer) 30.19(b-wer) ac automaton context graph: 8.13%(wer) 5.61%(u-wer) 30.26(b-wer)
aishell: old context graph: 12.86%(含热词) 5.42%(不含热词) ac automaton context graph: 12.99%(含热词) 5.42%(不含热词)

Can you provide some examples for us to try and learn from?

Context list of LibriSpeech can be obtained from here: https://github.com/facebookresearch/fbai-speech/tree/main/is21_deep_bias

@robin1001 robin1001 mentioned this pull request Sep 1, 2023
Closed
@WjMessi1
Copy link

What is the maximum number of hot words now? When I test 125000 hot words, there is an error, can I change the maximum number of hot words?

fclearner
fclearner reviewed Jan 26, 2024
View reviewed changes
runtime/core/kaldi/decoder/lattice-faster-decoder.cc
graph_cost -= context_score;
}
if (context_graph_ != nullptr && changed) {
e_new->val->context_state = tok->context_state;
Copy link
Contributor

@fclearner fclearner Jan 26, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

@pengzhendong 东哥，请教下，processnonemitting这里为什么要把热词增强这部分代码去掉，有点没看懂

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@fclearner fclearner fclearner left review comments

@robin1001 robin1001 robin1001 approved these changes

@kaixunhuang0 kaixunhuang0 Awaiting requested review from kaixunhuang0

@xingchensong xingchensong Awaiting requested review from xingchensong

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
6 participants
@pengzhendong @yangyyt @kaixunhuang0 @WjMessi1 @robin1001 @fclearner