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

Add CHiME4 recipe #746

Merged
merged 9 commits into from
Nov 19, 2021
Merged

Add CHiME4 recipe #746

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
413aa3d
init chime4 recipe
funcwj Oct 13, 2021
098ca05
update decoding recipe in run,sh
funcwj Oct 28, 2021
80fd5d5
fix tokenizer issue for char unit
funcwj Oct 29, 2021
76956fd
add initial results
funcwj Nov 2, 2021
1c79a0c
finalize recipe
funcwj Nov 10, 2021
20967cf
fix tab issues in run.sh
funcwj Nov 10, 2021
368d8e7
fix trailing whitespace
funcwj Nov 10, 2021
31e8de5
fix comments
funcwj Nov 17, 2021
958621c
commit missing changes
funcwj Nov 18, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
finalize recipe
  • Loading branch information
@funcwj
funcwj committed Nov 10, 2021
commit 1c79a0cd8722f2f4c595b7dc5f980d9dc8247edf
2 changes: 1 addition & 1 deletion examples/chime4/s0/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
* Decoding info: average_num 10

| decoding mode | dt05_real_1ch | dt05_simu_1ch | et05_real_1ch | et05_simu_1ch |
|-----------------|---------------|---------------|---------------|---------------|
|:---------------:|:-------------:|:-------------:|:-------------:|:-------------:|
| ctc_beam_search | 19.06% | 21.17% | 28.39% | 29.16% |
| att_rescoring | 17.92% | 20.22% | 27.40% | 28.25% |
Copy link
Collaborator

Choose a reason for hiding this comment

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

ctc_beam_search -> ctc_prefix_beam_search
att_rescoring -> attention_rescoring
rename so it is consistent with other recipes.


4 changes: 2 additions & 2 deletions examples/chime4/s0/conf/train_conformer.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,7 +63,7 @@ dataset_conf:
sort_size: 500 # sort_size should be less than shuffle_size
batch_conf:
batch_type: 'static' # static or dynamic
batch_size: 12
batch_size: 8

grad_clip: 10
accum_grad: 4
Expand All @@ -72,7 +72,7 @@ log_interval: 200

optim: adam
optim_conf:
lr: 0.001
lr: 0.0005
scheduler: warmuplr # pytorch v1.1.0+ required
scheduler_conf:
warmup_steps: 20000
12 changes: 6 additions & 6 deletions examples/chime4/s0/run.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,15 +104,15 @@ if [ $end -ge 5 ] && [ $beg -le 5 ]; then
dec_dir=$exp_dir/${subdir}_${mode} && mkdir -p $dec_dir
python wenet/bin/recognize.py \
--gpu 0 \
--mode $mode \
--mode $mode \
--config $exp_dir/train.yaml \
--test_data $data_dir/$subdir/data.list \
--checkpoint $exp_dir/avg_${average_num}.pt \
--beam_size 8 \
--test_data $data_dir/$subdir/data.list \
--checkpoint $exp_dir/avg_${average_num}.pt \
--beam_size 8 \
--batch_size 1 \
--dict $dict \
--ctc_weight $ctc_weight \
--result_file $dec_dir/text &
--ctc_weight $ctc_weight \
--result_file $dec_dir/text &
done
wait
done
Expand Down