Codestin Search App

satvik-dixit · 2024-04-03T20:27:40Z

What?

This is a new recipe for preparing the Makerere Radio Speech Corpus dataset and training automatic speech recognition models on it.

Details about the dataset:

The Makerere Radio Speech Corpus is a dataset in the Luganda language. It includes 20 hours of human-transcribed radio speech.

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 16.28%. Comparing base (844292e) to head (56a498c).
Report is 119 commits behind head on master.

❗ Current head 56a498c differs from pull request most recent head ac71eb4. Consider uploading reports for the commit ac71eb4 to get more accurate results

Additional details and impacted files

@@             Coverage Diff             @@
##           master    #5730       +/-   ##
===========================================
+ Coverage        0   16.28%   +16.28%     
===========================================
  Files           0      767      +767     
  Lines           0    70337    +70337     
===========================================
+ Hits            0    11453    +11453     
- Misses          0    58884    +58884

Flag	Coverage Δ
test_python_espnetez	`14.22% <ø> (?)`
test_utils	`20.92% <ø> (?)`

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

satvik-dixit · 2024-04-04T02:00:27Z

The error ImportError: cannot import name 'kaiser' from 'scipy.signal' (/home/runner/work/espnet/espnet/tools/venv/lib/python3.10/site-packages/scipy/signal/__init__.py) seems to be happening because kaiser should be imported from scipy.signal.windows according to the latest scipy version

sw005320 · 2024-04-05T14:58:42Z

@jctian98, can you review this PR?

jctian98 · 2024-04-06T00:04:35Z

@satvik-dixit Thanks for making this PR. I've left some comments above.
Additionally,
(1) You split the train/valid/test sets by yourself. Could you confirm that the official release of this dataset doesn't have that split? If they already have that split, we'd better follow it.
(2) You split the train_dev/test in the Python script; and then train/dev in the shell script. Could you put this logic together? either in Python or Shell should be fine.
(3) Seems the current recipe cannot pass the CI test due to the scipy version issue. After that issue is solved, please make sure to pass other CI tests.

sw005320 · 2024-04-08T23:23:07Z

@satvik-dixit Thanks for making this PR. I've left some comments above. Additionally,
(1) You split the train/valid/test sets by yourself. Could you confirm that the official release of this dataset doesn't have that split? If they already have that split, we'd better follow it.
(2) You split the train_dev/test in the Python script; and then train/dev in the shell script. Could you put this logic together? either in Python or Shell should be fine.

@satvik-dixit, please respond to them.

(3) Seems the current recipe cannot pass the CI test due to the scipy version issue. After that issue is solved, please make sure to pass other CI tests.

This is fixed.

satvik-dixit · 2024-04-09T00:37:36Z

That's right, the official release of this dataset doesn't have that split. Here are more about the dataset for reference:
Details about the dataset: https://zenodo.org/records/5855017
Paper on the dataset: https://arxiv.org/abs/2206.09790
Yes, I can include all of it in the Python script (data_prep.py)
Thanks Shinji

sw005320 · 2024-04-09T00:43:31Z

egs2/makerere/asr1/cmd.sh

+
+
+# Select the backend used by run.sh from "local", "stdout", "sge", "slurm", or "ssh"
+cmd_backend='slurm'


Please do not change some environment related configs.
Please submit them as they are (e.g., cmd.sh and slurm.conf)

I have made the suggested changes in cmd.sh and slurm.conf

sw005320 · 2024-04-09T00:43:56Z

egs2/makerere/asr1/README.md

+- espnet version: `espnet 202402`
+- pytorch version: `pytorch 2.0.1`
+- Git hash: `eed7751c910977290ef9a177ea0942a0e3c2fd35`
+  - Commit date: `Mon Mar 25 18:26:50 2024 +0000`


Please upload the model to the HF hub and add a link here

I have uploaded the model to the HF hub and added a link to this file. Here's the link: https://huggingface.co/satvik-dixit/asr_makerere.

jctian98 · 2024-04-15T18:57:18Z

Hi @satvik-dixit , is there anything I can help?

for more information, see https://pre-commit.ci

jctian98 · 2024-04-30T14:06:27Z

@satvik-dixit Thanks for the current progress. Shinji and I left some comments before. Please also kindly address those comments :)

jctian98 · 2024-04-30T14:09:17Z

@sw005320 It seems some CI tests fail at the Espnet/Python Installation stage, which should not be @satvik-dixit 's fault. Is there a good way to deal with it?

sw005320 · 2024-04-30T20:24:29Z

LGTM.
I'll merge it with some checks.

sw005320 · 2024-04-30T20:25:21Z

Please add a corpus description to https://github.com/espnet/espnet/blob/master/egs2/README.md

satvik-dixit · 2024-05-01T05:44:31Z

Added the corpus description to readme (https://github.com/espnet/espnet/blob/master/egs2/README.md)

sw005320 · 2024-05-01T11:31:32Z

Thanks a lot!

Satvik Dixit and others added 2 commits April 3, 2024 12:20

makerere asr

f2bdb6b

Rename RESULTS.md to README.md

dd346a1

mergify bot added ESPnet2 README labels Apr 3, 2024

[pre-commit.ci] auto fixes from pre-commit.com hooks

a4e4d3f

for more information, see https://pre-commit.ci

satvik-dixit added 2 commits April 3, 2024 18:03

Update data.sh

6280006

Update run.sh

9d82f88

sw005320 added Recipe ASR Automatic speech recogntion labels Apr 5, 2024

sw005320 added this to the v.202405 milestone Apr 5, 2024

Merge branch 'master' into tmp

d1198b4

sw005320 reviewed Apr 9, 2024

View reviewed changes

satvik-dixit and others added 9 commits April 24, 2024 19:38

Update README.md

20dfd96

Update cmd.sh

2286594

Update slurm.conf

859ecaf

Update data_prep.py

fe5b36f

[pre-commit.ci] auto fixes from pre-commit.com hooks

3bf4d86

for more information, see https://pre-commit.ci

Update data_prep.py

e18e4e4

Update data.sh

9c30d1d

[pre-commit.ci] auto fixes from pre-commit.com hooks

56a498c

for more information, see https://pre-commit.ci

Update data.sh

a7df460

Update README.md

ac71eb4

sw005320 merged commit 543f488 into espnet:master May 1, 2024



		# Select the backend used by run.sh from "local", "stdout", "sge", "slurm", or "ssh"
		cmd_backend='slurm'

Conversation

satvik-dixit commented Apr 3, 2024

What?

Details about the dataset:

Related Links

Uh oh!

codecov bot commented Apr 3, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

satvik-dixit commented Apr 4, 2024

Uh oh!

sw005320 commented Apr 5, 2024

Uh oh!

jctian98 commented Apr 6, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sw005320 commented Apr 8, 2024

Uh oh!

satvik-dixit commented Apr 9, 2024

Uh oh!

sw005320 Apr 9, 2024

Choose a reason for hiding this comment

Uh oh!

satvik-dixit Apr 30, 2024

Choose a reason for hiding this comment

Uh oh!

sw005320 Apr 9, 2024

Choose a reason for hiding this comment

Uh oh!

satvik-dixit Apr 30, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jctian98 commented Apr 15, 2024

Uh oh!

jctian98 commented Apr 30, 2024

Uh oh!

jctian98 commented Apr 30, 2024

Uh oh!

sw005320 commented Apr 30, 2024

Uh oh!

sw005320 commented Apr 30, 2024

Uh oh!

satvik-dixit commented May 1, 2024

Uh oh!

sw005320 commented May 1, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

codecov bot commented Apr 3, 2024 •

edited

Loading

jctian98 commented Apr 6, 2024 •

edited

Loading

satvik-dixit Apr 30, 2024 •

edited

Loading