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

Add a new EGS2 recipe 'reazonspeech' #4885

Merged
merged 4 commits into from
Feb 1, 2023
Merged

Add a new EGS2 recipe 'reazonspeech' #4885

merged 4 commits into from
Feb 1, 2023

Commits on Feb 1, 2023

  1. Add a new EGS2 recipe 'reazonspeech' 

    ReazonSpeech is >19000h Japanese corpus collected from TV programs.
This adds ASR recipe that trains Conformer model on it.

 * The dataset is automatically downloaded from Hugging Face using
   `datasets.load_download()`

 * The training recipe is almost the same as CSJ, except for minor
   parameter tweakings (to optimize it for our machine spec)

The pre-trained model is available on Hugging Face:

https://huggingface.co/reazon-research/reazonspeech-espnet-v1

Signed-off-by: Fujimoto Seiji <fujimoto@clear-code.com>
    @fujimotos
    Fujimoto Seiji authored and fujimotos committed Feb 1, 2023
    Configuration menu
    Copy the full SHA
    105e532 View commit details
    Browse the repository at this point in the history

  2. Make use of 'REAZONSPEECH' variable defined in db.sh 

    Based on a feedback on PR#4885. Instead of hard-cording the data path,
make it adjustable by db.sh

Signed-off-by: Fujimoto Seiji <fujimoto@clear-code.com>
    @fujimotos
    Fujimoto Seiji authored and fujimotos committed Feb 1, 2023
    Configuration menu
    Copy the full SHA
    9003799 View commit details
    Browse the repository at this point in the history

  3. Make use of installation scripts in 'tools/intstall' 

    Based on a feedback on PR#4885. Instead of letting users find out
how to install the required modules by themselves, let's automate
those steps.

Signed-off-by: Fujimoto Seiji <fujimoto@clear-code.com>
    @fujimotos
    Fujimoto Seiji authored and fujimotos committed Feb 1, 2023
    Configuration menu
    Copy the full SHA
    26210a2 View commit details
    Browse the repository at this point in the history

  4. Fix CI failure due to linter warnings 

    This resolves coding-style warnings emitted by linters:

 * Quote the `test_sets` variable to resolve a shellcheck worning.
 * Apply isort and black to reformat `local/data.py`.

Signed-off-by: Fujimoto Seiji <fujimoto@ceptord.net>
    @fujimotos
    fujimotos committed Feb 1, 2023
    Configuration menu
    Copy the full SHA
    a92301e View commit details
    Browse the repository at this point in the history