Skip to content

daewoung/RefWhisper

 
 

Repository files navigation

RefWhisper

This is a repository for our paper on DLfM 2023, "Aligning Incomplete Lyrics of Korean Folk Song Dataset using Whisper".

The original dataset is available at here

Requirements

We recommend pipenv for installing the requirements. The requirements are given in Pipfile.

Install pipenv and requirements

pip install pipenv
pipenv install
pipenv shell

Model

The pre-trained model is provided in OneDrive

Anthology of Korean Traditional Folksongs

Official website

The metadata we collected from the website is available in Finding Tori metadata.csv, including the url for each song. You can download the audio files from the website.

Lyric Transcribed Result

The transcribed result is provided in transcription_result_csvs.tar.gz. You can untar the file by running tar -xvzf transcription_result_csvs.tar.gz.

The transcription is given in three files:

  • {song_id}_transcribed.txt
    • The transcribed lyric of the song
  • {song_id}_word_align.csv
    • Word-level alignment between the transcribed lyric and the reference lyric
  • {song_id}_ref_align.csv
    • Line-level alignment between the reference lyric and the transcribed lyric
    • In this file, there are two alignment.
        1. Serial alignment between the transcribed and the reference. In this alignment, the two lyrics are aligned in serial order from the beginning to the end.
        1. Open-beginning/open-end alignment between each line of the reference lyric and the entire corresponding transcription. This trys to find the optimal alignment for that specific lyric line, not considering the alignment for entire lyrics.

Citation

If you use this code or the dataset, please cite our paper.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%