Lyrics-Transcription-Resources

A comprehensive list of open source resources for lyrics transcription task.

Datasets

• MUSDB - Suitable for music source separation.
• DAMP-VSEP - Vocal Separation.
• DAMP-MVP - Smule Multilingual Vocal Performance 300x30x2.
• DAMP-Vocal Performances - Smule DAMP-Multiple Songs.
• DAMP-VP1K - Subset of DAMP-Vocal Performances.
• DAMP-VPB - Smule Vocal Performances Balanced.

Pretrained Models

Music Source Separation

Trained on MUSDB

• Meta-TasNet - Repository and COLAB demo. Pytorch.
• Demucs - Repository and pretrained models. Pytorch.
• ConvTasNet - Repository and pretrained models. Pytorch.
• Spleeter by deezer - Repository and COLAB demo. Tensorflow.
• Open-Unmix - Repository and COLAB demo. Pytorch.