Sanidha: A Studio Quality Multi-Modal Dataset for Carnatic Music

TL;DR

Sanidha is a new high-quality multi-modal dataset for Carnatic music that enables better source separation, demonstrated by improved model performance and validated through listening tests.

Contribution

Introduces Sanidha, the first open-source, studio-quality Carnatic music dataset with multi-tracks and videos, and shows improved source separation results by fine-tuning Spleeter on it.

Findings

Enhanced SDR performance after fine-tuning on Sanidha

Improved source separation quality in listening tests

Availability of high-quality multi-modal Carnatic music data

Abstract

Music source separation demixes a piece of music into its individual sound sources (vocals, percussion, melodic instruments, etc.), a task with no simple mathematical solution. It requires deep learning methods involving training on large datasets of isolated music stems. The most commonly available datasets are made from commercial Western music, limiting the models' applications to non-Western genres like Carnatic music. Carnatic music is a live tradition, with the available multi-track recordings containing overlapping sounds and bleeds between the sources. This poses a challenge to commercially available source separation models like Spleeter and Hybrid Demucs. In this work, we introduce 'Sanidha', the first open-source novel dataset for Carnatic music, offering studio-quality, multi-track recordings with minimal to no overlap or bleed. Along with the audio files, we provide high-definition videos of the artists' performances. Additionally, we fine-tuned Spleeter, one of the most commonly used source separation models, on our dataset and observed improved SDR performance compared to fine-tuning on a pre-existing Carnatic multi-track dataset. The outputs of the fine-tuned model with 'Sanidha' are evaluated through a listening study.