Music sample identification improves with real training pairs and new model
Training Music Sample Identification Models on Real Sample Pairs
Sound
Summary
Identifying when a song uses parts of another song is hard, especially because there aren’t many examples to learn from. The authors created a new model called SI Embeddings (SIE) that trains on real examples of sampled music, not just made-up pairs. They found that training with real data and a better model design helps identify samples more accurately. This work provides a clearer way to train future systems that recognize musical samples.
What this means in practice
- •For music streaming platforms: Automatically detect sampled segments in tracks to improve metadata accuracy and royalty tracking.$Commercial implications: Enables better identification of sample usage for licensing and royalty management, improving service offerings.
- •For music producers: Find original sources of samples in music to assist with clearance and creative inspiration.
Authors
R. Oguz Araz, Joan Serrà, Xavier Lizarraga-Seijas, Emilio Molina, Xavier Serra, Yuki Mitsufuji, Dmitry Bogdanov
Abstract
Sample identification (SI) is the task of matching pairs of tracks, where one track is created by musically transforming an element of the other. In the absence of sample annotations at scale, the dominant training paradigm has depended on artificially creating sample pairs. Although a recently released dataset provides annotations of real sample pairs at scale, an effective training recipe is missing. In this work, we present SI Embeddings (SIE), an SI model that achieves state-of-the-art results on three benchmarks, including a large-scale test set. We show that the previous state of the art trained on artificial pairs generalizes only partially to real pairs, and that its training data limits its performance. We also show that real pairs do not fully account for SIE's performance: its architecture and training recipe contribute substantially. We provide the first fully supervised training recipe for real-world SI, establishing a strong foundation for future research in the field.