Codified audio language modeling learns useful representations for music information retrieval
arXiv:2107.05677
Abstract
We demonstrate that language models pre-trained on codified (discretely-encoded) music audio learn representations that are useful for downstream MIR tasks. Specifically, we explore representations from Jukebox (Dhariwal et al. 2020): a music generation system containing a language model trained on codified audio from 1M songs. To determine if Jukebox's representations contain useful information for MIR, we use them as input features to train shallow models on several MIR tasks. Relative to representations from conventional MIR models which are pre-trained on tagging, we find that using representations from Jukebox as input features yields 30% stronger performance on average across four MIR tasks: tagging, genre classification, emotion recognition, and key detection. For key detection, we observe that representations from Jukebox are considerably stronger than those from models pre-trained on tagging, suggesting that pre-training via codified audio language modeling may address blind spots in conventional approaches. We interpret the strength of Jukebox's representations as evidence that modeling audio instead of tags provides richer representations for MIR.
To appear in the proceedings of ISMIR 2021
References in corpus (8)
- Scaling Laws for Neural Language Models
- Generating Long Sequences with Sparse Transformers
- SampleRNN: An Unconditional End-to-End Neural Audio Generation Model
- Jukebox: A Generative Model for Music
- Evaluation of CNN-based Automatic Music Tagging Models
- Enriched Music Representations with Multiple Cross-modal Contrastive Learning
- Comparison and Analysis of Deep Audio Embeddings for Music Emotion Recognition
- Representation Learning of Music Using Artist, Album, and Track Information