Detailed captions that accurately reflect the characteristics of a music piece can enrich music databases and drive forward research in music AI. This paper introduces a multi-task music captioning model, SonicVerse, that integrates caption generation with auxiliary music feature detection tasks such as key detection, vocals detection, and more, so as to directly capture both low-level acoustic details as well as high-level musical attributes. The key contribution is a projection-based architecture that transforms audio input into language tokens, while simultaneously detecting music features through dedicated auxiliary heads. The outputs of these heads are also projected into language tokens, to enhance the captioning input. This framework not only produces rich, descriptive captions for short music fragments but also directly enables the generation of detailed time-informed descriptions for longer music pieces, by chaining the outputs using a large-language model. To train the model, we extended the MusicBench dataset by annotating it with music features using MIRFLEX, a modular music feature extractor, resulting in paired audio, captions and music feature data. Experimental results show that incorporating features in this way improves the quality and detail of the generated captions.
This study evaluates the baseline capabilities of Large Language Models (LLMs) like ChatGPT, Claude, and Gemini to learn concepts in music theory through in-context learning and chain-of-thought prompting. Using carefully designed prompts (in-context learning) and step-by-step worked examples (chain-of-thought prompting), we explore how LLMs can be taught increasingly complex material and how pedagogical strategies for human learners translate to educating machines. Performance is evaluated using questions from an official Canadian Royal Conservatory of Music (RCM) Level 6 examination, which covers a comprehensive range of topics, including interval and chord identification, key detection, cadence classification, and metrical analysis. Additionally, we evaluate the suitability of various music encoding formats for these tasks (ABC, Humdrum, MEI, MusicXML). All experiments were run both with and without contextual prompts. Results indicate that without context, ChatGPT with MEI performs the best at 52%, while with context, Claude with MEI performs the best at 75%. Future work will further refine prompts and expand to cover more advanced music theory concepts. This research contributes to the broader understanding of teaching LLMs and has applications for educators, students, and developers of AI music tools alike.
We present Myna, a simple yet effective approach for self-supervised musical representation learning. Built on a contrastive learning framework, Myna introduces two key innovations: (1) the use of a Vision Transformer (ViT) on mel-spectrograms as the backbone and (2) a novel data augmentation strategy, token masking, that masks 90 percent of spectrogram tokens. These innovations deliver both effectiveness and efficiency: (i) Token masking enables a significant increase in per-GPU batch size, from 48 or 120 in prior methods (CLMR, MULE) to 4096. (ii) By avoiding traditional augmentations, Myna retains pitch sensitivity, enhancing performance in tasks like key detection. (iii) The use of vertical patches allows the model to better capture critical features for key detection. Our hybrid model, Myna-22M-Hybrid, processes both 16x16 and 128x2 patches, achieving state-of-the-art results. Trained on a single GPU, it outperforms MULE (62M) on average and rivals MERT-95M, which was trained on 16 and 64 GPUs, respectively. Additionally, it surpasses MERT-95M-public, establishing itself as the best-performing model trained on publicly available data. We release our code and models to promote reproducibility and facilitate future research.
Recent years have witnessed the success of foundation models pre-trained with self-supervised learning (SSL) in various music informatics understanding tasks, including music tagging, instrument classification, key detection, and more. In this paper, we propose a self-supervised music representation learning model for music understanding. Distinguished from previous studies adopting random projection or existing neural codec, the proposed model, named MuQ, is trained to predict tokens generated by Mel Residual Vector Quantization (Mel-RVQ). Our Mel-RVQ utilizes residual linear projection structure for Mel spectrum quantization to enhance the stability and efficiency of target extraction and lead to better performance. Experiments in a large variety of downstream tasks demonstrate that MuQ outperforms previous self-supervised music representation models with only 0.9K hours of open-source pre-training data. Scaling up the data to over 160K hours and adopting iterative training consistently improve the model performance. To further validate the strength of our model, we present MuQ-MuLan, a joint music-text embedding model based on contrastive learning, which achieves state-of-the-art performance in the zero-shot music tagging task on the MagnaTagATune dataset. Code and checkpoints are open source in https://github.com/tencent-ailab/MuQ.
More music foundation models are recently being released, promising a general, mostly task independent encoding of musical information. Common ways of adapting music foundation models to downstream tasks are probing and fine-tuning. These common transfer learning approaches, however, face challenges. Probing might lead to suboptimal performance because the pre-trained weights are frozen, while fine-tuning is computationally expensive and is prone to overfitting. Our work investigates the use of parameter-efficient transfer learning (PETL) for music foundation models which integrates the advantage of probing and fine-tuning. We introduce three types of PETL methods: adapter-based methods, prompt-based methods, and reparameterization-based methods. These methods train only a small number of parameters, and therefore do not require significant computational resources. Results show that PETL methods outperform both probing and fine-tuning on music auto-tagging. On key detection and tempo estimation, they achieve similar results as fine-tuning with significantly less training cost. However, the usefulness of the current generation of foundation model on key and tempo tasks is questioned by the similar results achieved by training a small model from scratch. Code available at https://github.com/suncerock/peft-music/
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.