Analyzable Chain-of-Musical-Thought Prompting for High-Fidelity Music Generation

Autoregressive (AR) models have demonstrated impressive capabilities in generating high-fidelity music. However, the conventional next-token prediction paradigm in AR models does not align with the human creative process in music composition, potentially compromising the musicality of generated samples. To overcome this limitation, we introduce MusiCoT, a novel chain-of-thought (CoT) prompting technique tailored for music generation. MusiCoT empowers the AR model to first outline an overall music structure before generating audio tokens, thereby enhancing the coherence and creativity of the resulting compositions. By leveraging the contrastive language-audio pretraining (CLAP) model, we establish a chain of "musical thoughts", making MusiCoT scalable and independent of human-labeled data, in contrast to conventional CoT methods. Moreover, MusiCoT allows for in-depth analysis of music structure, such as instrumental arrangements, and supports music referencing -- accepting variable-length audio inputs as optional style references. This innovative approach effectively addresses copying issues, positioning MusiCoT as a vital practical method for music prompting. Our experimental results indicate that MusiCoT consistently achieves superior performance across both objective and subjective metrics, producing music quality that rivals state-of-the-art generation models. Our samples are available at https://MusiCoT.github.io/.

Kimi-Audio Technical Report

We present Kimi-Audio, an open-source audio foundation model that excels in audio understanding, generation, and conversation. We detail the practices in building Kimi-Audio, including model architecture, data curation, training recipe, inference deployment, and evaluation. Specifically, we leverage a 12.5Hz audio tokenizer, design a novel LLM-based architecture with continuous features as input and discrete tokens as output, and develop a chunk-wise streaming detokenizer based on flow matching. We curate a pre-training dataset that consists of more than 13 million hours of audio data covering a wide range of modalities including speech, sound, and music, and build a pipeline to construct high-quality and diverse post-training data. Initialized from a pre-trained LLM, Kimi-Audio is continual pre-trained on both audio and text data with several carefully designed tasks, and then fine-tuned to support a diverse of audio-related tasks. Extensive evaluation shows that Kimi-Audio achieves state-of-the-art performance on a range of audio benchmarks including speech recognition, audio understanding, audio question answering, and speech conversation. We release the codes, model checkpoints, as well as the evaluation toolkits in https://github.com/MoonshotAI/Kimi-Audio.

Text2Tracks: Prompt-based Music Recommendation via Generative Retrieval

In recent years, Large Language Models (LLMs) have enabled users to provide highly specific music recommendation requests using natural language prompts (e.g. "Can you recommend some old classics for slow dancing?"). In this setup, the recommended tracks are predicted by the LLM in an autoregressive way, i.e. the LLM generates the track titles one token at a time. While intuitive, this approach has several limitation. First, it is based on a general purpose tokenization that is optimized for words rather than for track titles. Second, it necessitates an additional entity resolution layer that matches the track title to the actual track identifier. Third, the number of decoding steps scales linearly with the length of the track title, slowing down inference. In this paper, we propose to address the task of prompt-based music recommendation as a generative retrieval task. Within this setting, we introduce novel, effective, and efficient representations of track identifiers that significantly outperform commonly used strategies. We introduce Text2Tracks, a generative retrieval model that learns a mapping from a user's music recommendation prompt to the relevant track IDs directly. Through an offline evaluation on a dataset of playlists with language inputs, we find that (1) the strategy to create IDs for music tracks is the most important factor for the effectiveness of Text2Tracks and semantic IDs significantly outperform commonly used strategies that rely on song titles as identifiers (2) provided with the right choice of track identifiers, Text2Tracks outperforms sparse and dense retrieval solutions trained to retrieve tracks from language prompts.

Towards Responsible AI Music: an Investigation of Trustworthy Features for Creative Systems

Generative AI is radically changing the creative arts, by fundamentally transforming the way we create and interact with cultural artefacts. While offering unprecedented opportunities for artistic expression and commercialisation, this technology also raises ethical, societal, and legal concerns. Key among these are the potential displacement of human creativity, copyright infringement stemming from vast training datasets, and the lack of transparency, explainability, and fairness mechanisms. As generative systems become pervasive in this domain, responsible design is crucial. Whilst previous work has tackled isolated aspects of generative systems (e.g., transparency, evaluation, data), we take a comprehensive approach, grounding these efforts within the Ethics Guidelines for Trustworthy Artificial Intelligence produced by the High-Level Expert Group on AI appointed by the European Commission - a framework for designing responsible AI systems across seven macro requirements. Focusing on generative music AI, we illustrate how these requirements can be contextualised for the field, addressing trustworthiness across multiple dimensions and integrating insights from the existing literature. We further propose a roadmap for operationalising these contextualised requirements, emphasising interdisciplinary collaboration and stakeholder engagement. Our work provides a foundation for designing and evaluating responsible music generation systems, calling for collaboration among AI experts, ethicists, legal scholars, and artists. This manuscript is accompanied by a website: https://amresearchlab.github.io/raim-framework/.

Expressive Timing in Hindustani Vocal Music

Temporal dynamics are among the cues to expres siveness in music performance in different cultures. In the case of Hindustani music, it is well known that expert vocalists often take liberties with the beat, intentionally not aligning their singing precisely with the relatively steady beat provided by the accompanying tabla. This becomes evident when comparing performances of the same composition such as a bandish. We present a methodology for the quantitative study of differences across performed pieces using computational techniques. This is applied to small study of two performances of a popular bandish in raga Yaman, to demonstrate how we can effectively capture the nuances of timing variations that bring out stylistic constraints along with the individual signature of a performer. This work articulates an important step towards the broader goals of music analysis and generative modelling for Indian classical music performance.

Tune It Up: Music Genre Transfer and Prediction

Deep generative models have been used in style transfer tasks for images. In this study, we adapt and improve CycleGAN model to perform music style transfer on Jazz and Classic genres. By doing so, we aim to easily generate new songs, cover music to different music genres and reduce the arrangements needed in those processes. We train and use music genre classifier to assess the performance of the transfer models. To that end, we obtain 87.7% accuracy with Multi-layer Perceptron algorithm. To improve our style transfer baseline, we add auxiliary discriminators and triplet loss to our model. According to our experiments, we obtain the best accuracies as 69.4% in Jazz to Classic task and 39.3% in Classic to Jazz task with our developed genre classifier. We also run a subjective experiment and results of it show that the overall performance of our transfer model is good and it manages to conserve melody of inputs on the transferred outputs. Our code is available at https://github.com/ fidansamet/tune-it-up

DanceMosaic: High-Fidelity Dance Generation with Multimodal Editability

Recent advances in dance generation have enabled automatic synthesis of 3D dance motions. However, existing methods still struggle to produce high-fidelity dance sequences that simultaneously deliver exceptional realism, precise dance-music synchronization, high motion diversity, and physical plausibility. Moreover, existing methods lack the flexibility to edit dance sequences according to diverse guidance signals, such as musical prompts, pose constraints, action labels, and genre descriptions, significantly restricting their creative utility and adaptability. Unlike the existing approaches, DanceMosaic enables fast and high-fidelity dance generation, while allowing multimodal motion editing. Specifically, we propose a multimodal masked motion model that fuses the text-to-motion model with music and pose adapters to learn probabilistic mapping from diverse guidance signals to high-quality dance motion sequences via progressive generative masking training. To further enhance the motion generation quality, we propose multimodal classifier-free guidance and inference-time optimization mechanism that further enforce the alignment between the generated motions and the multimodal guidance. Extensive experiments demonstrate that our method establishes a new state-of-the-art performance in dance generation, significantly advancing the quality and editability achieved by existing approaches.

CrossMuSim: A Cross-Modal Framework for Music Similarity Retrieval with LLM-Powered Text Description Sourcing and Mining

Music similarity retrieval is fundamental for managing and exploring relevant content from large collections in streaming platforms. This paper presents a novel cross-modal contrastive learning framework that leverages the open-ended nature of text descriptions to guide music similarity modeling, addressing the limitations of traditional uni-modal approaches in capturing complex musical relationships. To overcome the scarcity of high-quality text-music paired data, this paper introduces a dual-source data acquisition approach combining online scraping and LLM-based prompting, where carefully designed prompts leverage LLMs' comprehensive music knowledge to generate contextually rich descriptions. Exten1sive experiments demonstrate that the proposed framework achieves significant performance improvements over existing benchmarks through objective metrics, subjective evaluations, and real-world A/B testing on the Huawei Music streaming platform.

A Computational Cognitive Model for Processing Repetitions of Hierarchical Relations

Patterns are fundamental to human cognition, enabling the recognition of structure and regularity across diverse domains. In this work, we focus on structural repeats, patterns that arise from the repetition of hierarchical relations within sequential data, and develop a candidate computational model of how humans detect and understand such structural repeats. Based on a weighted deduction system, our model infers the minimal generative process of a given sequence in the form of a Template program, a formalism that enriches the context-free grammar with repetition combinators. Such representation efficiently encodes the repetition of sub-computations in a recursive manner. As a proof of concept, we demonstrate the expressiveness of our model on short sequences from music and action planning. The proposed model offers broader insights into the mental representations and cognitive mechanisms underlying human pattern recognition.

YuE: Scaling Open Foundation Models for Long-Form Music Generation

We tackle the task of long-form music generation--particularly the challenging \textbf{lyrics-to-song} problem--by introducing YuE, a family of open foundation models based on the LLaMA2 architecture. Specifically, YuE scales to trillions of tokens and generates up to five minutes of music while maintaining lyrical alignment, coherent musical structure, and engaging vocal melodies with appropriate accompaniment. It achieves this through (1) track-decoupled next-token prediction to overcome dense mixture signals, (2) structural progressive conditioning for long-context lyrical alignment, and (3) a multitask, multiphase pre-training recipe to converge and generalize. In addition, we redesign the in-context learning technique for music generation, enabling versatile style transfer (e.g., converting Japanese city pop into an English rap while preserving the original accompaniment) and bidirectional generation. Through extensive evaluation, we demonstrate that YuE matches or even surpasses some of the proprietary systems in musicality and vocal agility. In addition, fine-tuning YuE enables additional controls and enhanced support for tail languages. Furthermore, beyond generation, we show that YuE's learned representations can perform well on music understanding tasks, where the results of YuE match or exceed state-of-the-art methods on the MARBLE benchmark. Keywords: lyrics2song, song generation, long-form, foundation model, music generation