This paper proposes methods for unsupervised detection of AI-generated music in large-scale catalogs.

This paper proposes a framework for creating low-latency, interactive generative music AI using distillation in a streaming autoregressive latent space and music-aware consistency objectives.

The paper introduces HAIM, a new benchmark dataset designed to move AI music detection beyond simple binary classification by tracking specific stages and types of AI integration in music production.

This paper introduces a maintainable hybrid architecture for generating harmonies from melodies using quantum-inspired exploration and rule-based optimization.

This paper introduces WanSong, a diffusion-based model for long-form, commercial-grade song generation that directly generates high-fidelity, multilingual songs up to 5 minutes and outputs dual stems,…

This paper introduces a unified framework for generating high-quality full-length music from lyrics, text descriptions, and musical attributes, consisting of a semantic-aware tokenizer, hybird-LM, Ful…

This paper proposes RPPNet, a two-stage deep learning architecture for music generation with variable structural boundaries, which automatically derives grouping of Rhythm-Pitch Primitive sequences fr…

The paper introduces MMGenre, a benchmark for multi-genre singing voice synthesis diagnosis, revealing limited genre discrimination and proposing lightweight genre-specific continued training.

This paper proposes a method for learning a world model of piano sound using Joint Embedding Predictive Architectures (JEPA), treating music as an action-conditioned system.

The paper introduces Reflector, an interactive audio workstation that adapts pitch-class retrieval as compositions evolve, using a learned embedding space based on a hand-designed oracle.

This paper investigates the detectability of AI-generated artifacts in human-AI hybrid music tracks at the stem level, proposing a parallel architecture for detecting AI-generated stems.

The paper proposes an inference-time activation steering framework, utilizing orthogonalization, to achieve fine-grained, deterministic control over discrete musical attributes like Pitch and Duration…

This paper analyzes the effectiveness of synthetic data for pre-training music transcription models, combines it with fine-tuning on real music audio, and introduces conditioning on instrument presenc…

This paper identifies and quantifies two structural bottlenecks in certain state-of-the-art neural audio models that limit access to frequency-localized primitives, and proposes a lightweight interven…

A deep neural timbre trait predictor is introduced to evaluate neural audio synthesizers' performance using human judgments and correlate with average human ratings.