Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.38252

A Tutorial on Generative Models for Music Audio

Mateusz Modrzejewski·Joanna Kurkowska·Dorien Herremans
View original source ↗Read source document ↗Export metadata ↓

Abstract

Music audio generation spans instrumental clips, full songs with vocals, edits of existing recordings, and systems that play alongside a musician in real time. This monograph explains the methods behind these systems for readers familiar with neural-network training who are new to music generation. It introduces neural audio codecs and autoencoders, autoregressive and masked token models, diffusion and flow matching, sampling acceleration, and preference optimization in a shared notation. Worked examples connect training objectives and sampling procedures to published systems. The application sections relate the musical requirements of instrumental generation, songs, editing, and live performance to model design. Comparison tables record representations, methods, inputs, tasks, and release information documented in papers and official sources. Symbolic composition, speech, and sound effects are included where they connect to music audio. We explain how to interpret automatic metrics and learned judges, design and report listening tests, and assess evidence about data provenance, memorization, attribution, and output detection. Selected legal decisions and disputes illustrate the implications for training and release. The reviewed systems combine established generative paradigms with new representations, planning stages, and post-training procedures. Differences in data, scale, and evaluation limit what cross-system comparisons can establish about the sources of reported gains.

Explore related subjects

eess.AScs.AIcs.ETcs.MMcs.SD

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Mateusz Modrzejewski, Joanna Kurkowska, Dorien Herremans. 2026-10-28. A Tutorial on Generative Models for Music Audio. https://arxiv.org/abs/2610.38252

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Auditory and Visual Virtualization in 6DoF XR: Effects on Plausibility and Dynamic Auditory Distance Perception

Extended reality (XR) enables controlled yet naturalistic multisensory experiments, but its validity depends on whether virtualized cues are perceptually plausible and lead to behavior comparable to the real world. To the best of our knowledge, this is the first study to perceptually evaluate a six-degree-of-freedom (6DoF) XR system that combines real-time room acoustic simulation, binaural rendering, and an audio-visually matched virtual replica of a large laboratory room in terms of both plausibility and behavior. In two experiments with 45 participants, we investigated how auditory and visual virtualization affect plausibility and dynamic auditory distance perception (ADP). A plausibility experiment compared real and virtual sound sources in audio-augmented reality (AAR) and virtual reality (VR), while a locomotion-based ADP experiment assessed distance-judgment accuracy, signed error, and task effort across four virtualization conditions. The system produced plausible binaural rendering in both AAR and VR, with higher plausibility in VR, indicating that perceptual requirements for XR audio depend on the visual reference frame. In the ADP task, any form of virtualization reduced accuracy and increased effort, although performance remained high overall. Virtual audio produced a bias toward nearer perceived source distances, whereas visual-only virtualization reduced accuracy without a significant directional bias. These effects did not add up in full VR. Overall, the findings demonstrate the strong potential of current 6DoF XR systems for interactive perceptual research, while revealing that task-specific behavioral validation may be necessary.

eess.AS↗
arXiv2026

FUSEMOS: Perceptual Evaluation of Text-to-Music Generation with Dual-Encoder Fusion and Ranking-Aware Composite Loss

Perceptual evaluation of generated music is essential for developing text-to-music (TTM) generation systems. Existing automatic music perceptual evaluation methods rely solely on the contrastive language audio pretraining (CLAP) model to extract both audio and text features for predicting human-annotated mean opinion scores (MOS), particularly for overall musical impression (OMI) and textual alignment (TA). However, CLAP is limited in capturing fine-grained musical attributes such as timbre and expressiveness \emph{etc}. To address this, we propose FUSEMOS, a dual-encoder fusion framework that integrates CLAP with MERT, a self-supervised model pretrained on large scale music audio, to better capture both high level semantic alignment and detailed musical characteristics for more accurate OMI and TA estimation. Furthermore, to overcome the limitations of conventional regression losses in measuring human preferred relative ranking of music segments, we design a novel ranking-aware composite loss function. Experiments conducted on the MusicEval benchmark demonstrate that our approach outperforms the existing SOTA methods in both mean squared error and rank correlation metrics by a large margin.

eess.AS↗
arXiv2026

A Unified Multi-Encoder Framework for Music Lyric Intelligibility Assessment: Bridging Invasive and Non-Invasive Paradigms

Lyric intelligibility significantly impacts music engagement and comprehension, especially for hearing-impaired individuals. Existing speech intelligibility metrics perform poorly in musical contexts due to differences in prosody, timing, and spectral dynamics. We propose a unified multi-encoder framework that supports both invasive and non-invasive modeling for robust lyric intelligibility prediction. By capturing complementary local acoustic-semantic alignments, global musical semantics, and ASR-derived textual cues, our method achieves superior performance over baselines on the ICASSP 2026 Cadenza dataset, with the best model achieving 26.61 RMSE, a $9.2$\% improvement over the baseline, demonstrating its flexibility and robustness.

eess.AS↗