Saved in:
Bibliographic Details
Main Authors: Zhang, Junan, Zhang, Yunjia, Zhang, Xueyao, Wu, Zhizheng
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.14052
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914042787921920
author Zhang, Junan
Zhang, Yunjia
Zhang, Xueyao
Wu, Zhizheng
author_facet Zhang, Junan
Zhang, Yunjia
Zhang, Xueyao
Wu, Zhizheng
contents Singing Accompaniment Generation (SAG) is the process of generating instrumental music for a given clean vocal input. However, existing SAG techniques use source-separated vocals as input and overfit to separation artifacts. This creates a critical train-test mismatch, leading to failure on clean, real-world vocal inputs. We introduce AnyAccomp, a framework that resolves this by decoupling accompaniment generation from source-dependent artifacts. AnyAccomp first employs a quantized melodic bottleneck, using a chromagram and a VQ-VAE to extract a discrete and timbre-invariant representation of the core melody. A subsequent flow-matching model then generates the accompaniment conditioned on these robust codes. Experiments show AnyAccomp achieves competitive performance on separated-vocal benchmarks while significantly outperforming baselines on generalization test sets of clean studio vocals and, notably, solo instrumental tracks. This demonstrates a qualitative leap in generalization, enabling robust accompaniment for instruments - a task where existing models completely fail - and paving the way for more versatile music co-creation tools. Demo audio and code: https://anyaccomp.github.io
format Preprint
id arxiv_https___arxiv_org_abs_2509_14052
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AnyAccomp: Generalizable Accompaniment Generation via Quantized Melodic Bottleneck
Zhang, Junan
Zhang, Yunjia
Zhang, Xueyao
Wu, Zhizheng
Sound
Signal Processing
Singing Accompaniment Generation (SAG) is the process of generating instrumental music for a given clean vocal input. However, existing SAG techniques use source-separated vocals as input and overfit to separation artifacts. This creates a critical train-test mismatch, leading to failure on clean, real-world vocal inputs. We introduce AnyAccomp, a framework that resolves this by decoupling accompaniment generation from source-dependent artifacts. AnyAccomp first employs a quantized melodic bottleneck, using a chromagram and a VQ-VAE to extract a discrete and timbre-invariant representation of the core melody. A subsequent flow-matching model then generates the accompaniment conditioned on these robust codes. Experiments show AnyAccomp achieves competitive performance on separated-vocal benchmarks while significantly outperforming baselines on generalization test sets of clean studio vocals and, notably, solo instrumental tracks. This demonstrates a qualitative leap in generalization, enabling robust accompaniment for instruments - a task where existing models completely fail - and paving the way for more versatile music co-creation tools. Demo audio and code: https://anyaccomp.github.io
title AnyAccomp: Generalizable Accompaniment Generation via Quantized Melodic Bottleneck
topic Sound
Signal Processing
url https://arxiv.org/abs/2509.14052