Enregistré dans:
Détails bibliographiques
Auteurs principaux: Torres, Diego, Roebel, Axel, Obin, Nicolas
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2510.25566
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915585470758912
author Torres, Diego
Roebel, Axel
Obin, Nicolas
author_facet Torres, Diego
Roebel, Axel
Obin, Nicolas
contents We present PitchFlower, a flow-based neural audio codec with explicit pitch controllability. Our approach enforces disentanglement through a simple perturbation: during training, F0 contours are flattened and randomly shifted, while the true F0 is provided as conditioning. A vector-quantization bottleneck prevents pitch recovery, and a flow-based decoder generates high quality audio. Experiments show that PitchFlower achieves more accurate pitch control than WORLD at much higher audio quality, and outperforms SiFiGAN in controllability while maintaining comparable quality. Beyond pitch, this framework provides a simple and extensible path toward disentangling other speech attributes.
format Preprint
id arxiv_https___arxiv_org_abs_2510_25566
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PitchFlower: A flow-based neural audio codec with pitch controllability
Torres, Diego
Roebel, Axel
Obin, Nicolas
Audio and Speech Processing
Machine Learning
We present PitchFlower, a flow-based neural audio codec with explicit pitch controllability. Our approach enforces disentanglement through a simple perturbation: during training, F0 contours are flattened and randomly shifted, while the true F0 is provided as conditioning. A vector-quantization bottleneck prevents pitch recovery, and a flow-based decoder generates high quality audio. Experiments show that PitchFlower achieves more accurate pitch control than WORLD at much higher audio quality, and outperforms SiFiGAN in controllability while maintaining comparable quality. Beyond pitch, this framework provides a simple and extensible path toward disentangling other speech attributes.
title PitchFlower: A flow-based neural audio codec with pitch controllability
topic Audio and Speech Processing
Machine Learning
url https://arxiv.org/abs/2510.25566