Should Audio Front-ends be Adaptive? Comparing Learnable and Adaptive Front-ends

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Qiquan, Wickramasinghe, Buddhi, Ambikairajah, Eliathamby, Sethu, Vidhyasaharan, Li, Haizhou
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917913998393344
author Zhang, Qiquan
Wickramasinghe, Buddhi
Ambikairajah, Eliathamby
Sethu, Vidhyasaharan
Li, Haizhou
author_facet Zhang, Qiquan
Wickramasinghe, Buddhi
Ambikairajah, Eliathamby
Sethu, Vidhyasaharan
Li, Haizhou
contents Hand-crafted features, such as Mel-filterbanks, have traditionally been the choice for many audio processing applications. Recently, there has been a growing interest in learnable front-ends that extract representations directly from the raw audio waveform. \textcolor{black}{However, both hand-crafted filterbanks and current learnable front-ends lead to fixed computation graphs at inference time, failing to dynamically adapt to varying acoustic environments, a key feature of human auditory systems.} To this end, we explore the question of whether audio front-ends should be adaptive by comparing the Ada-FE front-end (a recently developed adaptive front-end that employs a neural adaptive feedback controller to dynamically adjust the Q-factors of its spectral decomposition filters) to established learnable front-ends. Specifically, we systematically investigate learnable front-ends and Ada-FE across two commonly used back-end backbones and a wide range of audio benchmarks including speech, sound event, and music. The comprehensive results show that our Ada-FE outperforms advanced learnable front-ends, and more importantly, it exhibits impressive stability or robustness on test samples over various training epochs.
format Preprint
id arxiv_https___arxiv_org_abs_2502_03260
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Should Audio Front-ends be Adaptive? Comparing Learnable and Adaptive Front-ends
Zhang, Qiquan
Wickramasinghe, Buddhi
Ambikairajah, Eliathamby
Sethu, Vidhyasaharan
Li, Haizhou
Audio and Speech Processing
Sound
Hand-crafted features, such as Mel-filterbanks, have traditionally been the choice for many audio processing applications. Recently, there has been a growing interest in learnable front-ends that extract representations directly from the raw audio waveform. \textcolor{black}{However, both hand-crafted filterbanks and current learnable front-ends lead to fixed computation graphs at inference time, failing to dynamically adapt to varying acoustic environments, a key feature of human auditory systems.} To this end, we explore the question of whether audio front-ends should be adaptive by comparing the Ada-FE front-end (a recently developed adaptive front-end that employs a neural adaptive feedback controller to dynamically adjust the Q-factors of its spectral decomposition filters) to established learnable front-ends. Specifically, we systematically investigate learnable front-ends and Ada-FE across two commonly used back-end backbones and a wide range of audio benchmarks including speech, sound event, and music. The comprehensive results show that our Ada-FE outperforms advanced learnable front-ends, and more importantly, it exhibits impressive stability or robustness on test samples over various training epochs.
title Should Audio Front-ends be Adaptive? Comparing Learnable and Adaptive Front-ends
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2502.03260