LMAC-TD: Producing Time Domain Explanations for Audio Classifiers

Eleonora Mancini, Francesco Paissan, Mirco Ravanelli, Cem Subakan
9/13/2024
cs.SDcs.AIcs.LG

No official code found. Best implementation: fpaissan/lmac-td-code (3 stars, last commit 2025-10-19).

Abstract

Neural networks are typically black-boxes that remain opaque with regards to their decision mechanisms. Several works in the literature have proposed post-hoc explanation methods to alleviate this issue. This paper proposes LMAC-TD, a post-hoc explanation method that trains a decoder to produce explanations directly in the time domain. This methodology builds upon the foundation of L-MAC, Listenable Maps for Audio Classifiers, a method that produces faithful and listenable explanations. We incorporate SepFormer, a popular transformer-based time-domain source separation architecture. We show through a user study that LMAC-TD significantly improves the audio quality of the produced explanations while not sacrificing from faithfulness.

View on arXivView PDF

Code Implementations(1)

fpaissan/lmac-td-codeCommunity71%

Official Implementation of LMAC-TD: Producing Time Domain Explanations for Audio Classifiers

30PythonNov 26, 20241 years ago

Similar Papers

Listenable Maps for Audio Classifiers

83% match2024

Listenable Maps for Zero-Shot Audio Classifiers

80% match2024

CleanUMamba: A Compact Mamba Network for Speech Denoising using Channel Pruning

78% match1 repo272024

XAI-based Comparison of Input Representations for Audio Event Classification

78% match2023

AudioMNIST: Exploring Explainable Artificial Intelligence for Audio Analysis on a Simple Benchmark

78% match1 repo3752018

Short-Term Memory Convolutions

78% match2023

Cite this paper

@article{mancini2024lmactd,
  title  = {LMAC-TD: Producing Time Domain Explanations for Audio Classifiers},
  author = {Eleonora Mancini and Francesco Paissan and Mirco Ravanelli and Cem Subakan},
  year   = {2024},
  eprint = {2409.08655},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2409.08655v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2409.08655.svg)](https://codeofpaper.com/paper/2409.08655)

Discussion