Neuro-MSBG: An End-to-End Neural Model for Hearing Loss Simulation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yuan, Hui-Guan, Zezario, Ryandhimas E., Ahmed, Shafique, Wang, Hsin-Min, Hua, Kai-Lung, Tsao, Yu
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908458046980096
author Yuan, Hui-Guan
Zezario, Ryandhimas E.
Ahmed, Shafique
Wang, Hsin-Min
Hua, Kai-Lung
Tsao, Yu
author_facet Yuan, Hui-Guan
Zezario, Ryandhimas E.
Ahmed, Shafique
Wang, Hsin-Min
Hua, Kai-Lung
Tsao, Yu
contents Hearing loss simulation models are essential for hearing aid deployment. However, existing models have high computational complexity and latency, which limits real-time applications and lack direct integration with speech processing systems. To address these issues, we propose Neuro-MSBG, a lightweight end-to-end model with a personalized audiogram encoder for effective time-frequency modeling. Experiments show that Neuro-MSBG supports parallel inference and retains the intelligibility and perceptual quality of the original MSBG, with a Spearman's rank correlation coefficient (SRCC) of 0.9247 for Short-Time Objective Intelligibility (STOI) and 0.8671 for Perceptual Evaluation of Speech Quality (PESQ). Neuro-MSBG reduces simulation runtime by a factor of 46 (from 0.970 seconds to 0.021 seconds for a 1 second input), further demonstrating its efficiency and practicality.
format Preprint
id arxiv_https___arxiv_org_abs_2507_15396
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Neuro-MSBG: An End-to-End Neural Model for Hearing Loss Simulation
Yuan, Hui-Guan
Zezario, Ryandhimas E.
Ahmed, Shafique
Wang, Hsin-Min
Hua, Kai-Lung
Tsao, Yu
Sound
Artificial Intelligence
Audio and Speech Processing
Hearing loss simulation models are essential for hearing aid deployment. However, existing models have high computational complexity and latency, which limits real-time applications and lack direct integration with speech processing systems. To address these issues, we propose Neuro-MSBG, a lightweight end-to-end model with a personalized audiogram encoder for effective time-frequency modeling. Experiments show that Neuro-MSBG supports parallel inference and retains the intelligibility and perceptual quality of the original MSBG, with a Spearman's rank correlation coefficient (SRCC) of 0.9247 for Short-Time Objective Intelligibility (STOI) and 0.8671 for Perceptual Evaluation of Speech Quality (PESQ). Neuro-MSBG reduces simulation runtime by a factor of 46 (from 0.970 seconds to 0.021 seconds for a 1 second input), further demonstrating its efficiency and practicality.
title Neuro-MSBG: An End-to-End Neural Model for Hearing Loss Simulation
topic Sound
Artificial Intelligence
Audio and Speech Processing
url https://arxiv.org/abs/2507.15396