Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Single Microphone Own Voice Detection based on Simulated Transfer Functions for Hearing Aids

Mathuranathan Mayuravaani, W. Bastiaan Kleijn, Andrew Lensen, Charlotte Sørensen

TL;DR

A data augmentation strategy based on simulated acoustic transfer functions (ATFs) that expose the model to a wide range of spatial propagation conditions that highlights the model's ability to generalize from simulated to real-world conditions, demonstrating practical viability and pointing toward a promising direction for future hearing aid design.

Abstract

This paper presents a simulation-based approach to own voice detection (OVD) in hearing aids using a single microphone. While OVD can significantly improve user comfort and speech intelligibility, existing solutions often rely on multiple microphones or additional sensors, increasing device complexity and cost. To enable ML-based OVD without requiring costly transfer-function measurements, we propose a data augmentation strategy based on simulated acoustic transfer functions (ATFs) that expose the model to a wide range of spatial propagation conditions. A transformer-based classifier is first trained on analytically generated ATFs and then progressively fine-tuned using numerically simulated ATFs, transitioning from a rigid-sphere model to a detailed head-and-torso representation. This hierarchical adaptation enabled the model to refine its spatial understanding while maintaining generalization. Experimental results show 95.52% accuracy on simulated head-and-torso test data. Under short-duration conditions, the model maintained 90.02% accuracy with one-second utterances. On real hearing aid recordings, the model achieved 80% accuracy without fine-tuning, aided by lightweight test-time feature compensation. This highlights the model's ability to generalize from simulated to real-world conditions, demonstrating practical viability and pointing toward a promising direction for future hearing aid design.

Single Microphone Own Voice Detection based on Simulated Transfer Functions for Hearing Aids

TL;DR

A data augmentation strategy based on simulated acoustic transfer functions (ATFs) that expose the model to a wide range of spatial propagation conditions that highlights the model's ability to generalize from simulated to real-world conditions, demonstrating practical viability and pointing toward a promising direction for future hearing aid design.

Abstract

This paper presents a simulation-based approach to own voice detection (OVD) in hearing aids using a single microphone. While OVD can significantly improve user comfort and speech intelligibility, existing solutions often rely on multiple microphones or additional sensors, increasing device complexity and cost. To enable ML-based OVD without requiring costly transfer-function measurements, we propose a data augmentation strategy based on simulated acoustic transfer functions (ATFs) that expose the model to a wide range of spatial propagation conditions. A transformer-based classifier is first trained on analytically generated ATFs and then progressively fine-tuned using numerically simulated ATFs, transitioning from a rigid-sphere model to a detailed head-and-torso representation. This hierarchical adaptation enabled the model to refine its spatial understanding while maintaining generalization. Experimental results show 95.52% accuracy on simulated head-and-torso test data. Under short-duration conditions, the model maintained 90.02% accuracy with one-second utterances. On real hearing aid recordings, the model achieved 80% accuracy without fine-tuning, aided by lightweight test-time feature compensation. This highlights the model's ability to generalize from simulated to real-world conditions, demonstrating practical viability and pointing toward a promising direction for future hearing aid design.
Paper Structure (23 sections, 12 equations, 11 figures, 7 tables)

This paper contains 23 sections, 12 equations, 11 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Method
  4. Geometry and basic formulas
  5. Analytical Modelling Approach
  6. Hpp(k)
  7. Hps(k)
  8. Hcp(k)
  9. Hcs(k)
  10. Numerical Modelling Approach
  11. Own voice detection algorithm
  12. Experimental setup
  13. Dataset
  14. Model
  15. Training methodology
  16. ...and 8 more sections

Figures (11)

  • Figure 1: Geometry of point source and rigid sphere.
  • Figure 2: Mapping the pressure between (a) a point source and a point receiver, (b) a point source and a rigid sphere, (c) a rigid sphere source with vibrating cap and a point receiver, and (d) vibrating spherical cap and location on the rigid sphere. Here, S denotes the source location and M denotes the receiver (microphone) location.
  • Figure 3: Mapping the pressure at an angle on the sphere when (a) a speech signal is from a point source from a distance (external speaker) and (b) a speech signal is from a vibrating cap on the rigid sphere (own voice).
  • Figure 4: Geometry of oscillating cap in a rigid sphere.
  • Figure 5: Directivity pattern $20 \log_{10}( D(\theta)/D(0))$ of the near-field pressure for a straight-ahead point source scattering from a 7 cm rigid sphere and a spherical cap ($\alpha=20^\circ$).
  • ...and 6 more figures