Quantitative Evaluation of the Performance of Monaural FDSI Beamforming Algorithm using a KEMAR Mannequin

Kyeongwon Cho, Kyoung Won Nam, Jonghee Han, Sangmin Lee, Dong-Wook Kim, Sung Hwa Hong, Dong Pyo Jang, In Young KimJournal of Biomedical Engineering Research 2013 · DOI

Abstract: To enhance the speech perception of hearing aid users in noisy environment, most hearing aid devices adopt various beamforming algorithms such as the first-order differential microphone (DM1) and the two-stage directional microphone (DM2) algorithms that maintain sounds from the direction of the interlocutor and reduce the ambient sounds from the other directions. However, these conventional algorithms represent poor directionality ability in low frequency area. Therefore, to enhance the speech perception of hearing aid uses in low frequency range, our group had suggested a fractional delay subtraction and integration (FDSI) algorithm and estimated its theoretical performance using computer simulation in previous article. In this study, we performed a KEMAR test in non-reverberant room that compares the performance of DM1, DM2, broadband beamforming (BBF), and proposed FDSI algorithms using several objective indices such as a signal-to-noise ratio (SNR) improvement, a segmental SNR (seg-SNR) improvement, a perceptual evaluation of speech quality (PESQ), and an Itakura-Saito measure (IS). Experimental results showed that the performance of the FDSI algorithm was -3.26-7.16 dB in SNR improvement, -1.94-5.41 dB in segSNR improvement, 1.49-2.79 in PESQ, and 0.79-3.59 in IS, which demonstrated that the FDSI algorithm showed the highest improvement of SNR and segSNR, and the lowest IS. We believe that the proposed FDSI algorithm has a potential as a beamformer for digital hearing aid devices.

Introduction

For hearing aid users, the "Cocktail Party Problem"—the ability to focus on a single speaker in a noisy environment—remains a significant challenge. Modern digital hearing aids typically employ beamforming algorithms to create directionality, prioritizing sounds from the front while attenuating ambient noise from other directions.

The industry standards have long been first-order (DM1) and two-stage (DM2) differential microphone algorithms. However, these methods suffer from a physical limitation: they struggle with low-frequency sounds. Due to the small spacing (a few millimeters) between microphones in a hearing aid, phase differences for long-wavelength sounds are minimal, leading to poor directionality and signal-to-noise ratio (SNR) loss in the bass range. This study evaluates a proposed Fractional Delay Subtraction and Integration (FDSI) algorithm using a physical KEMAR mannequin to validate its performance in a real-world acoustic environment.

Methods and Materials

The researchers compared four algorithms: DM1, DM2, BBF, and the proposed FDSI. The FDSI approach utilizes fractional delays and cross-correlation to align signals between two microphones more effectively. To ensure realistic testing, a KEMAR (Knowles Electronics Mannequin for Acoustic Research) was used. This mannequin mimics the average human head and torso, accounting for the "head shadowing effect"—the physical blocking and diffraction of sound by the human body.

Experimental setup showing the KEMAR mannequin in an anechoic chamber surrounded by an 8-speaker array.

The experimental setup consisted of a non-reverberant (anechoic) room with a behind-the-ear (BTE) hearing aid housing containing two microphones spaced 8mm apart. Eight speakers were arranged in a circle around the KEMAR at 45° intervals. The study utilized the IEEE speech database and the Phonak noise database (multi-talker babble) to evaluate performance across diverse acoustic conditions.

Results and Discussion

The FDSI algorithm demonstrated superior performance across nearly all objective indices. A critical advantage of FDSI is its low-frequency stability. While traditional DM1 and DM2 polar patterns lose their directional shape at 200Hz and 500Hz, FDSI maintains a consistent beam across the entire frequency spectrum.

Comparison of SNR and segmental SNR (segSNR) improvements across different noise directions for DM1, DM2, BBF, and FDSI algorithms.

Key quantitative findings include:

  • SNR Improvement: FDSI achieved gains of -3.26 to 7.16 dB, consistently outperforming the other algorithms.
  • Speech Quality (PESQ): FDSI scores ranged from 1.49 to 2.79, showing greater consistency across different noise directions compared to BBF.
  • Signal Distortion (IS): FDSI yielded the lowest Itakura-Saito scores (0.79 to 3.59), indicating minimal spectral distortion and better preservation of the original speech characteristics.

The use of the KEMAR mannequin confirmed that FDSI remains robust even when the physical presence of a head distorts incoming sound waves, a scenario where simpler algorithms often fail.

Conclusion

The FDSI algorithm represents a significant advancement in hearing aid beamforming, particularly in addressing low-frequency noise reduction without sacrificing speech quality. While the algorithm currently exhibits some high-frequency roll-off above 4kHz, its overall performance in SNR and signal preservation is promising. Future research will focus on expanding this monaural approach to binaural systems and optimizing the algorithm for dynamic, reverberant real-world environments.

Summary updated

Abstract

To enhance the speech perception of hearing aid users in noisy environment, most hearing aid devices adopt various beamforming algorithms such as the first-order differential microphone (DM1) and the two-stage directional microphone (DM2) algorithms that maintain sounds from the direction of the interlocutor and reduce the ambient sounds from the other directions. However, these conventional algorithms represent poor directionality ability in low frequency area. Therefore, to enhance the speech perception of hearing aid uses in low frequency range, our group had suggested a fractional delay subtraction and integration (FDSI) algorithm and estimated its theoretical performance using computer simulation in previous article. In this study, we performed a KEMAR test in non-reverberant room that compares the performance of DM1, DM2, broadband beamforming (BBF), and proposed FDSI algorithms using several objective indices such as a signal-to-noise ratio (SNR) improvement, a segmental SNR (seg-SNR) improvement, a perceptual evaluation of speech quality (PESQ), and an Itakura-Saito measure (IS). Experimental results showed that the performance of the FDSI algorithm was -3.26-7.16 dB in SNR improvement, -1.94-5.41 dB in segSNR improvement, 1.49-2.79 in PESQ, and 0.79-3.59 in IS, which demonstrated that the FDSI algorithm showed the highest improvement of SNR and segSNR, and the lowest IS. We believe that the proposed FDSI algorithm has a potential as a beamformer for digital hearing aid devices.
BibTeX
@article{cho2013quantitative,
  author = {Kyeongwon Cho and Kyoung Won Nam and Jonghee Han and Sangmin Lee and Dong-Wook Kim and Sung Hwa Hong and Dong Pyo Jang and In Young Kim},
  title = {{Quantitative Evaluation of the Performance of Monaural FDSI Beamforming Algorithm using a KEMAR Mannequin}},
  journal = {Journal of Biomedical Engineering Research},
  year = {2013},
  doi = {10.9718/jber.2013.34.1.24},
}
Figure