Objective Assessment of Speech Signal Filtering Algorithms For Hearing Aids

The aim of this work evaluates the performance of three speech improvement algorithms in hearing aids, Noise Reduction via Spectral Subtraction Enhancement (NRSE), Adaptive Frequency-Domain Range Compression (AFDRC), and Beamforming Noise Suppression (BNS). The action focuses on key objective metrics such as Speech Intelligibility Index (SII), Signal-to-Noise Ratio (SNR) Improvement, and Total Harmonic Distortion (THD). Outcomes indicate that all three algorithms perform similarly in no-noise environments, maintaining speech clarity and low distortion. However, in high-noise conditions, BNS is consistently profitable for both NRSE and AFDRC. BNS shows greater noise reduction capabilities, improving SNR by 10% compared to NRSE and 8% over AFDRC. Additionally, BNS reduces distortion more effectively, reducing THD by 9% and 7% compared to the other two methods. These findings help to highlight BNS's superior ability to improve speech clarity and reduce background noise, particularly in challenging environments. The work highlights the importance of using objective metrics like SII, SNR Improvement, and THD to evaluate and effective speech filtering algorithms in hearing aids. By improving performance in noisy conditions, BNS offers significant benefits for individuals with hearing impairments, enhancing overall hearing aid effectiveness in real-world scenarios.