Enhancing Speech Intelligibility in Background Noise; Possibility of a New Masking-Suppressor Digital Signal Processing Strategy as an Adjunct to Wide Dynamic Range Compression

Abstract

Objectives: To explore the efficacy of a new digital signal processing strategy (DSP) by comparing speech intelligibility in noise measures across different sound strategies. Introduction: Hearing aids are of limited benefit for the perceptual consequences that arise from a loss of frequency selectivity. By eliminating inaudible sound energy, and reducing ‘masking’ noise around speech peaks based on individual spreading functions, the new DSP effectively acts as a masking suppressor to enhance spectral contrast between speech peaks and dips. It is hypothesised that a combination of wide dynamic range compression (WDRC) and DSP will show greatest improvement in objective measures of speech intelligibility in noise. It is also hypothesised that no difference in subjective sound quality will be found. Methods: Twenty current hearing aid wearers (60-75 years) with symmetrical, normal/mild, sloping to a moderate/severe sensorineural hearing loss were recruited. Treble increase at low levels (TILL) prescription templates were used to achieve an approximation of real-ear, prescribed WDRC. An estimate of critical bandwidth (CB) was obtained using a consonant-nucleus-consonant (CNC) in white-noise tournament with different DSP types. Objective (QuickSIN™) and subjective (soundscape quality) speech intelligibility in noise results were compared across four sound strategies: No-Processing, DSP, TILL and DSP + TILL. Results: DSP type 1.0 had the highest CNC in white-noise phoneme and word score overall, but was not significantly different from No-Processing. Both DSP and DSP + TILL had significantly higher mean total QuickSIN™ scores than No-Processing. Mean DSP + TILL QuickSIN™ scores at +5 dB signal to noise ratio (SNR) were significantly higher than No-Processing. DSP + TILL also had a significantly higher mean score at +20 dB SNR than No-Processing or TILL. No significant difference in sound quality between strategies was found. Conclusions: This study suggests speech intelligibility can be enhanced in background noise using the new DSP either alone, or in combination with WDRC. The method used to estimate individual CB could be implemented clinically. The new DSP could also be incorporated at the front-end of WDRC hearing aid processing. Future directions include the possibility of integrating a more accurate, objective measure of auditory filter characteristics with the new DSP, and exploring the effects of an acclimatisation period on speech intelligibility in noise.