Benefits of the WaveNet-Based Speech Intelligibility Enhancement for Normaland Hearing Impaired Listeners

Speech perception becomes challenging in adverse listening conditions, hence increasing mental effort. Mod-ification of recorded clean speech with the goal of increasing its intelligibility is one approach for improvinglistening experience. Recently, we suggested a data-driven WaveNet-like speech intelligibility enhancementmethod which is based on the Spectral Shaping and Dynamic Range Compressions (SSDRC) approach. Bothapproaches achieve intelligibility gains by relocating energy from high sonorant to less sonorant portions ofspeech. In this study we have assessed the performance of the WaveNet intelligibility enhancer for both nor-mal and hearing impaired participants using formal listening tests, in terms of intelligibility and sound quality.Intelligibility was measured as a percentage of correct words recalled (CWR) and quality was assessed viaa comparative mean opinion score (CMOS). Furthermore, we compared performance among native and non-native listeners using English and Greek stimuli. We observed that modified speech was significantly moreintelligible than plain across all listener groups. Hearing impaired candidates rated modified speech 20% higheron average (on a CMOS scale) than normal hearing subjects. The positive impact for the hearing impairedlisteners might be attributed to the reallocation of energy to perceptually relevant frequency bands in which thehearing impaired are less sensitive.