A microscopic model of speech recognition for listeners with normal and impaired hearing

Meer over het boek

Degraded speech intelligibility is a common issue for listeners with sensorineural hearing impairment, affecting their ability to understand speech in both noisy and quiet environments. Understanding how hearing impairment impacts speech intelligibility is crucial for developing effective hearing-aid algorithms. However, the effects of sensorineural hearing impairment can vary significantly among individuals due to differences in the inner ear's functional deficits. Key factors influencing speech intelligibility include the audibility of speech signals, compressive properties, and active processes in the inner ear, collectively known as supra-threshold factors, which affect speech processing above the absolute threshold. Directly measuring these factors in humans is ethically challenging, but computer models can offer insights. This dissertation introduces a microscopic model of human speech recognition that focuses on phoneme recognition and processes the spectro-temporal structure of speech akin to the human auditory system. This model predicts phoneme recognition in normal-hearing listeners in noise and consonant recognition in both normal-hearing and hearing-impaired listeners in quiet. Additionally, it extends to predicting word recognition rates in German sentences and accurately assesses speech reception thresholds for various listeners. The research also evaluates supra-threshold auditory processing deficits through ps