Louis Braida’s influence on speech intelligibility research

Abstract

In the Sensory Communications Laboratory at MIT, several researchers have grappled with the question of how to predict listeners' perceptual performance using quantitative metrics. Most of the work has been motivated by the goal of understanding the effect of acoustic degradations on speech intelligibility for hearing impaired listeners and determining the best way to mitigate or counter those degradations. Louis Braida has been interested in this topic for many years. He worked with Ken Grant to investigate augmentation of the Articulation Index (AI) with visual information to obtain an audio-visual AI. He continued this work with other students modeling perceptual integration across modalities. Lou and I worked together to investigate the ability of the Speech Transmission Index (STI) to predict intelligibility for impaired listeners. Initially, we tried to use it as a way to try and capture acoustic differences between conversational and clearly-articulated speech. That evolved into the development of ...