Perception of Speech in the Presence of Competing Voices Perception of Speech in the Presence of Competing VoicesThis work has two primary goals: (i) to investigate the processes by which human listeners extract information from speech in the presence of competing sounds; (ii) to develop computational models of sound segregation and speech perception.
We are conducting laboratory studies with both synthesized and natural speech to study the effects of competing sounds on the perception of speech. When the competing sound source is another voice, listeners confront the difficult problem of separating two signals that are similar in their acoustic structure. This problem has serious implications for theoretical models of speech perception and has important practical consequences for two areas of applied speech research. First, many studies have found that competing voices present particular difficulties for individuals suffering from sensorineural hearing impairments. Research on the perceptual processes involved in speech-source segregation may provide insights into the problems faced by these listeners. It may also suggest forms of signal processing to enhance the intelligibility of speech corrupted by background noise. Second, competing voices severely degrade the performance of automatic speech recognizers. A better understanding of human performance will inevitably lead to improvements in the design of robust and noise-resistant devices for speech recognition. One of our long-term research objectives is the development of a preprocessor for speech analysis based on a model of human perception that can track an individual's voice through variable and unpredictable forms of background noise.
Project 1. Effects of waveform interactions on the use of fundamental frequency differences
Project 2. Role of formant transitions in the perception of concurrent vowels
Project 3. Pitches of concurrent vowels