Title :
Detection of Chinese Vowels in Noise and Its Auditory Simulation Model
Author_Institution :
Dept. of Commun. Sci. & Disorders, Univ. of Texas at Austin, Austin, TX, USA
Abstract :
The goal of this study was to examine detection thresholds of three Chinese vowels in long-term speech-shaped (LTSS) noise. Results showed that vowel detection thresholds were significantly affected by Mandarin tones, but not by vowel category. An auditory simulation model based on excitation-pattern computation was conducted to predict detection thresholds in LTSS noise and indicated that the pattern of predicted thresholds was not well matched with the pattern of behavioral thresholds. These results suggest that detection of Chinese vowels was dependent on not only vowel spectrum, but also tonal pitch, especially when tonal pitch changes dramatically over the time.
Keywords :
character recognition; speech recognition; Chinese auditory simulation model; Chinese vowel detection; LTSS; Mandarin tones; excitation-pattern computation; long- term speech-shaped noise; Acoustic noise; Acoustic signal detection; Analysis of variance; Computational modeling; Humans; Noise level; Noise measurement; Pattern matching; Predictive models; Speech enhancement;
Conference_Titel :
Bioinformatics and Biomedical Engineering (iCBBE), 2010 4th International Conference on
Conference_Location :
Chengdu
Print_ISBN :
978-1-4244-4712-1
Electronic_ISBN :
2151-7614
DOI :
10.1109/ICBBE.2010.5515704