DocumentCode :
2035773
Title :
Prediction of Arabic speech intelligibility for speech hall
Author :
El Awady, R. ; El Malawany, A.I. ; El Messiry, M.A. ; Fayed, H.S.
fYear :
2002
fDate :
2002
Firstpage :
214
Lastpage :
223
Abstract :
Different types of acoustical measures were made and compared as predictors of speech intelligibility in rooms of varied acoustical characteristic. These acoustical measurements include articulation index (Al), A-weighting signal to noise ratios (S/N) obtained from speech and background noise levels, useful to detrimental sound ratios obtained from early to late sound ratios. Some room acoustic parameters were also considered like reverberation time RT, early decay time for 10 dB (EDT), early to late sound ratios (C50) and early to total sound energy ratios (D50). The relationships of these measures to speech intelligibility scores were studied to evaluate which were most appropriate in room for speech, and the best predictors of speech intelligibility scores were identified. It is found that the most successful measure is the useful to detrimental sound ratios based. on a 50 msec. early time interval as a predictor of speech intelligibility. From the results the optimum acoustical conditions in room for speech were determined in terms of the 50 msec. useful/detrimental ratios or articulation index or from the combinations of the reverberation time and signal to noise ratio.
Keywords :
acoustic variables measurement; architectural acoustics; natural languages; noise; reverberation; speech intelligibility; A-weighting signal to noise ratio; Arabic speech intelligibility prediction; SNR; acoustical characteristic; acoustical measurements; acoustical measures; articulation index; background noise level; early decay time; early to late sound ratios; early to total sound energy ratios; optimum acoustical conditions; reverberation time; room acoustic parameters; speech hall; useful to detrimental sound ratios; Acoustic measurements; Artificial intelligence; Background noise; Energy measurement; Noise measurement; Reverberation; Signal to noise ratio; Speech analysis; Speech enhancement; Time measurement;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Radio Science Conference, 2002. (NRSC 2002). Proceedings of the Nineteenth National
Print_ISBN :
977-5031-72-9
Type :
conf
DOI :
10.1109/NRSC.2002.1022624
Filename :
1022624
Link To Document :
بازگشت