Some experiments on the use of one-channel noise reduction techniques with the Italian SpeechDat Car database
Tóm tắt
The use of noise reduction techniques for hands-free speech recognition in a car environment is investigated. A set of experiments was carried out using different speech enhancement algorithms based on noise estimation. In particular, linear spectral subtraction and MMSE estimators are considered with various parameter settings. Experiments were conducted on connected and isolated digits, extracted from the Italian version of the SpeechDat Car database. Recognition rates do not agree with acoustically perceived quality of noise reduction. As a result, the best performance is obtained by spectral subtraction with a suitable choice of the oversubtraction factor and a quantile noise estimator. It provides more than 30% relative performance improvement, from 94.4% of the baseline to 96.2% digit recognition accuracy.
Từ khóa
#Noise reduction #Speech enhancement #Working environment noise #Speech recognition #Low-frequency noise #Databases #Background noise #Road safety #Additive noise #Noise robustnessTài liệu tham khảo
evans, 0, Noise estimation without explicit speech, non-speech detection: a comparison of mean, median and modal based approaches, Proc of Eurospeech 2001 accepted for publishing
10.1007/978-3-322-92773-6
10.1109/ICASSP.1979.1170788
evans, 2000, An assessment of local nonlinear spectral subtraction for remote speech recognition, Proc of 1st meeting on Speech Technology
10.1109/ICASSP.1983.1171938
10.1109/TASSP.1985.1164550
cristoforetti, 2000, Annotation of a multichannel noisy speech corpus, Proc of LREC
junqua, 1996, Robustness in automatic speech recognition, 10.1007/978-1-4613-1297-0
10.1016/0167-6393(92)90016-Z
nolazco flores, 0, Continuous speech recognition in noise using spectral subtraction and HMM adaptation, Proc ICASSP 1994, i, 409
matassoni, 0, Use of real and contaminated speech for training of a hands-free in-car speech recognizer, Proc of Eurospeech 2001 accepted for publishing
10.1109/ICASSP.1995.479387
10.1109/89.928915
10.1016/S0167-6393(98)00030-2
10.1109/ICASSP.1999.758387
10.1109/ICASSP.2000.862122