METHODS OF PREPROCESSING SPEECH SIGNALS 

Valeriy V. Kozlov, Candidate of technical sciences, associate professor of the sub-department of information and measuring equipment and metrology, Penza State University (40 Krasnaya street, Penza, Russia), val.iit@mail.ru
Aleksey A. Trofimov, Doctor of technical sciences, associate professor, professor of the sub-department of information and measuring equipment and metrology, Penza State University (40 Krasnaya street, Penza, Russia), iit@pnzgu.ru
Ekaterina A. Fokina, Postgraduate student, Penza State University (40 Krasnaya street, Penza, Russia), ekaterina.isay1997@gmail.com
Vladislav N. Ponomarev, Postgraduate student, Penza State University (40 Krasnaya street, Penza, Russia), revik2296@gmail.com
Timofey O. Zhukov, Student, Penza State University (40 Krasnaya street, Penza, Russia), iit@pnzgu.ru

Background. Automatic speech recognition technology faces a lot of difficulties. The complex structure of human speech combined with various noise effects makes this task very time-consuming. To simplify this task, preprocessing of speech signals is used, which can be carried out using various methods. Materials and methods. Preprocessing of speech signals includes a number of techniques and methods aimed at improving the quality of speech data and preparing them for further analysis. The article discusses the methods of preprocessing speech signals and provides directions for the application of these methods. An overview of some techniques of preprocessing speech signals is presented. Results and conclusions. The analysis of the most suitable methods for preprocessing speech signals is presented.

speech signals, preprocessing, empirical modes decomposition, wavelet transform, Fourier transform

Kozlov V.V., Trofimov A.A., Fokina E.A., Ponomarev V.N., Zhukov T.O. Methods of preprocessing speech signals. Izmerenie. Monitoring. Upravlenie. Kontrol' = Measuring. Monitoring. Management. Control. 2023;(4):43–49. (In Russ.). doi: 10.21685/2307-5538-2023-4-5