| Autor: |
Dobrushkin, G. O., Danilov, V. Ya. |
| Předmět: |
|
| Zdroj: |
Naukovi visti NTUU - KPI; 2010, Vol. 2010 Issue 2, p34-42, 9p |
| Abstrakt: |
Based on the components of the wavelet-transform theory, we present an automated method of speech signals denoising, interference cleaning and ADC artifacts refining. We propose a new method of detecting interphoneme transitions in the continuous speech signals as well as logarithmic and moving average criteria that help make a decision on the speech signals segmentation. In addition, we analyze advantages and disadvantages of the proposed criteria. We also show the complexity of automation of the problem solving of speech signal segmentation under the conditions of absence of a priori knowledge on phonemes natures and transition between them. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Supplemental Index |
| Externí odkaz: |
|
