Abstract: Digital signals including human speeches are very popular and important digital data due to the large number of computerized applications that required them, some of these applications must have an immediate response generating a response in significantly small time. Digital speech usually has a big size, and to reduce a processing time we have to represent the speech by a unique identifier or classifier called speech features. In this paper research we will study the wavelet packet tree decomposition, and we will introduce a method to calculate speech histogram. This histogram can be used as an input data set to the signal decomposition to fix the levels of decomposition and to fix the number of elements in the approximation or the detailed in a selected level of decomposition.
Keywords: Speech Signal, WPT, Histogram, Level, Approximations, Details, Features.
| DOI: 10.17148/IJARCCE.2020.9620