ALGORITHMS FOR SPEAKER RECOGNITION BASED ON WAVELET TRANSFORM AND MACHINE LEARNING IN SPEECH SIGNAL ANALYSIS

Аннотация

In this work, features were isolated from the speech signal using Haar, Daubechies, and Morlet wavelet transforms, which were subsequently used for speaker recognition in SVM and LSTM models. The research was conducted on the basis of a dataset of 100 speakers. The obtained results made it possible to analyze the effectiveness between wavelet types and machine learning models.