Algorithm Allows Changing the Required Voice to a Predetermined One Using a Surrogate Voice

This article describes the algorithm for changing the input voice to a predefined voice. The main principle and new principle of the algorithm is the creation of a surrogate voice, the use of which will significantly reduce the time spent on the operation of the algorithm. The work consists of 3 main stages: training the neural network model according to the given parameters of the initial voice; calculation of filters that affect the characteristics of speech: tempo, speed, intonation, defects, as well as musical characteristics; applying the received filters to the input - variable voice.