Text-independent speech conversion system based on HMM model state mapping

Abstract

The invention discloses a text-independent speech conversion system based on HMM model state mapping, which is composed of a data alignment module, a spectrum conversion model generation module, a rhythm conversion model generation module, an online conversion module and a parameter voice synthesizer; wherein, the data alignment module receives the voice parameters of the source and target speakers, and aligns to the input data according to phoneme information to generate state-aligned data pairs; the spectrum conversion model generation module receives the aligned data pairs and establishes a voice spectrum parameter conversion module based on source and target speakers according to the data; the rhythm conversion model generation module receives the aligned data pairs and establishes a voice rhythm parameter conversion module based on source and target speakers according to the data; the online conversion module obtains the converted voice spectrum parameter and rhythm parameter according to the conversion modules generated by the spectrum conversion model generation module and the rhythm conversion model generation module, and voice data of the source speaker for conversion; the parameter voice synthesizer module receives the converted spectrum information and rhythm information from the online conversion module and outputs the converted voice result.

Description

technical field [0001] The invention relates to a speech conversion system, in particular to a text-independent speech conversion system based on hidden Markov model state mapping. Background technique [0002] Harmonious human-computer interaction technology has always been the object of people's attention. Voice conversion technology for personalized voice is an important part of it. It can process a person's voice and make it into another person's voice. The research results It is of great significance to the development of personalized speech generation and man-machine dialogue. Most of the existing speech conversion technologies are generally based on text-related technologies. This technology must require the source speaker and the target speaker to provide the same speech training samples of the text, which is also called parallel corpus training. In real life, the requirements for parallel corpus are relatively high, and technical users need to spend a lot of energy...

AI Technical Summary

Problems solved by technology

The situation of non-parallel corpus is much more complicated, because in this case, the source speaker and the target speaker can say different things, it is difficult to align the voice of the source speaker with the voice of the target speaker

In response to this problem, some scholars try to use the distance between speech spectrum parameters as the criterion for speech alignment training. Although this method can produce a relatively smooth transition function or rule based on the shortest distance criterion, it only relies on the distance of speech spectrum parameters to establish The mapping relationship is prone to phoneme misplacement and reduces the accuracy of conversion

Images