Speech synthesis based corpus augmentation method, system, device and medium

Abstract

The application relates to the technical field of voice synthesis, in particular to a corpus expansion method and system based on voice synthesis, equipment and a medium, wherein the method comprises the following steps: based on collected audio and corresponding text of a target speaker, pre-processing including data labeling is carried out; acoustic characteristics are extracted from the pre-processed audio; based on a pre-trained acoustic model, individual fine-tuning is carried out by using labeled data and acoustic characteristics, and multiple speaker individual acoustic models are simultaneously trained through multi-thread parallel computing; the trained individual acoustic model is called, and a voice corpus of a target text is synthesized in combination with a vocoder; and based on the trained individual acoustic model, a corpus library is continuously expanded by replacing texts. The application can quickly generate individual voice corpus through a small amount of voice samples, significantly reduces data collection cost, and improves corpus construction efficiency.

Images