Speech synthesis method and device, electronic equipment and readable storage medium

Abstract

The embodiment of the invention provides a speech synthesis method and device, electronic equipment and a readable storage medium. The method comprises the steps of obtaining phoneme features of a to-be-synthesized text; taking the phoneme features as input of a trained acoustic model, and generating acoustic information corresponding to the phoneme features based on a processing layer in the acoustic model; the processing layer comprises at least one pair of first sampling module and second sampling module; the first sampling module is used for reducing the feature dimension, and the second sampling module is used for recovering the feature dimension; the trained acoustic model is obtained based on sample text training with the length not smaller than a preset length threshold value; and acquiring voice corresponding to the to-be-synthesized text based on the acoustic information. Therefore, the first sampling module and the second sampling module in the acoustic model are used for firstly reducing the feature dimension of the phoneme sequence participating in calculation and then recovering the feature dimension after processing, so that speech synthesis can be directly carried out without carrying out forced sentence segmentation on a long text to a certain extent.

Description

technical field [0001] The present invention belongs to the technical field of speech, and in particular relates to a speech synthesis method, apparatus, electronic device and readable storage medium. Background technique [0002] With the continuous development of computer technology, human-computer voice interaction technology has also made great progress, and in the field of human-computer voice interaction, there are two key technologies of speech recognition and speech synthesis. Speech synthesis is to convert input text information into corresponding speech. In the application scenarios of speech synthesis, long text synthesis is often encountered, such as news scenarios. [0003] In the prior art, speech synthesis is generally performed after the text is forcibly segmented. This synthesis method will result in poor final synthesis effect. SUMMARY OF THE INVENTION [0004] The present invention provides a speech synthesis method, device, electronic device and readab...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a speech synthesis method, device, electronic equipment and readable storage medium, so as to solve the problem of poor synthesis effect when synthesizing long texts

Images