Speech synthesis method and device, storage medium and electronic equipment

Abstract

The invention relates to a speech synthesis method and device, a storage medium and electronic equipment, and the method comprises the steps: extracting a first feature vector of the speech of a target speaker, and extracting the sound features of the target speaker from the speech of the target speaker through a speaker feature extraction network; performing parameter adjustment on a first decoder according to the first feature vector, the target speaker voice feature and the target speaker voice; constructing a target speech synthesis model through the first decoder and the second encoder after parameter adjustment; and inputting the to-be-synthesized text and the sound features of the target speaker into the target speech synthesis model to synthesize the obtained target speech. Therefore, the method does not need to completely depend on the capability of extracting the voice features of the speaker authorized by the user for use by the speaker feature extraction network, and does not solidify the noisy tone quality information in a voice synthesis system during adjustment according to the voice parameters of the target speaker authorized by the user for use. And the stability and precision of speech synthesis are ensured.

Description

technical field [0001] The present disclosure relates to the technical field of audio processing, and in particular, to a speech synthesis method, device, storage medium and electronic equipment. Background technique [0002] In the field of speech synthesis, in general application scenarios, the synthesis requires a large amount of data (more than 5h) for support to have a relatively stable effect. For most users, it is unrealistic to record 5h data according to strict specifications, and for regular users, when synthesizing their own voice, they pay more attention to the effect of synthesized voice and their own voice in terms of timbre and tone. How to enhance the pronunciation stability of the speech synthesis system itself and improve the sound quality as much as possible while ensuring the user's timbre effect is a problem that needs to be solved. [0003] Existing speech synthesis schemes usually need to rely absolutely on the speaker feature extraction network with ...

AI Technical Summary

Problems solved by technology

[0003] Existing speech synthesis schemes usually need to rely absolutely on the speaker feature extraction network with strong decoupling ability, that is, the synthesized speech and the target speaker speech that needs to be synthesized and authorized by the user are absolutely dependent on speech However, the ability of the speaker feature extraction network in the prior art cannot fully meet the needs of this scenario; in addition, there are speech synthesis solutions that first use the target speaker's voice that has been authorized by the user. Retrain the pre-trained speech synthesis system to achieve the effect of synthesizing timbre. However, since the purpose of the speech synthesis system is to synthesize speech with sound quality information, if the target speaker's speech that has been authorized by the user Noisy, the trained speech synthesis system will also include the noisy sound quality information, which will lead to the problem that the subsequent speech synthesized from the text is noisy

Images