Speech synthesis method and system, storage medium and computer equipment

Abstract

The invention is suitable for the technical field of speech synthesis, and provides a speech synthesis method. The method comprises the following steps of: acquiring a speech input signal and converting the speech input signal into text data; analyzing a field and intention information corresponding to the text data, and generating corresponding feedback information according to the field and the intention information; analyzing and evaluating the fluency and audio quality of the feedback information according to the intention information and / or the rhythm of the voice input signal; and selecting a corresponding speech synthesis back-end strategy through a preset processing strategy according to the evaluation result of the fluency and audio quality, so as to synthesize the feedback information into feedback speech. A speech synthesis system, a storage medium for storing a computer program for performing the method, and a computer device implementing the method are also provided. Therefore, the speech synthesis method and system can adapt to voice speech in different scenes, and the user experience is improved.

Description

technical field [0001] The invention relates to the technical field of speech synthesis, in particular to a speech synthesis method, system, storage medium and computer equipment thereof. Background technique [0002] TTS (Text To Speech, from text to speech) technology essentially solves the problem of "conversion from text to speech", by which the machine can speak. But this process is not easy. In order to reduce the difficulty of machine understanding, this conversion process is divided into two parts - the front-end system and the back-end system; figure 1 , the front end is responsible for converting the input text into an intermediate result, and then sends the intermediate result to the back end, which generates sound. The intermediate results generated by the front-end system are called "linguistic specifications". [0003] At present, there are two methods in the mainstream back-end system: one is based on waveform splicing, and the other is based on parameter ge...

AI Technical Summary

Problems solved by technology

The disadvantage is that the quality is worse than the splicing method

[0004] The splicing method needs to obtain the acoustic model before it can perform online splicing and synthesis. Because of the splicing operation, although it is better than the parametric method in terms of TTS broadcast quality, it is worse than the parametric method in terms of fluency, especially In some scenarios that are particularly sensitive to traffic, because the parameter method can finally synthesize offline TTS locally, but the splicing method must be spliced ​​and synthesized in the cloud relying on a large voice library, so it is more dependent on the network environment

[0005] Therefore, in the face of the scene of voice control terminal equipment in life, what kind of TTS synthesis method should be selected in different fields or intentions, the existing technology has not provided a corresponding solution, so it is necessary to improve

Images