Model management method, device and system, voice synthesis method, device and system, and storage media

Abstract

The embodiment of the invention provides a model management method, a model management device and a model management system, a voice synthesis method, a voice synthesis device and a voice synthesis system, and storage media. The model management method comprises the following steps: S110, receiving user request information, wherein the user request information comprises address information used for indicating an address of a storage file of a target acoustic model; S120, searching for the target acoustic model in an acoustic model pond; and S130, under the condition that the target acoustic model is not searched, loading the target acoustic model to the acoustic model pond according to the address information. With the technical scheme, the hot-loading of the acoustic model can be realized, and thus great degree of freedom can be provided for a user. According to the technical scheme, the intervention of the user is not needed, thus the operation of the user can be reduced, then the work efficiency of the voice system can be favorably improved, errors are reduced, and meanwhile, the user experience can be effectively promoted. With the technical scheme, the voice processing demandin the super-large-scale scenes with many pronouncers can be well satisfied.

Description

technical field [0001] The present invention relates to the technical field of speech, and more specifically relates to a model management method, device, system, and storage medium, and a speech synthesis method, device, system, and storage medium. Background technique [0002] Speech technologies, such as speech recognition, speech synthesis, etc., have flourished in recent years. Speech synthesis systems can synthesize speech with smooth pronunciation, rich emotions, and realistic sound quality. The synthesis effect is getting better and better, and its use is becoming more and more popular. [0003] At present, with the gradual development of speech synthesis technology, the demand for large-scale multi-speaker speech synthesis systems is becoming stronger and stronger, but the existing speech synthesis systems basically only support the application of acoustic models of several fixed speakers . [0004] In the existing speech synthesis technology, it is usually necess...

AI Technical Summary

Problems solved by technology

If users want to use a new acoustic model, they often need to modify the configuration file and restart the service. This method is inconvenient, inefficient, error-prone, and the user experience is relatively poor. At the same time, it is difficult to meet the speech synthesis needs of large-scale multi-speakers

Images