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

Abstract

The embodiment of the invention provides a model management method, device and system, a storage medium, a voice synthesis method, device and system and a storage medium. The model management method comprises the steps of receiving identity information of a current user; performing identity authentication based on the identity information to determine whether or not the current user is a legal user; searching an engine module for an acoustic model of a target object on the basis of identification information of the target object under the condition that the current user is a legal user, wherein the target object is one of authorized objects in an authorized object set corresponding to the current user; loading the acoustic model of the target object into an engine module under the condition that the acoustic model of the target object is not found. According to the scheme, dynamic loading of the acoustic model can be achieved, a great degree of freedom can be provided for the user, theuser experience can be effectively improved, and the voice processing demand under an extremely large-scale scene where there are many speakers can be well met.

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 technology is a technology that converts text information into sound information. Speech synthesis technology can provide speech synthesis services for users and third-party applications. 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] Most of the current speech synthesis services are provided by the service provider with a fixed speaker model list, such as "standard synthesis-female voice-Jiaojiao", "sp...

AI Technical Summary

Problems solved by technology

That is to say, the existing technical solutions have certain restrictions on the use of speaker models, that is, they cannot dynamically adjust the speaker models provided by the speech synthesis service in real time, thereby limiting the system’s application of a large number of speaker models.

Images