Research On Evolution Technology Of Open Source Software Oriented To Speech Recognition Application

The emergence of the open source software community provides a wealth of reference prototype resources for software development,from software technology prototypes to machine learning models and so on.However,these prototype resources often depend on specific application backgrounds and computing environments,and it is difficult to adapt to individual needs.How to effectively use these open source prototypes is a new challenge.This article combines the application of speech recognition to study how to use software evolution technology to solve the problem of open source speech model adaptation.The reflection mechanism is used to dynamically perceive changes in the environment and changes in voice recognition requirements,and adaptively adjust open source software to meet the needs of personalized voice recognition applications.In the research,model-driven and data-driven evolutionary algorithms are proposed to solve the personalization problem of voice recognition model evolution caused by changes in voice recognition accuracy requirements and changes in the voice model environment.Take the application of a speech recognition system for table tennis techniques and tactics as an example to test the effectiveness of the proposed algorithm.Designed and developed a prototype system for collecting speech recognition of table tennis techniques and tactics.It adopts model library,method library and database architecture.Driven by different methods in the method library,the model-driven and data-driven speech recognition adaptive process is completed.