Method and system for improving multi-round man-machine interaction performance

• Main Authors: WANG YING, WEI YONGQIANG, XIA GUANGYU
• Format: Patent
• Language: Chinese; English
• Published: 02.04.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a method and system for improving multi-round man-machine interaction performance, and the method comprises the steps: inputting real-time voice data into a trained voice conversion network model, so as to convert and output text data; inputting the text data into a trained emotion recognition network model to recognize and obtain an emotion capture result; inputting an emotion capture result into the trained dialogue system model so as to output an automatic optimization strategy through interaction with the environment based on a reinforcement learning algorithm; and inputting the automatic optimization strategy into the trained GPT-4 model so as to output and obtain a corresponding semantic response. According to the method, the problems of conversation flexibility and adaptability, fixed and single reply due to insufficient conversation strategies and conversation quality stability can be effectively solved. 本发明公开了一种提升多轮人机交互性能的方法和系统，该方法，包括将实时语音数据输入至训练好的语音转化网络模型以转化输出文本数据；将文本数据输入至训练好