基于非单调共轭梯度算法的声纹识别机器人控制系统设计

Design of voiceprint recognition robot control system based on non-monotone conjugate gradient algorithm

摘要:

传统声纹识别人控制系统识别准确率低,存在语音识别噪声鲁棒性问题。针对上述问题,基于非单调共轭梯度算法设计了一种新的声纹识别机器人控制系统,采用BioVoice 2.0 标准声纹采集器采集数据,提取声纹特征,根据提取的声纹特征建立模型库,同时引用了两个声纹数据采集终端,型号分别是TMC104-B和TMC104,选用型号为AS-MrobotR的机器人配合采集器和采集终端实现工作。在Windows平台下使用C/C+语言研究了一种专用的程序,在程序内部添加mde-api数据库,完成训练程序和识别程序。实验结果表明,基于非单调共轭梯度算法的声纹识别机器人控制系统能够很好地解决语音识别噪声鲁棒性问题,在有噪声环境下识别准确率提高15.24%,在无噪声环境下识别准确率提高21.55%。

Abstract:

The recognition accuracy of traditional voiceprint recognition system is low, and there is a problem of speech recognition noise robustness. Aiming at the above problems, a new voiceprint recognition robot control system is designed based on non-monotone conjugate gradient algorithm. The BioVoice 2.0 standard voiceprint collector is used to collect data, extract voiceprint features, and build a model library based on the extracted voiceprint features. At the same time, two voiceprint data acquisition terminals are cited, the models are TMC104-B and TMC104 respectively, and the robot with model AS-MrobotR is selected to cooperate with the collector and the collection terminal to achieve work. In the Windows platform, a special program is studied using C/C+ language. The mde-api database is added inside the program to complete the training program and the recognition program. The experimental results show that the voiceprint recognition robot control system based on non-monotone conjugate gradient algorithm can solve the problem of speech recognition noise robustness. The recognition accuracy is improved by 15.24% in noisy environment. The accuracy rate increased by 21.55%.