基于人工智能调度的失语症康复治疗仪的端点检测和声韵母切分

Detection of Endpoint for Segmentation Between Consonants and Vowels in Aphasia Rehabilitation Software Based on Artificial Intelligence Scheduling

为了提高患者康复训练的效率,本研究在失语症康复治疗仪中添加了人工智能调度功能,使软件的性能进一步完善。结合患者语音的特点及康复治疗仪中人工智能调度功能的需要,设计一套端点检测算法。仿真实验表明,该方法从检测参考端点出发,从而确定字的实际端点及声韵母合理切分点,准确率较高,适用于失语症患者语音的检测。

For the purpose of improving the efficiency of aphasia rehabilitation training, artificial intelligence-scheduling function is added in the aphasia rehabilitation software, and the software＇s performance is improved. With the characteristics of aphasia patient＇s voice as well as with the need of artificial intelligence-scheduling functions under consideration, the present authors have designed a set of endpoint detection algorithm. It determines the reference endpoints, then extracts every word and ensures the reasonable segmentation points between consonants and vowels, using the reference endpoints. The results of experiments show that the algorithm is able to attain the objects of detection at a higher accuracy rate. Therefore, it is applicable to the detection of endpoint on aphasia-patient＇s voice.