论文部分内容阅读
降频助听算法是改善听障患者声音辨识能力的最安全有效的方法.本文以主观测试实验为手段,通过分析当前算法的声音识别能力的不足,提出一种自适应慢放降频算法.算法结合慢放算法和频移算法的优点,并能根据信号的频谱结构,自适应调整慢放因子,降低时域不同步性.并且,通过分析含噪信号和噪声信号的频谱关系,提出一种噪声下的慢放因子评估方法.实验结果显示,同其他降频算法相比,该算法可以提高15%到20%的识别率.在对听障患者的测试中,同传统的助听设备相比,平均识别率也获得显著改善.
The frequency-down hearing-aid algorithm is the most safe and effective way to improve the speech recognition ability of hearing-impaired patients.In this paper, a subjective test experiment is taken as the means to analyze the current algorithm’s lack of voice recognition capability, and an adaptive slow-down frequency-down algorithm is proposed. The algorithm combines the advantages of slow-release algorithm and frequency-shift algorithm, and can adaptively adjust the slow-release factor and reduce the time-domain unsynchronization according to the spectral structure of the signal.And by analyzing the spectral relationship between the noisy signal and the noisy signal, The results show that this algorithm can improve the recognition rate of 15% to 20% compared with other down-conversion algorithms.In the test of hearing-impaired patients, with the traditional hearing aid equipment Compared with the average recognition rate has also been significantly improved.