文件名称:Research_on_DTW-based_speech_recognition_for_voice
介绍说明－－下载内容均来自于网络，请自行研究使用
研究了将语音识别中的 ( 动态时间规整) 算法用于声纹鉴别的技术。通过引入
DTW Dynamic Time Warping,
, 由所给的有限个样本建立最大相似于样本点的样本域, 计算被测样本的相似度。该算法提高了语
“样本域”的概念
音鉴别( 区分不同发音者) 的效率。有限人数的实验结果显示该算法辨伪率为 ( 人次) , 识别率 (98.75 400 81~93 80
人次) 。-Will be studied in speech recognition (dynamic time warping) algorithm for voiceprint identification technology. Through the introduction of DTW Dynamic Time Warping,, by the limited samples to establish the most similar samples in the sample domain, the similarity calculation of the measured samples. The algorithm improves the phrase " sample domain," the concept of the sound identification (to distinguish between persons of different pronunciation) efficiency. Limited number of laboratory tests show that the algorithm for the detection of forgeries rate (people), recognition rate (98.75 400 81 ~ 93 to 80 trips).
DTW Dynamic Time Warping,
, 由所给的有限个样本建立最大相似于样本点的样本域, 计算被测样本的相似度。该算法提高了语
“样本域”的概念
音鉴别( 区分不同发音者) 的效率。有限人数的实验结果显示该算法辨伪率为 ( 人次) , 识别率 (98.75 400 81~93 80
人次) 。-Will be studied in speech recognition (dynamic time warping) algorithm for voiceprint identification technology. Through the introduction of DTW Dynamic Time Warping,, by the limited samples to establish the most similar samples in the sample domain, the similarity calculation of the measured samples. The algorithm improves the phrase " sample domain," the concept of the sound identification (to distinguish between persons of different pronunciation) efficiency. Limited number of laboratory tests show that the algorithm for the detection of forgeries rate (people), recognition rate (98.75 400 81 ~ 93 to 80 trips).
(系统自动生成,下载前可以参看下载内容)
下载文件列表
Research_on_DTW-based_speech_recognition_for_voiceprint_identify.pdf