摘要
针对语音识别中声学特征提取存在的局限性,提出一种基于自适应特征加权的声学特征优化方法。首先,分析声学特征提取在语音识别中的作用,介绍传统梅尔频率倒谱系数(Mel-scale Frequency Cepstral Coefficients,MFCC)方法的基本原理和存在的问题。其次,提出自适应特征加权方法,通过计算自适应权重优化MFCC特征。最后,进行实验分析。实验结果表明,优化方法在语音识别任务中具有较好的效果和实用性。
The article proposes an acoustic feature optimization method based on adaptive feature weighting to address the limitations of acoustic feature extraction in speech recognition.Firstly,analyze the role of acoustic feature extraction in speech recognition,and introduce the basic principles and existing problems of the traditional Mel-scale Frequency Cepstral Coefficients(MFCC)method.Secondly,an adaptive feature weighting method is proposed to optimize MFCC features by calculating adaptive weights.Finally,conduct experimental analysis.The experimental results show that the optimization method has good effectiveness and practicality in speech recognition tasks.
作者
杨波
YANG Bo(School of Mathematics and Information,Longnan Teachers College,Longnan 742500,China)
出处
《电声技术》
2024年第4期51-53,共3页
Audio Engineering
关键词
语音识别
特征提取
声学特征
speech recognition
feature extraction
acoustic characteristics
