BriefGPT.xyz
Jun, 2012
一种用于说话人识别的高效MFCC计算的新型窗口技术
A Novel Windowing Technique for Efficient Computation of MFCC for Speaker Recognition
HTML
PDF
Md. Sahidullah, Goutam Saha
TL;DR
本文提出一种基于离散时间傅里叶变换的微分性质，采用新型窗函数计算音频中的MFCC特征，用于自动语音识别和说话人识别，实验证明该方法可显著提高系统性能和识别准确度。
Abstract
In this paper, we propose a novel family of
windowing technique
to compute Mel Frequency Cepstral Coefficient (
mfcc
) for automatic
speaker recogn
→