cepstral
共 5 篇文章
cepstral 相关的电子技术资料，包括技术文档、应用笔记、电路设计、代码示例等，共 5 篇文章，持续更新中。
基于STM32单片机的嵌入式语音识别系统设计
<p>设计了一款以STM32F103为核心的自然语言识别系统,为满足实时语音识别系统对内存资源和运算速度的要求,基于硬件资源合理设计语音处理算法,在嵌入式平台上实现了对孤立词语的语音识别。首先根据背景噪声和语音信号的时域特征差异设定相应门限值,从而实现了对语音信号的端点检测。然后针对语音识别中传统梅尔倒谱系数对语音的高频信息敏感度较低,对语音信号分别提取梅尔倒谱系数(MFCC)与翻转梅尔倒谱系数(
Signal Processing for Telecommunications
This paper presents a Hidden Markov Model (HMM)-based speech<br />
enhancement method, aiming at reducing non-stationary noise from speech<br />
signals. The system is based on the assumption that the
cepstral analisys for pitch and F0 detection
cepstral analisys for pitch and F0 detection
CEP2POW convert cepstral means and variances to the power domain
CEP2POW convert cepstral means and variances to the power domain
Generalized Mel frequency cepstral coefficients for large-vocabulary Speaker-Independent Continuous-
Generalized Mel frequency cepstral coefficients for large-vocabulary Speaker-Independent Continuous-Speech Recognition 关于MFCC算法的很好的英语文章