摘要
频谱分析技术在语音信号处理中得到了广泛应用。短时频谱分析是各种语音频域处理方法的基础,广泛应用于语音编解码、语音合成、语音识别等领域。通过时频分析理论,对语音短时频谱分析与合成进行理论推导,表明语音短时频谱分析与合成可以通过加窗和滤波语音信号实现,并据此提出具体算法,按照该算法处理实际语音数据。合成语音与原始语音相比,差异很小。
Spectrum analysis technology has been widely used in speech signal processing. Short-term spectrum analysis is the basis of various speech frequency domain processing methods, which is widely used in speech coding and decoding, speech synthesis, speech recognition and other fields. Through the time-frequency analysis theory, the short-time spectrum analysis and synthesis of speech are theoretically deduced. It shows that the short-time spectrum analysis and synthesis of speech can be realized by adding windows and filtering speech signals. Based on this, a specific algorithm is proposed to process actual speech data according to the algorithm. The difference between synthetic speech and original speech is very small.
作者
吴树兴
刘新红
Wu Shuxing;Liu Xinhong(Beijing Information Technology College, Beijing 100015, China)
出处
《信息与电脑》
2019年第15期50-52,共3页
Information & Computer
基金
北京市教育委员会科技计划面上项目“汉语语音合成中影响自然度的若干关键技术研究”（项目编号：KM201410857001）
关键词
语音短时频谱分析
语音短时频谱合成
窗口滤波器
speech short-time spectrum analysis
speech short-time spectrum synthesis
window filter
