SPEECH SYNTHESIS BASED ON SINUSOIDAL MODELING
SPEECH SYNTHESIS BASED ON SINUSOIDAL MODELING
复制标题
基于正弦建模的语音合成
DOI：
--
复制
发表时间：
2004
期刊：
影响因子：
--
通讯作者：
K. Kumar
中科院分区：
文献类型：
--
作者：
K. Kumar
本报告介绍了语音合成的一些方法及其相关问题的简要概述。详细讨论了利用语音波形的正弦表示来产生合成语音。正弦模型基于从短时傅立叶变换中提取正弦波分量的振幅、频率和相位，并将其用于合成语音的产生。该报告是基于语音修改，以实现理想的语音特性，使用音调缩放，时间缩放和频谱扭曲进行的一些工作。获得愤怒、情绪、喜悦等表情，合成语音是目前人们关注的主要领域之一。这方面的表达，使语音使用正弦模型的简要介绍。
This report presents an introduction to speech synthesis with a brief overview of some methods and their associated problems. The usage of sinusoidal representation of speech waveform for producing synthetic speech is discussed in detail. The sinusoidal model is based on extracting the amplitudes, frequencies, and phases of the component sine waves from the short-time Fourier transform and using them for the production of synthetic speech. The report is based on some of the works carried out on speech modification to achieve desirable characteristics in speech using pitch scaling, time-scaling, and spectral warping. Obtaining expressions like anger, emotion, joy etc., in synthetic speech is one of the major areas of concern these days. This aspect of bringing expression into speech using the sinusoidal model is presented in brief.