首页> 外国专利> AUDIO CODE METHOD BASED ON MULTI-BAND EXITATED MODEL

AUDIO CODE METHOD BASED ON MULTI-BAND EXITATED MODEL

机译:基于多频带激励模型的音频编码方法

摘要

In the speech coding method, the voiced sound component and the unvoiced sound component can be mixed in all frequency bands to provide high quality voice service. The weighting of the voiced excitation signal spectrum V (n) and the unvoiced excitation signal spectrum U (n) Modeling the sum to predict the excitation signal spectrum E (n), and Predicted spectral envelope And generating a synthesized sound spectrum by estimating the synthesized sound spectrum, and estimating a function p (n) representing a mixing ratio in the frequency domain by using an analysis method or an analysis synthesis method. We propose a MMBE speech model that allows the mixing of voiced and unvoiced components in all frequency domains, and the application of synthesis methods and simple assumptions to estimate the model parameters by minimizing the RMS error between the original and synthesized speech spectra. In this paper, the frequency domain mixing ratio was estimated without comparing with the threshold through an analytical synthesis method that estimates the model parameters with much less computation amount than the synthesis method.
机译:在语音编码方法中,可以在所有频带中混合浊音成分和清浊音成分以提供高质量的语音服务。浊声激励信号频谱V(n)和清声激励信号频谱U(n)的加权对总和建模,以预测激励信号频谱E(n),并预测频谱包络,并通过估算合成的声谱来生成频谱,并通过使用分析方法或分析合成方法估计表示频域中混合比的函数p(n)。我们提出了一种MMBE语音模型,该模型允许在所有频域中混合浊音成分和清浊成分,并应用合成方法和简单假设通过最小化原始语音频谱和合成语音频谱之间的RMS误差来估计模型参数。在本文中,通过解析合成方法估算频域混合比而无需与阈值进行比较,该方法以比合成方法少的计算量来估算模型参数。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号