Introduction - If you have any usage issues, please Google them yourself
Abstract: speech enhancement algorithm based on phonological state model is a hot topic of speech signal processing. When the normal LPC voice model is modified, two speech models are obtained: the time variant AR model and the time-variant double AR model. However, when using these models to enhance the speech, the pronunciation is not taken into account in the voice voicing and voicing. This paper introduced the voice clear voiced state space model, this model in describing the voice than time-varying AR model, the time-varying double AR model, and the physical meaning is obvious At the same time in the speech signal with noise prediction purity voice signal, the forgetting factor and particle filter algorithm is introduced to reduce the computational complexity, reduce the computational complexity. The experimental results show that the improved speech signal-to-noise ratio is improved. It is superior to the traditional LPC model.