Speech VTF similarity property and its application on coding analysis

Similarity is one of the most important properties of speech, especially of voiced speech signal. In this paper, speech similarity property is analyzed and applied to speech coding. Firstly, vocal tract filter (VTF) and its similarity of voiced speech are analyzed. Secondly, similarity property is applied on speech coding and a threshold is set up to determine whether the linear spectrum pair parameter of each frame needs to be sent. Finally, simulation experiments show that this method could reduce the bit rates about 28.5% for Mixed-Excitation Linear Predictive without distinct damage to the synthesis speech quality.