Research On Stereo Encoding And Decoding Algorithm Combined With EVS

With the rapid development of IP networks and multimedia services,the voice quality provided by the existing voice codecs has been unable to meet users' needs because of the limitations of audio bandwidth and non-voice signal processing.In order to adapt to the future broadband mobile communications,voice codecs should support high-definition,full-band voice and audio.Beside this,immersive communications,online live and other occasions also require a new generation of codecs to support stereo signals and achieve high quality codec.The stateof-the-art full-band voice codec,Enhanced Voice Services Codec(EVS),dramatically improves the quality of the communication by improving voice and audio coding techniques.EVS makes sound after transmission closer to the original sound by providing full band audio transmission to improved speech intelligibility and listening comfort.However,EVS currently does not support stereo signal processing.In order to be more widely used in daily life,the stereo signal processing technology are urgently needed in EVS.After many years of development of codec technology,a variety of stereo signal processing technologies based on different principles have emerged.These technologies have tended to be mature and have more or less patent problems.Therefore,new development directions need to be explored.Principal component analysis(PCA)is widely used in data dimensionality reduction and has certain similarities with the codec.Combined with EVS codec,this paper presents a scheme of PCA in codec based on the parametric stereo processing technology,which includes the modules of signal preprocessing,PCA,quantization and stereo processing.The algorithm is evaluated from the aspects of compression effect,audio reconstruction quality and spatial sense.The results of tests show that this PCA-based stereo codec algorithm has a high feasibility and a lot of room for development.