This　paper　made　further　research　in　audio　coding　based　on　adapted　wavelet　packet,　and　optimized　the　exist　basis　selection　algorithm　in　wavelet　packet　transform.　A　new　improved　adapted　wavelet　packet　transforming　algorithm　was　implemented　in　view　of　real-time　computing　complexity　and　energy　of　audio　signal　mainly　distributed　on　low　frequency.　The　algorithm,　which　adopts　zero　coding　based　on　psychoacoustic　model,　mabes　the　systems　has　a　compression　at　a　large　degree　on　audio　signal　keep　anticipant　real　time　performance　and　objectively　resume　SNR.