A perceptually biased linear prediction scheme is proposed for audio coding. Using only simple modifications of the coefficients defining the normal equations for a least-squares error, the spectral masking effects are mimicked in the prediction synthesis filter without using an explicit psychoacoustic model. The main advantage of such a scheme is reduced computational complexity. The proposed approach was implemented in a Laguerre-based linear prediction scheme, and its performance was evaluated in comparison with a Laguerrebased linear prediction approach controlled by the ISO MPEG-1 Layer I–II model, as well as with one of the latest spectral integration–based psychoacoustic models. Listening tests clearly demonstrate the viability of the proposed method.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.