Linear prediction (LP) has traditionally been used in speech coding. Recently, variants of LP have also shown to be appropriate for audio coding. In this paper we introduce a new prediction scheme, called pure linear prediction (PLP), which combines important features from previous approaches. We show that the modelling capability of the PLP cna be tuned in a psychoacoustically relevant way making it suitablbe for speech and audio coding. Moreover, under certain restrictions, this new scheme is directly realizable, stable and retains the whitening property of conventional linear prediction. The processing of the prediction coefficients to perfrom operations such as quantisation, interpolation and spectral broadening is also addressed. As an example of the application of the PLP, we describe its use in the context of the sinusoidal coder proposed by Philips which is being standardised in "MPEG-4 Extension 2".
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.