5.2.2 Levinson‑Durbin algorithm

26.1903GPPAdaptive Multi-Rate - Wideband (AMR-WB) speech codecSpeech codec speech processing functionsTranscoding functionsTS

The modified autocorrelations and are used to obtain the LP filter coefficients a_k,k=1,…,16 by solving the set of equations.

( 9 )

The set of equations in (9) is solved using the Levinson-Durbin algorithm [2]. This algorithm uses the following recursion:

The final solution is given as ,j=1,…,16.

The LP filter coefficients are converted to the ISP representation [4] for quantization and interpolation purposes. The conversions to the ISP domain and back to the LP filter domain are described in the next two sections.