GainCodebookSearch

Finds pitch and fixed gains.

Syntax

IppStatus ippsGainCodebookSearch_G729_32f( const Ipp32f* pSrcCorrFactors, Ipp32f valPredictedCodebookGain, const int* pSrcCand, Ipp32f* pDstIdxs, Ipp32s valTimeFlag);

IppStatus ippsGainCodebookSearch_G729D_32f (const Ipp32f* pSrcCorrFactors, Ipp32f valPredictedCodebookGain, const int* pSrcCand, Ipp32f* pDstIdxs, Ipp32s valTimeFlag);

Parameters

pSrcCorrFactors	Pointer to the correlations.
valPredictedCodebookGain	Predicted codebook gain.
pSrcCand	Pointer to the pre-searched candidates.
pDstIdxs	Pointer to the output indexes.
valTimeFlag	Taming process indicator.

Description

The functions ippsGainCodebookSearch_G729 and ippsGainCodebookSearch_G729D are declared in the ipps.h file. As part of LSP quantization algorithm (refer to ippsGainQuant_G729 for details), these functions operate in floating-point arithmetic and perform selection of the gain vector in the codebook GA and GB in the range limited by pSrcCand[0] and pSrcCand[1]. The search algorithm consists of minimizing the following expression:

E = pSrcCorrFactors[0] x G_p² + pSrcCorrFactors[2] x G_c² + pSrcCorrFactors[1] x G_p + pSrcCorrFactors[3] x G_c + pSrcCorrFactors[1] x G_p x G_c,

where G_pand G_c are optimum pitch and code gains, respectively, in selected domain.

The ippsGainCodebookSearch_G729_32f function uses the 7-bit codebook, whereas the function ippsGainCodebookSearch_G729D_32f uses the 6-bit codebook.

Return Values

ippStsNoErr	Indicates no error.
ippStsNullPtrErr	Indicates an error when one of the specified pointers is NULL.
ippStsRangeErr	Indicates an error when valTimeFlag is not 0 or 1.
ippStsRangeErr	For the G729 function flavor: indicates an error when pSrcCand[0] is not in the range [0,4) or pSrcCand[1] is not in the range [0,8). For the G729D function flavor: indicates an error when pSrcCand[0] or pSrcCand[1] is not in the range [0,2).