GainQuant

Quantizes the codebook gain using a two-stage conjugate-structured codebook.

Syntax

IppStatus ippsGainQuant_G729_16s(const Ipp16s* pSrcAdptTarget, const Ipp16s* pSrcFltAdptVector, const Ipp16s* pSrcFixedVector, const Ipp16s* pSrcFltFixedVector, Ipp16s* pSrcDstEnergyErr, Ipp16s* pDstQGain, Ipp16s* pDstQGainIndex, Ipp16s tameProcess);

IppStatus ippsGainQuant_G729D_16s(const Ipp16s* pSrcAdptTarget, const Ipp16s* pSrcFltAdptVector, const Ipp16s* pSrcFixedVector, const Ipp16s* pSrcFltFixedVector, Ipp16s* pSrcDstEnergyErr, Ipp16s* pDstQGain, Ipp16s* pDstQGainIndex, Ipp16s tameProcess);

Parameters

pSrcAdptTarget	Pointer to the target signal for adaptive codebook search vector [40], in Q11.
pSrcFltAdptVector	Pointer to the filtered adaptive codebook vector [40], in Q11.
pSrcFixedVector	Pointer to the pre-weighted fixed-codebook vector [40], in Q12.
pSrcFltFixedVector	Pointer to the filtered fixed codebook vector [40], in Q12.
pSrcDstEnergyErr	Pointer to the previous predicted energy error vector [4], in Q10.
pDstQGain	Pointer to the quantized gain (G_P and G_C ).
pDstQGainIndex	Pointer to the gain codebook indices, GA and GB.
tameProcess	Taming process indicator.

Description

The functions ippsGainQuant_G729 and ippsGainQuant_G729D are declared in the ippsc.h file. These functions use a two-stage conjugate-structured codebook. The function ippsGainQuant_G729_16s quantizes the gain using the 7-bit codebook, whereas the function ippsGainQuant_G729D_16s uses the 6-bit codebook. The gain codebook search is done by minimizing the mean-squared weighted error between original and reconstructed speech:

E = x^tx + g_p²y^ty + g_c²z^tz - 2g_px^ty - 2g_cx^tz + 2g_pg_c y^tz,

where x is the adaptive target, y is the filtered adaptive vector, and z is the filtered fixed vector; g_p is the pitch gain, g_c is the fixed gain.

Return Values

ippStsNoErr	Indicates no error.
ippStsNullPtrErr	Indicates an error when one of the specified pointers is NULL.
ippStsRangeErr	Indicates an error when tameProcess is not 0 or 1.