Function faad_imdct is called only for N =256 or N=2048 where divisor are 2 and 4. Called are relative simple functions passf2pos and passf4pos. I add calls faad_imdct followed by reverse function faad_mdct and for N=256 reversing is OK (small differences with numerical float point)
For N=240 is not OK. I add cos_sin tables for mdct192 and mdct200. For N=192 is still OK, for 200 results differs.
This means that error is for divisor 5 - passf5. This function is long and complicated. Where can I find algorithm description for divisor 5?
FAAD uses only 2^n length frames, but I want use FAAD's MDCT/IMDCT function in other project and want use different frame length.
Function faad_imdct is called only for N =256 or N=2048 where divisor are 2 and 4. Called are relative simple functions passf2pos and passf4pos. I add calls faad_imdct followed by reverse function faad_mdct and for N=256 reversing is OK (small differences with numerical float point)
For N=240 is not OK. I add cos_sin tables for mdct192 and mdct200. For N=192 is still OK, for 200 results differs.
This means that error is for divisor 5 - passf5. This function is long and complicated. Where can I find algorithm description for divisor 5?
FAAD uses only 2^n length frames, but I want use FAAD's MDCT/IMDCT function in other project and want use different frame length.