shameem
Member level 5

hi,
i am using vdsp ver 4.0 and a voip board having BF 533 procesor. My system is xp. Actually i need to use the ifftN_fr16 and cfft_fr16 library functions of vdsp 4.0.
My problem is that when i take 2048 frquency domain samples of a signal and use ifftN_fr16 followed by cfftN_fr16 functions, i should get the same frequency domain samples of the original signal, but i get very small values of the corresponding samples which are very inaccurate and cannot be recovered by scaling by a constant factor.
If anybody have idea to modify these functions or if anybody have modified routines for these functions, please send me.
i am using vdsp ver 4.0 and a voip board having BF 533 procesor. My system is xp. Actually i need to use the ifftN_fr16 and cfft_fr16 library functions of vdsp 4.0.
My problem is that when i take 2048 frquency domain samples of a signal and use ifftN_fr16 followed by cfftN_fr16 functions, i should get the same frequency domain samples of the original signal, but i get very small values of the corresponding samples which are very inaccurate and cannot be recovered by scaling by a constant factor.
If anybody have idea to modify these functions or if anybody have modified routines for these functions, please send me.