+2
−0
lib_com/fft_cldfb_fx.c
0 → 100644
+1070
−0
File added.
Preview size limit exceeded, changes collapsed.
Loading
Significance: ============= [x] Changes pertaining to MDCT/IMDCT fixed point conversion [x] Added new unit tests under FFT - includes testing of fft_rel function [x] The version of fft_rel_fx taken from EVS fails conformance [x] Two alternate versions of fft_rel_fx are been created version 1: By extending 16-bit input of evs fft_rel_fx to 32-bit. This fails Conformance for 512 point FFT. version 2: Through fixed point conversion of fft_rel function in IVAS This passes conformance for all lengths supported by fft_rel. [x] Converted ifft_rel to fixed point as well. [x] Input scaling is needed sometimes for the random input generated as the input values generated are at times very small causing relative error in output to exceed 0.005% (threshold for unit tests currently) [x] Changed real basops to complex basops in few FFT functions. Testing: ======== [x] Unit tests of FFT and MDCT Known Issues: ============= [x] Unit test of RFFTN for 640 length is failing with random input.
File added.
Preview size limit exceeded, changes collapsed.