Skip to content

Resolve "Double precision arithmetic in DFT stereo"

reutelhuber requested to merge 811-double-precision-arithmetic-in-dft-stereo into main

Reason why this change is needed

  • Use of float functions in DFT stereo instead of equivalent functions with double precision.

Description of the change

  • Change atan2() to atan2f() and floor() to floorf() in DFT Stereo-related code.

Affected operating points

  • Changing floor() to floorf() should always be BE. It might be possible that changing atan2() to atan2f() for computing IPDs could in rare cases change an IPD value which would affect bit-exactness of both bitstream and output.
  • BC is not affected

Related to #811 (closed)

Merge request reports