Skip to content

[allow regression]Add fix FIX_1416_MDCT_CLASSIFIER for TCX/HQ classifier

Closes #1416 (closed)

The calculation of avrg_h and avrg_l in mdct_classifier_ivas_fx.c sacrifices the precision already in the summation. By just moving the right shift out of the loop, the precision is improved and the decision was aligned.

Merge request reports

Loading