Skip to content

Proposed fix for 810

vaillancour requested to merge 810-issue-in-decoding-transition-mode-at-32-kbps into main

A fix is proposed to solve #810 (closed). It is not bit exact, but most of the time the difference is minimal, except when there is a large amplitude onset coded with TC that could have saturated before and now it is not. The fixed point synthesis gets closer to floating point, but from a perception point of view, the fix doesn't have a big impact

Merge request reports