wine/gecko
0
0
Fork 0
mirror of https://gitlab.winehq.org/wine/wine-gecko.git synced 2024-09-13 09:24:08 -07:00
Code Issues Packages Projects Releases Wiki Activity

b=865241 don't try to compute a phase for group delay from the DC and nyquist components of the DFT r=ehsan

Browse Source 
Given imagP[0] is really realP[halfSize], it doesn't make sense to calculate a
phase from placing the DC and Nyquist components in different directions.
The DC component doesn't have a phase offset and lastPhase is already
initialized to zero.

This makes extractAverageGroupDelay() consistent with addConstantGroupDelay()
and to some extent interpolateFrequencyComponents().

The effect of the bug would have been small for reasonably sized FFTs.

--HG--
extra : rebase_source : b2931d625bd7e014f123b576f28aad7f719c585d
This commit is contained in:
Karl Tomlinson 2013-08-08 21:37:36 +12:00
parent a7e95885f3
commit 0e2cb4c001
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
content/media/webaudio/blink/FFTFrame.cpp
View File

@ -193,7 +193,7 @@ double FFTFrame::extractAverageGroupDelay()
const double kSamplePhaseDelay = (2.0 * piDouble) / double(fftSize());
// Calculate weighted average group delay
for (int i = 0; i < halfSize; i++) {
for (int i = 1; i < halfSize; i++) {
Complex c(realP[i], imagP[i]);
double mag = abs(c);
double phase = arg(c);