"Unfix" the filter length calculation
commitfdd5070c3b1945eec3553e5395faf1a99c5883f3
authorChris Robinson <chris.kcat@gmail.com>
Fri, 18 Aug 2017 09:48:32 +0000 (18 02:48 -0700)
committerChris Robinson <chris.kcat@gmail.com>
Fri, 18 Aug 2017 09:48:32 +0000 (18 02:48 -0700)
tree586ab37375720faf112a065cb182899819fccfb1
parent5048956ff02ec6d84aed2f79c65e4c130adf6e43
"Unfix" the filter length calculation

Partially reverts 3f3a3ac4f1d069542ca2399a8b5e63d9d1a4df3b. The l*2 + 1 is
correct when you want an odd number of sample points, which avoids an
unnecessary phase offset in the fitler. However, the rounding is still needed
to calculating the left offset (l), or else the transition width can increase
with an odd-numbered order.
utils/makehrtf.c