len at ovenwerks.net
Tue Jan 3 07:37:03 UTC 2012
I did find a copy of the HDA spec. It is a bus spec in many ways. It is
(like AC97... that is about the only likeness) centered on 48khz. This is
it's design sample rate and all other sample rates are derived from
that... except spdif, though it is handled at a multiple of 48khz in any
case. If there is DSP sample rate conversion it would be part of the codec
as the HDA controller doesn't deal with that itself. The way HDA deals
with rates greater than 48k is to send more than one sample per period. I
don't know, but it sounds as if fewer channels are available at higher
rates. I would guess that the PCI bus speed would have this effect on any
kind of audio interface. All sample rates are divided down from 24mhz. 48k
is divide by 500... 44.1k is 147/80000. I can see why the "digital" mics
are straight 48k as they would be able to run off of the sync signal with
no divider. It appears that unless a codec uses DSP conversions there are
none (the spec and bus timing tend to discourage it too). I guess Intel
learned from the earlier AC97 short comings.
HDA spec is available here:
Just wondering on the mic question... the one that has left normal and
right inverted, if the line from the mic to the codec is actually balanced
and would benefit from inverting the right channel before mixing the two.
If the mic was left stereo I could test that easy enough as my recording
software would let me invert a channel.
I am going through the AC97 spec now.
More information about the Ubuntu-Studio-devel