On Sat, 30 Aug 2014, dale wrote:
On Fri, 2014-08-29 at 08:12 -0700, Len Ovens wrote:
Intel says that nobody really needs that low of a latency anyway in
response some sound cards problems with one of their implementations of
USB3. (the problem was USB2 cards with USB3 ports which are supposed to be
compatable)
In recording latency is not the issue it is with live work. For recording
latency needs to known and constant and reasonably low for monitoring. I
have heard/read people who say what does it matter if you move your head a
foot or two closer or father from the speaker? But in live work if the
same audio comes from different places it is called a filter.
Yes if it is coming from two sources of different distances you get a
comb filter but that's not what we're talking about with distance/delay
here. It's more absolute latency.
Examples I was presented at while at college/university were:
* A piano player striking a key and hearing a sound, from the downwards
motion to hearing the sound is about 6ms. Taking into account both
mechanical transference from key to string and the sound to the ear.
* For comparison this would be the same as a guitarist standing six feet
away from his guitar amplifier if we lived in a ideal world (where
electricity travelled the speed of light) but there is obviously
propagation delay plus any added by stomp boxes (s)he may have. This is
not a distance at which a guitarist finds it difficult to play!
I sometimes think the hunt for super-low latency is a bit absurd! 3ms,
to give you a 6ms round trip, should be a workable amount for pretty
much anybody and most I expect could cope with quite a lot higher (not
many working methods require the full round trip!)
Lets take my ice1712 based device. With jack set -p16 (as low as it can
go), there is 3ms round trip:
1ms each way for the ice1712 internal monitor mixer and whatever else
routing, plus .6ms each way for 3.2ms. So set -p32 for 4.4ms or -p64 for
6.8ms.
Most USB2 IFs can do that fine in a USB2 only world. Some of
intel's USB3 inplementations make -p64 for a USB2 unit full of xruns. So
the "low" latency I was talking about is somewhat higher than 6ms.
But full round trip is probably also the time when delay effects (com
filter and echo) become important, such as using a PC as an LMS or FOH
mixing desk. I would consider these specialist cases though.
Not that many years ago, the norm in computer based audio was sequencing
and digital to disk (or memory) recording of audio was a specialist
case... because the HW couldn't handle it. Now the HW can handle it and we
do much more than just record audio to disk. So maybe the same thing is
true here, The HW can't handle it so it becomes special case. The
difference (from what I can tell) is that in this case, we seem to be
going backwards. Latency is increasing. The reason it is increasing seems
to be that someone has decided it is not important. So what might make
sense in a very low latency environment, may never happen based on that
decision.
What I am saying, is that to become standard application, the HW has to be
cheap enough for most people to own it.
--
Len Ovens
www.ovenwerks.net
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user@xxxxxxxxxxxxxxxxxxxx
http://lists.linuxaudio.org/listinfo/linux-audio-user
