Page 1 of 1

Theoretical Limit to Latency

Posted: Wed Jul 18, 2012 9:57 pm
by GroverGrooves
Hi guys,

I have a MOTU 8Pre, and I use Live. I've been using it on a pretty powerful Windows box (dual 3.0 cores) running XP, but I've recently bought my first Mac, dual 2.8 with 8GB or RAM, and I plan on moving all of my audio and music apps over to that, and retiring or retooling my Windows box.

I am trying to get my latency down to a level at which I can record myself playing guitar, and have it feel natural, without any noticable delay. In my Audio Settings I turned the buffer down to 128 samples, and it says my latency is about 3-4 miliseconds, but I swear I still hear a delay between the time I pick a string, and the time I hear it in my headphones. Actually I don't really notice any difference between 128 samples and 512.

Is this a limitation of my Audio hardware, my drivers or my computer? I know I can buy a new audio interface with real-time monitoring, but I'd rather hear all of the audio effects while I'm playing the parts, and plus who wants to spend more money?

Any ideas or suggestions? Thanks!

Aaron

Re: Theoretical Limit to Latency

Posted: Wed Jul 18, 2012 10:13 pm
by Tone Deft
it's difficult for a DAW to report the exact latency, you need a latency checker, I forget the name.

sounds takes 1 millisecond to travel 1 foot.

play a mono track out of your speakers. stand next to one speaker and away from the other. do you hear a delay? if you're like most of us your speakers are probably 5-6 feet apart. can you hear that 5-6 milliseconds of delay?

I don't look at the numbers, I listen.

currently my computer is powerful enough to get playable latency. prior to that, when my latency was high I'd use Direct Monitoring on my sound card, IOW I listened to the input of the card, not Live when I played.

be sure to do the tutorial on delay compensation.

Re: Theoretical Limit to Latency

Posted: Wed Jul 18, 2012 11:31 pm
by 3dot...
4 ms input latency..
+
variable processing latency ( software fx )
+
4 ms output latency..
=
more than 8 ms total..

you need to go to 64 or less sample buffer if possible..to get around 4-5 ms. total
you'll still feel it but it's way easier to compensate..

Re: Theoretical Limit to Latency

Posted: Thu Jul 19, 2012 12:27 am
by simmerdown
..or go back to pc ,xp and asio

Re: Theoretical Limit to Latency

Posted: Thu Jul 19, 2012 1:19 am
by login
LOL seriously.

10 ms is the standard latency of monitors in a Live show.

Re: Theoretical Limit to Latency

Posted: Thu Jul 19, 2012 1:40 am
by simmerdown
try recording an instrument to a beat at 10ms.....no

Re: Theoretical Limit to Latency

Posted: Thu Jul 19, 2012 12:04 pm
by GroverGrooves
Thanks everyone for your input.

Re: Theoretical Limit to Latency

Posted: Thu Jul 19, 2012 2:20 pm
by Tarekith
simmerdown wrote:try recording an instrument to a beat at 10ms.....no
I record guitar in just about all my songs with my latency set to 256 samples for years now, never notice any delay or I must be compensating manually.

Re: Theoretical Limit to Latency

Posted: Thu Jul 19, 2012 2:48 pm
by simmerdown
...that was refering back to login's comment, 10 ms is no good to rec, it'll make you insane...live show latency is not relevant

i rec at 256 also, and if there is a delay i cant hear it....humans generally can detect greater than 3 ms, so it must be under that

Re: Theoretical Limit to Latency

Posted: Fri Jul 20, 2012 9:24 am
by twisted-space
simmerdown wrote:try recording an instrument to a beat at 10ms.....no
Ermmmm, sorry but yes.
Thats like standing 10 ft from your drummer while playing live and for most of the drummers I've worked with, that's way too close.

Re: Theoretical Limit to Latency

Posted: Fri Jul 20, 2012 1:52 pm
by simmerdown
it takes 10ms for sound to travel 10 ft? i think not...i'll do some math

but anyway, i was saying, set your delay to 10ms, put on some headphones and try to record, not happnin

or take one instrument and move the track delay 10ms, hows that sound?

Re: Theoretical Limit to Latency

Posted: Fri Jul 20, 2012 2:51 pm
by twisted-space
simmerdown wrote:it takes 10ms for sound to travel 10 ft? i think not...i'll do some math
Have you done your math yet?

In air, sound travels @ roughly 1000ft/s, depending on air pressure, humidity, temperature, etc.

Re: Theoretical Limit to Latency

Posted: Fri Jul 20, 2012 5:24 pm
by simmerdown
great, thanks for that

sidetopic to the op, but good to know for real