Skip to main content

I understand what latency is, and I know it perfectly well when I have the monitor button in Cubase on, but what I am confused about is why very low latency should be an issue, like when software is marketed as having very low latency. It seems that Cubase applies whatever inherent latency there is across all tracks, and I have never had any problem tracking. Even incredibly small changes in one side of a pair of double tracks is very obvious, and I have not yet heard defects I could attribute to latency, let alone what might manifest from the say 31 milliseconds of output latency my DAW claims to have.

So, why do people feel the need for very low DAW/plug in latency, if it is not a problem in real world conditions? Or, if it is a problem, exactly what kind of applications might latency become a more obvious problem if values are not sufficiently low?

Comments

hueseph Thu, 01/06/2011 - 19:59

Latency is a problem when you want to monitor your input with effects. If the latency is too large and you are using a virtual instrument it is a huge problem. You want to play in time with the backing track but if what you hear ends up being delayed 50ms or more, it can sound very off. In that sense the less latency the better. When you're recording audio it's not so much of an issue as long as you can enable direct monitoring but as soon as you start to factor in real time effects and virtual instruments, you want the lowest latency possible.

anonymous Fri, 01/07/2011 - 07:38

Same card. 10 millisecond buffer setting, with 21 and 31 milliseconds of respective input/output latency. This is the default setting of the USB 6. What do you think about these values? I changed the buffer setting to 1 ms, and see that the latency could be as low as 3 and 4ms, but perhaps not without suffering some issue.

TheJackAttack Fri, 01/07/2011 - 08:37

Delay in miliseconds is roughly 1ms/foot or slightly less rough 3ms/m. A live non amped musician that plays in a tight group knows its easier to play together when the group is physically close than across a 21' stage.

Now in a DAW there is automatic compensation for laying in the new track. Except the monitoring. That's why your direct monitoring remains on many interfaces. It is fairly difficult for those unused to it to, to be precise inre time while hearing the offset. The computer portion of the delay is simply due to processing ability versus resource requests from the memory controller.

TimOBrien Fri, 01/07/2011 - 11:54

jmm22, post: 360691 wrote: 21 and 31 milliseconds of respective input/output latency.

That's why you're hearing latency.
MOST humans can't hear much under 30ms in double-blind studies (you may be more sensitive) and you're right on the border at 31ms.

I keep my daw around 15-18ms and have no problem.

Knock the setting down (watching for any new problems like clicks, pops and dropouts) and you'll be fine.

NOTE: You do NOT want to crank it down all the way, either. Just enough to get rid of the latency.
Pushing it lower than you need to only stresses the cpu more, giving you less track/plugin counts or the click/pop/dropouts.

Remember that standard install settings don't mean much. You have to make it work properly...