1

My understanding of latency is that, if I'm recording vocals/guitar via a line-in, there will be some delay between my playing/singing and it registering with the computer; therefore, if the latency is really high, this means that the computer will think I sang a note slightly after I actually did. Upon playback, then, I'll be off-beat and will have to nudge my vocal track slightly earlier, in order to have my downbeats line up with the computer's timeline (and any other tracks that are on-beat). (Please correct me if I'm wrong about this.)

I'm running my instruments through an audio interface so I don't really notice any latency -- it registers at 24ms through a Focusrite @ 512 samples. When I adjust the driver error compensation, though, I would expect it to affect something. Here's what happened:

+300ms overall latency: Recorded myself going "dut" on each eighth note. Playback is in-time.

-300ms overall latency: Reocrded myself going "dut" on each eighth note. Playback is in-time.

Shouldn't both of my vocal tracks be shifted slightly ahead of, or behind, the beat, when I intentionally set the overall latency so far off?

I'm asking here because I figure I must be wrong about something, and I don't know where my misunderstanding lies.

  • 2
    I don't know Ableton Live, but it sounds like it just uses the correct latency compensation regardless of what you manually choose. Really, any software should just do this without you having to think about it; it's mostly because the industry standard Pro Tools failed to implement this that many engineers still feel latency “must always be compensated by hand”. – leftaroundabout Dec 18 '17 at 23:29
  • 2
    @leftaroundabout - agreed. The only time these days you ought to be able to hear the latency is if you're using play-through, i.e. listening to what you're recording as you're recording it. After that, the software just slides it to where it ought to be, seamlessly. – Tetsujin Dec 19 '17 at 7:38
  • How does it work during live monitoring? If it play the mix earlier/later, wouldn't I adjust my playing too, to compensate? Then we still have a timing problem, only instead of it going mix@20ms, monitor@50ms, we have mix@25ms, monitor@55ms. Wouldn't they both be shifted later in time, but retain the same timing difference in relation to each other? – Alex G Dec 19 '17 at 8:12
  • 2
    @AlexG there's no way to compensate latency in monitoring, so the preferred option is to use a signal path that doesn't actually pass through the workstation – either analogue split of the signal before the interface, or digital latency-free routing within the interface. Of course that precludes you from hearing virtual effects rendered in the machine; if that's necessary, the only option is to make the buffer size so small (64 or 128 samples) that the latency becomes almost inaudible. This should be possible on a fast computer if the FX load isn't too high. – leftaroundabout Dec 19 '17 at 10:32
2

Upon playback, then, I'll be off-beat and will have to nudge my vocal track slightly earlier, in order to have my downbeats line up with the computer's timeline (and any other tracks that are on-beat). (Please correct me if I'm wrong about this.)

You’re wrong about that. The computer knows how much input latency there is and automatically records the new audio at the correct point in the timeline for when you played it while you were overdubbing. No need to nudge it afterwards.

There are at least two situations where latency can still be a factor. The first is where you are monitoring through the computer as you record. The problem there is round-trip latency. The computer can compensate for the input latency and place your audio in the correct spot, but the round trip latency means you pluck a string or press a key and there is a slight delay before you hear the sound. That often makes it heard for musicians to play comfortably and in time because the latency ruins the “hand ear coordination”.

The second situation is when you are playing audio out of the computer to a speaker or effect unit and then recording the results back in. In this case, you usually do have to nudge the audio back into place to compensate for the output latency, since the software usually doesn’t know that the output latency is a factor. I believe some DAWs have figured out how to deal with this also, but I’m not sure.

Also I recommend not messing with driver error compensation. It’s just as likely you’ll make things worse or have no effect. Set it back to automatic.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.