R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1]   Go Down

Author Topic: Digital System Alignment  (Read 727 times)

LawrenceF

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 89
Digital System Alignment
« on: December 18, 2004, 01:25:58 pm »

This is a copy of a post of mine from the 3daudio forum that may be of use to some here...

This is a VERY long post so please bear with me.  

I wonder how many people with hybrid systems and/or stand alone daws actually test to see how (where) the daw actually places recorded audio.  We all know that analog decks required alignment (for different reasons) but what about your digital system's (daw) timing and measurements?

Just curious.  I've tested my system through and through to the point of being anal and made a final adjustment to my daw of -29 samples to make sure that everything I track through my d8b to daw is sample accurate in relation to where it was heard during tracking.  I can now play an audio track from daw through the d8b and record it back to daw and it will completely cancel with the original.  SX automatically adjusts every track to make up for the hardware's input latency plus the other 29 samples that I entered in the hardware setup screen.

Now I know everything I record is being played back exactly (to the sample) where I heard it as it passed through my digital console during tracking.

I suspect that many so called sonic flaws in daws that some people complain about, loss of imaging, "feel" etc, could be due to a lack of this kind of digital "time alignment" with or without a supporting digital (or analog?) mixer.  Many take for granted that the software does exactly what it claims to do perfectly.  Often that's not the case and it may be (usually is) just a little bit off.  Enough to cause some of these issues?  Maybe.  The measurements (ASIO anyway) rely on the accuracy of the hardware reporting so most daws have options to manually adjust for that if it's a little off or if there is a "wildcard" like the throughput latency of a  digital mixer.

Edit: Of course that option is useless if you don't know how much to adjust for! Smile  You gotta take the time to find out.

If you can't send a digital track from your daw (and through your digital console?) and back in again without conversion and have the daw place it in the exact same spot where it will cancel 100% you might want to maunally adjust for that.

Then again who cares if it sounds good?                               Smile

... a little off topic but of possible value to SX users..

During these tests I found a serious bug in my system.  Unless I manually reset the audio engine after loading a project, SX (2.01) does not automatically adjust for input latency even though that option is checked.                           Sad  Could be a bug in SX or it could be a stray plugin that's mucking things up.  Once I reset the engine I can work with sample accurate recordings.


Back on topic...

This timing adjustment/correction effectively keeps overdubs exactly where they were performed down to the sample.

When recording the basic tracks at the same time, all at once, this is not an issue.  The daw could be off by 9ms and it wouldn't matter because all of the tracks would be off by 9ms and it would still capture the true performance timing.  The only timing reference the players used were each other and it's tight and all good.  Although if you had a drummer playing against a click track you'd hear it immediately on playback if you let the click continue playing with the tracks.

It's when overdubbing against those daw tracks where this becomes (subjectively) important.  The overdubs would be 9ms later than the basic tracks.  The effect doesn't actually repeat (compound) unless you use a soloed overdubbed track as the timing reference for a new overdub.

Example:  You record the song to a stereo track to conserve cpu while overdubbing vocals.  You record your overdubs  while that stereo track plays.  Guess what?  When you turn the original tracks back on those overdubbed vocals will be 18ms late because the timing reference you used was already 9ms later than the first!  My use of 9ms as a reference system latency is random but you get the point, this is one situation where the number could double during a recording session.  If it was 2ms to start with you may not hear it when it hits 4ms, if it starts with 4-5ms you should hear it at 8-10ms with timing critical tracks.  The entire feel will change from what you just heard being recorded.  Not good.

If you've ever shifted vocals back a little during mixdown to get it to "feel" better (I have) this could be why.  It's not immediately noticable during tracking because the actual performance sounds fine.  When you play it back from the daw the timing of the overdubs changes because the daw placed the audio in a different place from where you just heard it live against the tracks.  If it's very subtle you may not notice it until much later or not at all.  I would often feel something was "different" but couldn't put my finger on it.

If you tracked midi parts to daw (midi playing alongside recorded audio) and the song "felt"  a little "different" this is probably why.  Especially if they are midi drum parts.  Where did the "groove" go?

If you typically mix (not render, mix through a digital console) back to daw, open up a project with a mix file still in it and play the mix along with the individual tracks.  You'll immediately see and hear what I mean.

During overdubbing every audio track you record is going to be off in relation to the other audio tracks playing from the daw when it/they passed through the digital console during tracking, when you play it back, by the amount of time that's causing the flamming or phasing you'll probably hear when you play your mix track against the original tracks.

If you strap something with a huge latency across your LR during mixdown your actual figure will be less than what you'll hear.  You'll have to test with a click track to find it anyway.  Record a click to the daw and then play that click through your console or whatever you use to monitor and record it back to another track.  Zoom way in and compare the start times of the hits.

Surprised?  I was.  Especially when the "adjust for input latency" option wasn't working right.

Edit: It occured to me that some newer daw users may not know how to find the number to adjust for so here it is...

To get the sample value change your timeline display to samples, zoom WAY in to the first hit and write the down the sample number where that hit landed, zoom WAY in to the second hit and write that sample number down.  Subtract the first number from the second and enter result into your daw setup as the sample adjustment.  If it's a "record latency adjustment" you may not have to enter it as "- xx", you'll know when you copy the click again and see where it lands.  You may have to tweak +/- 1 or 2 samples to get full cancellation or as close as possible.  When you record the click again and zoom in you'll see they'll be at almost identical spots with the difference being too small for a +/- 1 sample adjustment to get closer.  Reverse the phase of one...  nothing... at least nothing audible! Repeat it 3 or four more times.  Cancelled every time?  You're all set.  Go make music.  You've just aligned your daw for perfect overdubs!


Thanks for reading the VERY long post.  I hope it was helpful to some here.

Edit: I would recommend anyone using a daw to perform this basic test and make sure your system is as close to sample accurate as you can make it.

Lawrence
Logged
Pages: [1]   Go Up