Recorded audio placement on timeline is not consistent

[aanon]

Using DP 8.06 64 bit
Windows 7 64 bit
MOTU 828 mk3 hybrid connected via USB, no USB hub involved, no other USB devices
Latest MOTU drivers

- -

I'm fussy about timing. I don't do MIDI with DAWs at all. I record an audio click track into a song first (at the same time as audio of anything I've got hardware sequenced) and then I overdub instrument tracks (all as audio) one by one on top of that. I use Cuemix for "zero latency" monitoring during recording.

Even doing this, I often didn't feel that the timing of my new instrument tracks on playback was quite the same as when performed. I don't have perfect timing, of course, but I am sensitive to it and I never sensed this before using my Roland VS2000cd multitrack or analog tape machines way before that.

I started to wonder whether the newly recorded audio was being put in precisely the right place on the timeline or not. So I did a "loopback" experiment to take the human factor out of it:

- Take my recorded audio click track from DP out a spare MOTU analog output
- Bring it back into DP via a MOTU input and record back into the song
- Zoom in and compare the position of the re-recorded click track vs the original click track

I initially found that the re-recorded click track tended to be placed behind the original click track by about 250-270 samples.

So I worked with audio I/O timing fine tune and managed to reduce this lag somewhat. However, I never got the re-recorded audio to align exactly with the original audio click track.

More worryingly than that, the discrepancy between them wasn't the same every time I repeated the test. I could sometimes get to a point when I thought I had the fine tuning set very well, and get several takes that were nearly perfectly placed, then on the next take my re-recorded click track could suddenly be out by 100 samples or more. Also, when I tried to re-record the click and also simultaneously record additional audio tracks, this also seemed to affect the amount of discrepancy observed.

So, as things stand I have concluded that I can't really rely on DP (or perhaps I should say my total current approach and setup) to record something I played in the appropriate position on the timeline.

What I currently do as a workaround (not being aware of other solutions) is to re-record the click like this EVERY TIME I lay down any instrument track, and then use this as a reference for timing (i.e. to slide both the re-recorded click and the instrument track by whatever amount is needed to get the clicks to line up perfectly).

This works, but it's tedious and there has to be a better way! So I'm here to ask the help of the community. Are the symptoms I'm describing common, or an aberration? Have other overcome similar issues? Where should I start in resolving these issues?

I would very much appreciate any guidance the community can offer.

Aanon

[HCMarkus]

It seems likely that the variation is coming from DP's generated click, not the audio you record. Might I suggest recording the click at the start of a project, then using the recorded click as your timing reference?

[aanon]

Thank you for the reply HCMarkus,

I have not been relying on dp to create the click track live. Rather i have recorded my own click track as audio, much as you have suggested. This is necessary in order to compare original vs re-recorded tracks.

[HCMarkus]

Curious what sample rate you are working at.

Are you seeing timing issues note to note or is the entire recording uniformly delayed or advanced?

Also, are tracks recorded simultaneously (i.e. stereo or multitrack, like drums) in sync between tracks?

To put 100 samples into timing perspective, at 44.1k, it is like standing about 2 feet from a sound source. Not saying your point is moot, DAWs should deliver perfect timing but, unless we are talking simultaneously recorded track-to-track discrepancies (where phase issues between tracks could affect blended sound), monitoring with an all-analog signal path wearing headphones would make a bigger difference.

[aanon]

Audio is running at 44.1K.

Any tracks that go down "onto tape" simultaneously are in time with each other.
Also, my re-recorded click track, once manually aligned with the original at any single point, is perfectly in time with the original click track for the whole duration of the recording. So yes, digital audio recording is doing its job in that sense.

I acknowledge that in my original post I was talking about small discrepancies in timing/placement. However, from the information I've been able to find I did have the impression that once a person a) worked out the exact system delay from a loop test like I was doing and b) set a "recording offset" to match, it should be possible to get sample accurate placement of recorded audio.

Are you able to comment on that assumption? Is it realistic? Do you see the same kind of variability as I'm getting on your own systems?

Thank you again for your input.

Aanon

[MIDI Life Crisis]

It's those damn electrons! The universe is not perfect. In fact, nothing is. The very "best" performances (recorded or otherwise) are never in "perfect" sync. It's just not the way the universe works. Not that it "shouldn't be" when you want it to be, but if the difference if negligible, I tend to move on and make music.

Of course, if one hears a difference, then one needs to act to make it what one wants. That said, it of reminds me of the old trick of providing five identical mixes to a producer or artist and let them decide which is the best mix. Invariably, they will pick one for whatever reason and be cock sure it's the best mix.

I'm not saying you are wrong, I have no doubt that sync is never perfect in anything if you dig in deep enough. The question is: at what point is it acceptable and time to move on? For me, that point comes a lot quicker than it apparently does for you. Kind of like people with perfect pitch cringing when a piano is a few cents under of over their idea of concert pitch (which is, of course, somewhat variable).

The thing to remember is that data has to be stored in RAM and written to a hard drive. While that is super fast, it does introduce a lag. I'm not sure you will ever find a way to compensate for that short of changing your Rx for Perfection ™ - now with an ingredient originally found in jellyfish (cue cymbal roll).

[MIDI Life Crisis]

ps - It would be interesting to run your experiment on various machines (Mac and PC) with different hardware configurations. Damn! Now you've got me wanting to try this on my system...

[HCMarkus]

Any tracks that go down "onto tape" simultaneously are in time with each other.

Also, my re-recorded click track, once manually aligned with the original at any single point, is perfectly in time with the original click track for the whole duration of the recording.

This is very good news! Phasing between simultaneously recorded tracks would be a real problem.

If I get a chance, I will run some timing experiments on my Mac.

PS: I have been reminded by this thread that I have not checked timing since I upgraded the CPUs and got an 828es... must not be far off (I've done a ton of recording projects since I got the gear running and all has been groovelicious), but certainly a good idea to check!

[aanon]

Thanks to you both for your comments and (reasonable level of) tolerance for my madness!

[HCMarkus]

Madness runs deep within music and musicians.

Did little testing on my Mac Pro (see my sig below for specs)... everything running within 20 samples without fine tuning. Fine tuned to perfection for all audio by setting record offset of 20 samples and playback offset of 0. This is at a DP buffer of 128. Internal timing is maintained at buffer 1024.

Not sure if I am recreating the OP's test, but I can record (thru my 828es interface in digital domain via AVB Mixer) with consistent placement of audio within less than one sample. Also tested using a VI (BFD snare side stick sample) routed thru 828es and we are within 1 sample at buffer 128 and 1024.

Using an external instrument (Yamaha S90ES - acoustic piano sample) at Buffer 128, I get about a 7-8 millisecond delay from MIDI note transmission from DP; this is likely due to combined synth and MIDI response times, very respectable for MIDI. Interestingly (and this is something I noted and posted about here at MOTUNation several years back), the same test with Buffer 1024 results in about 28 ms delay. So the timing relationship between Audio and MIDI output from DP varies depending on DP Buffer size.

Takeaways:
1. At least on a Mac, DP audio recording timing looks rock solid.
2. Use VIs for precise timing and consider recording all external MIDI instrument output as audio before mixing

[cuttime]

Using an external instrument (Yamaha S90ES - acoustic piano sample) at Buffer 128, I get about a 7-8 millisecond delay from MIDI note transmission from DP;

Interesting. How do you measure the delay?

Also, to the OP, I ask the obvious: Is there any plugin on the audio track?

[HCMarkus]

Using an external instrument (Yamaha S90ES - acoustic piano sample) at Buffer 128, I get about a 7-8 millisecond delay from MIDI note transmission from DP;

Interesting. How do you measure the delay?

Place some MIDI Note Events on the timeline.

Transmit to external synth (playback DP) and record audio from the synth.

Compare placement of MIDI Event and onset of corresponding audio on your screen. Zoom waaaay in!

Variables include MIDI baud rate (over USB in my case), Instrument response time, and, in my case, how tight piano sample played by synth is trimmed to zero. The really interesting variable is DP's lag time depending on buffer size!

[aanon]

Also, to the OP, I ask the obvious: Is there any plugin on the audio track?

No plugins or VI/VST being used anywhere.

[aanon]

Madness runs deep within music and musicians.

Did little testing on my Mac Pro (see my sig below for specs)... everything running within 20 samples without fine tuning. Fine tuned to perfection for all audio by setting record offset of 20 samples and playback offset of 0. This is at a DP buffer of 128. Internal timing is maintained at buffer 1024.

Not sure if I am recreating the OP's test, but I can record (thru my 828es interface in digital domain via AVB Mixer) with consistent placement of audio within less than one sample.

Thank you.

That all looks great. In my test I was going out to analog then back in to digital. Not sure from your description whether you were doing the same, or looping back the click fully within the digital domain. In any case, it sounds to me like once you determine the required number of samples you are able to perfectly compensate with offset, and that is what I was wondering.

Since my last post, I also played around with installing an alternative light-weight DAW (apparently I can't say the name here ) on another computer (Vista 32 bit, old but well spec'd). Using the same MOTU 828mk3 interface via USB, I was very quickly able to find the right recording offset in samples and thenceforth get sample-accurate re-recording of the click every single time, whether recording one track or several simultaneously.

So it looks like I need to do some further testing with my Windows7 system to see where the "problem" really lies.

Forums and constructive respondents are great things. Thanks once again.

[MIDI Life Crisis]

Zoom waaaay in!

Therein lies the rub. You can zoom in far better visually than your ears can. Which raises the question: to what extent does it actually affect sync between parts?