Audio Timing from one DAW+CPU+soundcard to another, bullet-proof sync/timing?

[Saxophonick]

Hi Slutz,

being a looong time lurker here, my 2008 resolution include diving more into Gearslutz, so here I am:

Currently working on an album of a bassplayer.
Budget is,of course, tight (are there ANY bands/artists recording their album around that aren't broke?......)

We agreed he would record the drums and basses(pilot-basses for some) himself(with a friend of him having decent room,mics) and that every track had to remain unedited, unaltered, so everything would sync, and not being preprocessed destructively either.

This was recorded in Cubase with a Presonus Firepod on a Dell-Laptop.
The drummer is very steady and has a lot of experience of working with a clicktrack.

I got a harddrive from them with everything recorded at 24bit, 44.1kHz.
I also got every precise tempo of the songs.

I first opened every session in my Cubase Studio4 to check the tempos,
then I imported every song,ie the tracks, in brand-new Logic8 sessions,
at 44.1KhZ as well, but on a MacPro Quad 3GhZ, using a Metric Halo 2882
interface, and gave each session the appropriate tempo.

Sound quality very ok, tried briefly with a click-track, no problems.

Two days ago: we started recording the rest of the instruments starting with guitars.
It went well in the beginning until we got further in the first track and noticed the previously recorded tracks weren't in sync anymore with the click(after 32 bars)!?
What is strange, is that sometimes the timing gets back with the click
after a certain period, and then goes away again.
It's not that it speeds up constantly or down, it's a bit (=1/32 or 1/64 note) after a while up, then down again.
It is also NOT systematic the same speed-up/down on each song,
nor at the same places in the songs.

I have already done sessions with this drummer where I had to add rythmic loop-material (percussions,beats etc...) without problems, so I doubt it's his playing, however the fact that it's not systematically the same in each song whispers in my ear that he could have had a slightly lesser day regarding accuracy in his playing.

He told me he listened carefully with a click-track after, and that it was
real ok, even perhaps tighter than what he'd already done with me.

SO here are the questions:
- Could it be one computer(PC-laptop) to another(Mac) doesn't translate/read audio 100% accurately?
- Is the difference in quality between Presonus and Metric Halo to blame?
- Could the combination of PC+Cubase and Mac+Logic have a different
"way" of recording/reading those "IO"digital bits
- Could it be the Presonus doesn't always record at 44.1Khz, but sometimes at 44.095Khz, or 44.105Khz and/or that the Metric Halo
reads/records also in the same fluctuating way?
- Is any soundcard/combined system of CPU+DAW+card out there 100% steady regarding bitrate or audio timing?
- Is that the major reason that Protools HD keeps his top-of-the-foodchain place, ie absolute accuracy regardless the studio you're in, because you are bound of using the software with one same brand of hardware,
unlike systems where everybody has a different DAW+soundcard?

As this is a pretty fundamental issue that puts the session on hold,
not knowing if we must record everything again, I'd appreciate your slutty input and thank you, in advance, for your time

Best

Saxophonick

[hackenslash]

If it worked well with his click, it's simply a matter of getting the tempo of the click right. It could be down to something as simple as a different clock. However, constancy is constancy, so if it worked with his click, it will work with yours. have you tried the Warp tool?

[Saxophonick]

Thanks for answer,

My click track's fine.
However I noticed listening to a drumtrack with click
can be tricky, because it is not an existing sound of the kit and is high pitched, and as
it adds another sound, it blends more easily even if there
are a few flams....
When I placed a simple kick-on1&3+snare-on2&4 loop that is quantized
accurately, there was a BIG difference with the click.
I think that recording over a straight basic kicksnare loop might do it better than a click...

I've tried many things....for one song I winded up using
Ableton live's warp tool and didn't change the feeling, only adjust
the beginnings of every four bars(shortening some 4bars a few
making some longer).

It worked out pretty well, without audible degradation of the sound
and no messing up of the players' feel, but it took me 2h30 to do it
properly, maintaining accuracy as well as musicality was the hard part.

I'm suggesting heavily the most problematic songs be re-recorded
in my place,

Still did anybody ever experience this, and could there be a techological flaw somewhere?

Thanks for sharing

S

[shaddai]

- Could it be one computer(PC-laptop) to another(Mac) doesn't translate/read audio 100% accurately?

Possibly

I've been in a situation where we had the audio drift off over a couple of hours from one computer to another. One was a G5 running Final Cut, and the other a PC running Sonar. After a couple of unusually tense conversations with my video buddy, I finally noticed out that the guide track I was given to sync the 48khz master audio to was 44.1khz (which it shouldn't have been)...and Sonar happily doesn't care if you import a 44.1 file next to a 48 file. It just goes on it's merry way like nothings wrong. If you compared the 44.1 sync file inside Sonar to the 48 master, they were perfectly the same length, and would play back in perfect sync. Once you loaded the 48k master up next to the 44.1k file in Final Cut, the 48k file was visibly & audibly out of whack...like a bunch...1 second or so..

Sonar was convinced they were the same length, Final Cut wasn't. So, I had him boot out a 48khz sync track & re-aligned the master audio to match. Final Cut said the first 44.1 sync file and the 48 sync file were the same length..Sonar said they weren't. Of course when I gave him the 48khz master made from his 48khz sync, it plopped right in & off to the duplicators it went. I should have done a 44.1 master off his 48 sync track just to see if it would match his original 44.1 sync file in Final Cut.

So, the moral of the story, and hopefully some answers:

a.) Double check the sample rates, playback times, and file sizes of each track on both machines to make sure they match...if they don't match perfectly, find out why.

b.) Don't export/mixdown/bounce the individual tracks out of Cubase for import to Logic...especially to a different sample rate than what they were recorded at. Find the root directory where the .wav files exist, then copy THOSE to the mac... That would eliminate the possibility of a mixdown process tweaking your playback times somehow..

c.) Will Cubase to the multiple sample rate trick like Sonar? If it does, then can it record that way?

d.) Did you import a 48khz click track into a 44.1khz project?

Sorry for writing a book, but that's all I know that could be going on. Hope it helps..
Todd

[Brian Brock]

I guess you can eliminate drummer error by checking in cubase again. fluctuating inconsistency is pretty unlikely to be caused by the computer, right?

[boody]

A friend of mine has the same problem importing a guitar file recorded with protools LE back in Logic: the guitar file was late. They corrected the placement on hearing but the feel was off. I checked the files and by comparing the system noise (there was no audible bleed from a headphone) came to a slightly different offset. Feel was better now.

The problem is that feel and timing are extremely time critical, more then you would think at first. The notes that are spot on are not the trouble; it are the slight offsets in 'feel', the notes that are within a certain marge, that will not sound right when there is a slight offset in overall timing. Best thing to do and be save is to re record (a part of) the guide track in the other system to have a solid time reference. Mic/headphone bleeds can help as well.

I have done a lot of repairs fixing rap tracks recorded on modern pcs that hat to be placed over tracks recorded from an atari: the clock of an atari is slightly slower than a modern pc clock.

[Saxophonick]

Hi Slutz,

thanks for your replies.

Of course I double-triple-checked samplerates.
Also I didn't export the files out of Cubase.
By "click-track" I meant simply enabling the metronome
in Logic, so obviously it was at the right samplerate.
I just opened their Cubase SX-PC version with my Cubase Studio4 Mac
version to check the tempo(to be sure I had written down the correct tempos given by the bassplayer)

Then I created new Logic-sessions and imported(without converting)
the files straight from the root-directory of where the files were recorded
in Cubase, so no conversion would occur.

After more listening and in-depth look, I must conclude it wouldn't have been the drummers best performance, because of the inconsistency
of the speed-up/down ie not gradual.

What about the possibility of an audio interface not recording/writing
at a steady samplerate(like 44.095 KHz or 44.105 KHz) and oscillating
a bit?
For sure such a small oscillation would be very hard to hear because,
the micropitch differences would be too small I think to interfere
with newer overdubs, or be audible to the human ear?

Is that phenomenon possible?
How can any manufacturer be 100% sure the sample rate is locked?
(If I remind correctly this is advertised with RME products)
and that even their better product doesn't oscillate a bit?

We're about to re-schedule a session, but before we do so,
if any of you could assure me there is no such thing as sample-rate oscillation with audio interfaces, that would be great.

Thanks

Best

S

[shaddai]

What about the possibility of an audio interface not recording/writing
at a steady samplerate(like 44.095 KHz or 44.105 KHz) and oscillating
a bit?

I kinda doubt it honestly. Assuming no oscillation, if the PC recorded at 44095 samples in a second, and the mac played 44105 samples in a second, the only thing that would be off is the amount of time the mac took to play back that same section of audio. IE: the PC would record 44095 samples in 1.0000 second, the mac would playback the same 44095 samples in .99977 seconds....and it would be constant throughout..not rising & falling like you described.

However, if one of the clocks was oscilating..ie: going from 44095 to 44105 then back again....that possibly would create the phenomenon you're talking about...but I always thought digital clocking worked off the principle of dividing a much higher frequency that was at a resonance (ie:feeding back)...like 25mhz or something? That would take a pretty hefty swing on the 25mhz source frequency to get the 44.1k dividend frequency to move around enough to hear.

the clock of an atari is slightly slower than a modern pc clock.

Wouldn't that just drift the PC source tracks out of time with the Atari sourced tracks? It wouldn't speed up & slow down...

Dunno...wish you the best with it though. Ever think the drummer just had a few before showing up to record?

Todd

[Saxophonick]

Originally Posted by shaddai

I kinda doubt it honestly. Assuming no oscillation, if the PC recorded at 44095 samples in a second, and the mac played 44105 samples in a second, the only thing that would be off is the amount of time the mac took to play back that same section of audio. IE: the PC would record 44095 samples in 1.0000 second, the mac would playback the same 44095 samples in .99977 seconds....and it would be constant throughout..not rising & falling like you described.

However, if one of the clocks was oscilating..ie: going from 44095 to 44105 then back again....that possibly would create the phenomenon you're talking about...but I always thought digital clocking worked off the principle of dividing a much higher frequency that was at a resonance (ie:feeding back)...like 25mhz or something? That would take a pretty hefty swing on the 25mhz source frequency to get the 44.1k dividend frequency to move around enough to hear.



Wouldn't that just drift the PC source tracks out of time with the Atari sourced tracks? It wouldn't speed up & slow down...

Dunno...wish you the best with it though. Ever think the drummer just had a few before showing up to record?

Todd

Hi Todd,

thanks for insight.
In the meanwhile we eliminated all possible technological flaws....
So no more headaches, and drums will have to be re-done.
(the drummer doesn't drink before playing;was also giving it a thought...)

Best

S

[shaddai]

Best of luck to ya!