Topic: Excuse me,another basic SBD question

[mattmiller]

I also don't find identifying an identical point in both sources down to the sample level all that easy.  Audacity may also not be the greatest software for this as the zoom on it I find a bit uneven compared to, say, Audition.

It's not always easy, but the tough ones are the ones that take a few tries to get as close as ~50 samples difference (in my experience), which I consider to be plenty close enough.  And the hour it might take me to accomplish that is nothing compared to the time I spend applying volume envelopes, etc., to adjust the mix throughout the recording.

[SmokinJoe]

^^^ I have to agree, as someone who runs SBD/AUD pretty frequently, and who has also tried to sync in yet more sources (in this case, from another R-44), I've had limited success doing that sync in post, even after reading the thread.

What thread are you referencing?  Am I over-simplifying things?  Here's my process for syncing the two sources:

1.  Open MIC source in Cool Edit Pro and make sure the timeline is displaying in samples
2.  Identify a distinct point near the beginning and zoom in enough to identify it down to the sample.  That's point A.
3.  Identify a distinct point near the end and identify the sample number.  That's point B.
4.  Subtract A from B.  That's the length of the selection in samples.
5.  Open SBD source in Cool Edit Pro 2
6.  Identify the same points A and B, subtract their sample numbers, yielding the length of the same selection on the SBD source
7.  Divide the two lengths, which will yield a ratio of either just over or under 1 depending on which number you divided by which (it doesn't matter)
8.  Multiply that ratio by the sample rate to yield the sample rate that you need to adjust to.
9.  Use r8brain to resample the shorter source by the ratio greater than 1, OR the longer source by the ratio less than 1
10. Line up the resampled source with the other source

An example (I keep notes of all of my work):

Last year at a show I acquired a KM140 > HD-P2 MIC source and a SBD > M10 SBD source.  In CEP2, I determined that the length from points A to B of the MIC source was 295,825,752 samples.  The same A-B on the SBD source was 295,818,010 samples.  My HD-P2 source is always longer, and I always divide the bigger number by the smaller number.  Doing so results in a ratio of 1.000026171496.  Multiplying this by the sample rate of 48,000 gives 48001.2562318.  That's the rate that the shorter (SBD) source needs adjusted to.  After letting r8brain chew on that at the highest quality setting (takes just a few minutes), I opened the resampled file in CEP2 and re-measured A-B.  The result:  295,825,754 samples.  Exactly two samples difference from my HD-P2 source.  And ready for lining up using any single point that's easily identifiable.

This whole process takes as little as 15 minutes and has never taken more than an hour.  The worst I've ever done (on a first attempt) is to end up with the resampled source still differing by ~200 samples, which, at 48000 samples/second, is about 4 ms.  Many wouldn't even notice this (myself included), but I give it another try and invariably end up with something matched to less than 50 samples (1 ms) without ever spending any more than an hour on it.

That's pretty much how I do it too.  For me, this kind of math makes perfect sense, but I have an engineering degree.  Some people really struggle with this kind of math.  Just like when someone who can play guitar fluently shows me a few chords and says "that's all there is to it, go play".  For me, it's not that easy to play.  We all have our different abilities.

But that's how it's done.  If you can handle the math, great.  If you can't, no shame, but you are probably better off to buy a 4 track recorder.

[acidjack]

Also, I just realized I did NOT use your method - I used the other one that involves stretching the time.  I don't have r8brain, but maybe I should get it.  Your method seems easier.

I also don't find identifying an identical point in both sources down to the sample level all that easy.  Audacity may also not be the greatest software for this as the zoom on it I find a bit uneven compared to, say, Audition.

It's not always easy, but the tough ones are the ones that take a few tries to get as close as ~50 samples difference (in my experience), which I consider to be plenty close enough.  And the hour it might take me to accomplish that is nothing compared to the time I spend applying volume envelopes, etc., to adjust the mix throughout the recording.

[mattmiller]

I don't have r8brain, but maybe I should get it.  Your method seems easier.

http://www.voxengo.com/product/r8brain/

The maximum file size for the free version is 2 GB, so you have to chop a larger file in half before resampling if you don't want to pay for the full version.  Just plug your calculated new sample rate (I always go to 8 decimal places) in the drop-down-looking box.  Your base/reference rate goes below it (48,000, or whatever).  I always choose very high quality because it only takes a few minutes, and I've been using 32-bit float for the output with seemingly fine results.  I've never really tried to understand what the whole 32-bit floating workspace is, but I know CEP2 utilizes it so I've been using this option in r8brain.  Would love an explanation of this from somebody.

[TNJazz]

    Based my knowledge ，the signal from SBD's line out is digital source which has been processed by the preamp and ADC built in SBD or  some connected to the SBD on the stage, which we can come to a conclusion that m10 is a bit bucket.
    Is there anything wrong with description above?

Yes.  line out signal is NOT digital.  Digital (SPDIF/AES, etc) is not often found as an output option unless it's a digital console (which is increasingly common, but not THAT common yet).  It's usually analog, so the M10 would be doing the ADC and would not technically be a bit bucket.