Topic: Matrixing 6 sources

[Duffalo_Soldier]

Hi all, first post so go easy on me!

I’ve got a bunch of stealth recordings of shows, pretty much all digital. Some go back to days of using mini discs whereas everything since 2013 is made on digital recorders (Tascam DR-05, DR-40x etc) and I have some friends recordings of the same gigs done on various Edirol or Zoom recorders.

I’m trying to matrix some of them and obviously they are nearly always subject to clock drift. A friend who runs a very successful studio basically talked me out of timestretching any of the clips I have as he insists they will be out of time and/or out of phase and that I should basically move one of the files every song/every 3-4 mins. I have tried basically timestretching them and the ‘cut up’ method he suggested with varying degrees of success. Is there any means of importing files without timecode into a program and it will automatically align them for you? I’m using Windows 10, Audition, Audacity and Izotope RX for all my other processes but the one thing I am banging my head repeatedly over is ensuring the tracks are synced properly and without phasing in the first place!

I thought I would start with the most complicated one I have which is 6 audience stereo pairs, some are reasonably easy to line up albeit reversing the polarity on some of the recordings. I can attach a screenshot or two with a later post but right now I am having enough trouble trying to post this!

I did try the trial version of Syncaila, and an old version of Plural Eyes but both of those needed XML files done in something like Premiere Pro and the only issue with that is that I don’t have a recent version of Premiere to be able to make those files in the first place so didn’t feel like shelling out money on Syncaila as I couldn’t test things properly!

Any help or suggestions would be greatly appreciated!

Thanks for your patience!

😉

[hoserama]

I'll disagree with your studio buddy pretty aggressively. If you do the "chop & align" method, it will be aligned at your start, but immediately go out of sync depending on how much variance on the speed. You're already going to align again in 3-4 minutes, so why not just time stretch between those two points?

I've been using Adobe Audition for the time alignment. AA3 is by far the most accurate algorithm I've found in terms of exact percentages.

Additionally, even if they were completely aligned correctly (which is a stretch for aud recordings), consider they're different PA sources and you'll likely get phasing between them. I typically advise against matrixing separate audience recordings unless there's a specific element or issue you're looking to address.

I did a video years ago on time alignment with Adobe Audition:
https://youtu.be/MOz2GUdL_vw?si=ZRm1XdDvCIWKG-S8

Halfway through you can hear my cats barreling through disrupting everything.

[nulldogmas]

I'll disagree with your studio buddy pretty aggressively. If you do the "chop & align" method, it will be aligned at your start, but immediately go out of sync depending on how much variance on the speed. You're already going to align again in 3-4 minutes, so why not just time stretch between those two points?

I'll second this strenuous disagreement. If you align the files at the start, pick a spot near the end of them with an obvious loud spike (snare hits are good), and time-stretch them from there to match, you will be fine.

It's very easy to do this in Audacity, as you can simply select the two spans, take the length of one in samples, select the other and do Effects > Change Pitch and Tempo, and paste the length of the first recording and paste it into the second as the new length. If your selected marker for the end of the span is close enough to the end of the recording, you don't even need to worry about whether the remainder is slightly out of sync, as the drift won't have time to become apparent before the recording runs out.

[RyanJ]

I've never had an issue with syncing at the start and then like everyone else said, stretch to near the end of the recording. Usually takes a few minutes.

I cut that out completely and have been using this. I know it's a bit pricey but if you will do this a lot. It's definitely a good option! https://www.soundradix.com/products/auto-align/

[fanofjam]

I also agree with the others.  I've had lots of success time stretching recordings and getting them to synch just fine. 

The problem you run into, and this is probably the situation that your studio friend encounters and the reason he says what he says, is that it becomes more difficult if, for some reason, one or the other of the two recordings that I'm synching has any time breaks, like say between songs or at set break.  If they were made, it's never going to be obvious on a recording where those breaks are...you'll only know that your tracks are synching until at some point they go out of synch.  Studio's don't record like we do at concerts...where we just hit record and let it run for a complete set or a complete show. 

That's when you need to start breaking the recording up into smaller and smaller pieces...say track-by-track.  And yes, in that case, I agree with your studio friend...it's a real pain in the ass getting those recordings to line up.

I remember one time I tried synching up two recordings and they just wouldn't synch up right.  I was pulling my damn hair out and couldn't figure out what was going on.  I ended up using the zoom tool and finally found out that one of the recordings was glitchy with occasional random data drop-outs.  The drop-outs didn't produce noise...they were literally loss of data on one of the recordings where a few milli-seconds was just gone and, at that point, the two recordings were out of synch by those milli-seconds.  The data gaps were so short they weren't even audible when listening to the affected recording, but there were enough of them to unsynch the two tracks.  I had to give up on that one because it would have taken me weeks to go through that recording manually to find and correct for all of those glitches.

PS:  After writing the above, I thought about it some more and I remembered back to the situation I was referring to.  One of the two recordings I was trying to synch was some sort of transfer from DAT tape to file.  Can't remember details other than when you transfer DAT, you can have issues with variation in DAT tape transport speeds and whatnot.

[jefflester]

Aligning/matrixing six audience pairs, sounds like a nightmare.

[morst]

Sounds like a nightmare, especially if you do NOT squash/stretch.


6 sources drifting out of alignment at all times except the few spots where they are lined up every 3-4 minutes?


hard no.


Here is my blog post from a long time ago about aligning sources using squash/stretch (I prefer squashing all longer sources to match the shortest, so that no samples have to be invented.)
https://morst.blogspot.com/2010/03/this-is-follow-up-of-sorts-to-my-post.html

[opsopcopolis]

Add me to the list of people that disagree with your friend. That is the correct way to do it

[Duffalo_Soldier]

Thanks to everyone who has provided their thoughts and advice. I think it’s only going to be five sources as one is incomplete, has tons of mic bumps due to the taper “dancing” 😂 and ultimately doesn’t really add anything to the mix. I’m very tempted to try out Auto Align 2 and despite the cost, if it makes my life easier in terms of lining up sources and phase correcting them then it sounds like it’s worth it. With this in mind, is it possible to use Auto Align 2 as a standalone and line up the various recordings and then export them for use in Adobe Audition (I have version 3)? Or do I have to have something else (for example, Reaper as I use a PC) and use AA2 as a plugin? If so, can I export the wavs from Reaper once time aligned (if necessary adding some silence to the start if the track is missing portions of the audio comparative to some of the other sources?

Thanks again for all your advice, it’s been very helpful and is greatly appreciated 😊

[morst]

If any of the sources were not made with a mic at a fixed position, the delay value will fluctuate constantly for those versus a fixed source, due to the changing distance.
Lining up starts and ends on a moving source (as with the online tools you propose, or with my mathematical method) does not guarantee that the middle sections will be aligned.
This is in addition to all the other things that make sync difficult; the speed of sound varies with temperature and air pressure, of course.)

[Scooter123]

Do it anyway you want.

You’ll figure it out.

The time clocks will drift. Whether that is five minutes or fifteen no one knows.

I used the synch at two points, stretch between them method, usually about 10-15 minutes. Easiest to sync at the beginning of he track

[aaronji]

Pick the best one and be done with it! In my personal experience, these matrices usually don't sound better than the best stereo recording in the absence of some advanced planning. Every once in a while, maybe, but generally not.

[Gutbucket]

^ I agree with Aaron for the most part.

What is it you hope to achieve by mixing so many sources?  I ask because I see far more hassle than benefit in mixing 6 separate stereo sources very likely recorded using very similar microphone techniques.

As others have stated, there are methods and tools you can use to achieve this.  The better ones will stretch/shrink each additional source to match up with synchronization target points you choose in the baseline master source.

Complications arise with sources that require not only basic initial synchronization (those that share the same clock or are close enough to only need initial alignment) and perhaps a single time-stretch/shrink over the course of the entire recording (because they were recorded using clocks running at slightly different but steady clock rates), but additionally do not maintain a constant rate over the course of the recording (those that speed up and slow down over time).  That last category is a real PITA as it will require many sync points with different stretch/shrink factors between each, and is more likely to occur with analog tape sources.

But if your goal is producing a better sounding standard 2-channel stereo recording than any of the individual sources, I strongly suspect you will quickly reach the point of diminishing returns upon combining two or maybe three sources..  when done correctly on those that will actually benefit from it.  The primary value in mixing together more than one source is typically to either cover problems in one or both sources, or to combine several sources that are different enough from each other such that their sum ends up sounding better than any one of them in isolation. 

Its unlikely that each of your six sources will provide something usefully different enough from the others that the combination of them produces a recording which sounds better than just one or two of them properly treated and combined, particularly if you are doing other things such as EQing. IME, you are likely to get far better results from the careful manipulation of EQ and dynamics on just one or two of those sources, than by mixing all six together with or without the other improvements.

Some will argue against using any EQ or dynamics processing.. and they aren't wrong as its easy to do more harm than good.  But the mixing of separate sources can be considered a similar "bastardization" of the original source material, so if you are set on doing that kind of thing anyway, you might consider EQ and dynamics on the most promising source or two rather than attempting to mix a bunch of unintentionally-related recordings together.

^
I say all this as a guy who regularly runs a dedicated 8 channel microphone arrangement that can be thought of consisting of 4 stereo pairs which I subsequently mix down to 2channel stereo. Obviously I find recording this way worthwhile, even though I know many other tapers feel what I am doing is already "jumping the shark".  However I can justify using each of those four pairs because each is configured very differently from the others, and they are arranged in a way that's specifically designed for them to be mixed together in a way that provides more benefits than potential problems.  For instance, none of the four pairs use the same stereo pair configuration, and one of them faces away from the stage in the opposite direction.  Additionally, I record all four pairs into a single recorder which keeps all four pairs in sync.

But do your own thing and let us know how it goes.  Share your results here if you can.  I'll be more than happy to be proven wrong if the results are good.  Beware of things sounding smeared or blurred, and its probably best to start with a few simple ones that require sync'ing just two sources to get familiar with the process and the benefits before tackling the bigger projects.

I'm now trying to imagine strategies that might be useful once something like 3 or 4 separate unrelated AUD sources have been fully sync'd, and can envision some advanced mixing methods like starting with the most decent stereo source and adding some mono-ish mixdown of an alternate X/Y pair to solidify the center image, along with maybe some stereo-widened content from an omni pair to enhance the ambient/reverberant/audience aspects.  Something like that is making each source different enough that they all contribute something different to the end result, I can see that working, but its a rather complicated approach that may not be worth the trouble.

[Duffalo_Soldier]

Thank you once again! I suppose what I am trying to achieve is the best sounding recording I possibly can. As I mentioned previously, I have a lot of recordings that are predominantly one stereo recording, all of which are digital. There are however, a significant number where I have a second recording and a small handful where there are numerous additional recordings (like in this case)
I was reasonably confident in my timestretching and phase inversion ability until I had a conversation with that studio friend! Granted, it was over messenger and being completely honest, he deleted all of his comments the next day (possibly because he may have felt foolish or perhaps alcohol was involved?!). In short, he basically said at the time, that I would completely **** it up entirely by timestretching it despite it being necessary due to the drift.

   I will absolutely for certain, post some snippets of the individual sources and the finished version of a couple of songs for people to let me know what they think, once it’s done and I totally hear you Morst (and others) with regards to some peoples preferences to NOT using EQ or dynamic processing. That being said, my feeling is to NOT correct musicians who were out of tune or time or inadvertent feedback squalls from the mics/PA. Ultimately I am trying to preserve the live performances as close to what they were heard on the night in question although with corrections to the recorder or recordings in the process. By this I mean, supposing someone was recording half an inch away from the drum kit (not an issue I tend to have but I use it as an example), that might be all you can hear on the whole tape and maybe the other instruments or vocals are completely drowned out. Obviously this is most likely not what most of the audience heard on the night in question so a bit of a rebalance in RX works for me in this example.

I will be doing some studio recording with some friends in the not too distant future so getting my head around Auto Align 2 sounds like a good idea regardless. Hoserama’s timestretching video about Adobe Audition 3 has been my go to manual for aligning sources but would now like to try Auto Align in the fashion that Ryan mentions above. Maybe I should message him directly or perhaps he can outline his process of utilising it so there’s less need to timestretch in the first place and also (the key thing for me) to see how it ensures that your mics are as constantly phase aligned as possible?

Thanks again for all your support and patience!
😉

[hoserama]

Auto Align 2 will not be helpful in this situation, it does not do any time stretching. It would be useful if all your sources are in different locations, but recorded on the same deck.