Elastic Audio in Depth

[2fly]

Thought I'd chip in again..

With one tune where I had issues with EA on grouped drums I exported all tracks to Ableton Live8 did the tightening in Live and recorded it back into Pro tools (using rewire). The results were much more natural with no (to my ears) noticeable phasing issues but the time it took made it not practicable in an everyday studio session. I mention this because I was under the impression that Ableton helped PT develop it's EA algorithms.... actually I might try this again...will report back if I do

[DigiTechSupt]

Originally Posted by pfo

Totally understandable, thanks for taking the time to chime in.


The thing is, my drum tracks are always grouped. I would never perform an edit on any multi-mic instrument without them grouped. And with regard to #5, if you have an analysis marker at the start of a transient in the overheads, and somewhere in the middle of the transient on the room mics, even if they do both apply to all tracks, wouldn't that cause odd results when you warp? I mean, that essentially means the same transient is represented in multiple locations in each track. And for that matter, close mics vs overheads, the markers will always be in different locations. I always assumed the point of the markers was to say "here's transient x in the snare track, and here's where transient x is represented in the overhead track." It seems odd to apply both markers to each track. But then maybe I'm not fully understanding analysis markers.

I want to reply to this, but I can see it's going to take some serious time and well thought out examples to do it properly.

I'm just going to hit on a couple things I think are missing from what you're doing.

'Anchor' markers - should prevent the 'swishing' that you're getting.

Edits on an EA track - definitely inadvisable and likely the cause of many of your problems. Consolidate before using EA.

I'd also be curious about your workflow - I mean, just how much you're warping, how organic you expect things to sound, etc. Is it unreal expectations? By that I mean it's like having perfect pitch - someone with it can hear things others can't and their expectations may be higher than what is achievable by anyone without perfect pitch. Does anyone you work with (many, some, most?) say the same thing about EA? Have you found other EA type products that do a better job? I only ask because sometimes it's these little details that can make all the difference. I'm trying to establish a baseline of your experience and what you know, what you've tried, etc. Complex applications like this can have so many variables affecting the results, some of which aren't obvious.

For instance I had a case the other day where a customer was having a problem with MIDI. I had made an assumption - EVERYONE knows that! kind of thing - and it ended up that my assumption was the root of the problem. It seemed, to me, to be something so basic that it was beyond the realm of even the most inexperienced to overlook. But there it was.

Throw the fact that you've got edits on the track and I think there may be a couple things that are contributing to a less-than-satisfactory experience on your part. I'll get some good examples together and get back on this thread as soon as I can. I'm on staycation this coming week (taking time off work to mix a record), but will try to get something together over the next 4 or 5 days.

If you have any way of posting concrete examples of sessions exhibiting what you're experiencing, that would help immensely. That way I can see exactly what you're doing and hearing and determine what may be going wrong.

[pfo]

Originally Posted by DigiTechSupt

I want to reply to this, but I can see it's going to take some serious time and well thought out examples to do it properly.

I'm just going to hit on a couple things I think are missing from what you're doing.

'Anchor' markers - should prevent the 'swishing' that you're getting.

Edits on an EA track - definitely inadvisable and likely the cause of many of your problems. Consolidate before using EA.

I'd also be curious about your workflow - I mean, just how much you're warping, how organic you expect things to sound, etc. Is it unreal expectations? By that I mean it's like having perfect pitch - someone with it can hear things others can't and their expectations may be higher than what is achievable by anyone without perfect pitch. Does anyone you work with (many, some, most?) say the same thing about EA? Have you found other EA type products that do a better job? I only ask because sometimes it's these little details that can make all the difference. I'm trying to establish a baseline of your experience and what you know, what you've tried, etc. Complex applications like this can have so many variables affecting the results, some of which aren't obvious.

For instance I had a case the other day where a customer was having a problem with MIDI. I had made an assumption - EVERYONE knows that! kind of thing - and it ended up that my assumption was the root of the problem. It seemed, to me, to be something so basic that it was beyond the realm of even the most inexperienced to overlook. But there it was.

Throw the fact that you've got edits on the track and I think there may be a couple things that are contributing to a less-than-satisfactory experience on your part. I'll get some good examples together and get back on this thread as soon as I can. I'm on staycation this coming week (taking time off work to mix a record), but will try to get something together over the next 4 or 5 days.

If you have any way of posting concrete examples of sessions exhibiting what you're experiencing, that would help immensely. That way I can see exactly what you're doing and hearing and determine what may be going wrong.

First, thanks for your input, it's great having you guys around to field questions like this.

It's a good point you make about expectations. When I say "natural," I guess I mean unaffected. Meaning I don't want to hear any artifacts or other side effects of the pocketing. I want it to sound like the drummer played it just like it is. And I understand that really can't be done in cases where he's really really out, I'm just talking about small moves, nothing drastic.

I must have said something earlier that made it sound like I'm trying to warp tracks with edits on them. That's not the case; I always consolidate everything before I even think of using EA.

Regarding other similar products, I really haven't used anything else like EA. I've always used Beat Detective for this type of work. I never had the time to learn how to do EA right, and I just didn't want to use it unless I was using it right. But I'm tired of Beat Detective in a lot of ways. It takes a ton of work to avoid over-pocketing things, you have to watch out for transients sneaking through the crossfades, and those thousands of fades bog my system down. But the big thing that draws me to EA is it's ability to shift one beat and have all the other's adjust around it. It means you can edit less, and keep things a lot more natural when you do. As an aside, is it possible to spot warp markers back to their original location, like you can with regions? That would be really helpful. Also, it would be awesome if markers changed color when you warped them, so you'd know at a glance which ones are just anchoring the track, and which ones to delete to go back to the original performance. But I digress...

I've tried all kinds of workflows, ranging from selecting and quantizing, to going through each track and moving each analysis marker to the start of each transient before making any adjustments. I never want to do the latter again; my eyes are bleeding by the time I'm ready to begin pocketing! My current workflow is pretty simple. Once I've done whatever comping, editing, etc the tracks need, I consolidate. With all drum tracks grouped, I roll the track, and when I find a beat that doesn't sit right, I find the nearest beats on either side that are correct, and drop warp markers on them. Then I adjust the beat til it sits right, and move on. After I'm done, I'll zoom in and go through making sure no warp markers are sitting directly on any transients, and I'll also delete extra analysis markers (often found after a big cymbal crash, etc). The thing I still don't know is what affect those analysis markers, or their locations have on the results.

The good news is that I think I may have solved the problem I was having in that screenshot, though the answer doesn't really make a lot of sense to me – tell me if it makes any sense to you.

On this particular track, I had been experimenting with the shift+click method of dropping warp markers, which automatically creates anchor markers before and after the one you're editing. However, and I don't know why this didn't occur to me at the time, that method drops warp markers wherever the analysis markers are, which is almost never in front of a transient, and of course different on each track. My normal method places warp markers in the same spot on each track, so everything is stretched equally, but I guess this was creating a situation where the distance between warp markers was different for each track.

The reason that doesn't make sense to me, is that's the normal operation of EA. It chose the locations for the analysis and warp markers, and that caused a change in the phase relationship. I'm beginning to realize that the only way to work with EA is to treat groups of tracks just like a single track, so that all warp markers are in the same location on each track. But what about the analysis markers? Is it okay if they sit directly on a transient?

Assuming this really was the culprit, then my workflow is good to go. I haven't had any phasing issues other than that one lately. But like I said, when you find one, you start to question everything.

I'd still like to learn more about the EA plugin parameters, and learn how to really use them as opposed to just guessing what they're doing.

I haven't finished the EA process on any of these tracks, so I haven't committed anything yet, and I haven't tried X-form. I haven't gotten any of these tracks quite right enough that I want to take the 8 hours X-form will take on my quad 2.66, especially if it's true what folks are saying, that it may not be the best choice for drums anyway. Am I correct that if I were to use X-form, window size, decay rate, etc would no longer matter, as it's a different algorithm entirely?

Thanks again for your help.

[pfo]

Any more thoughts here? Have you had a chance to delve into any of this? I know it's a lot and you guys are certainly busy, but I'd really love to see a clear explanation of EA, specifically the questions I mentioned in the original post (correct placement of analysis/warp markers, deeper explanation of EA parameters, etc).

Here's another screenshot of a single kick drum hit to illustrate my question. From top to bottom, tracks are kick, snare, hat, toms, oh, room. Does it matter that these analysis markers aren't accurate? Of course they shouldn't all be in the same place, but it seems like they should be in the same place relative to the start of the transient, right?

Complicating things even further, they have different relative locations throughout the song. So where the marker on the hat mic is ahead of the one on the snare mic on this beat, on other beats it's the other way around, and it varies infinitely from hit to hit.

So what should it look like? Where should these analysis markers sit? Does it matter?

Thanks again for your time and insight, I really want to fully understand what's going on under the hood of EA. The manual really doesn't explain this stuff with any degree of depth, but I know somebody there must know how it works and what these controls really do!

[SessionKat]

Hey all - I apologize for my "drop in and leave" approach I take sometimes with my involvement of threads as my job's tasks tend to keep me from any length of focus here on the DUC. That being said, besides "pfo's" last question on analysis/transient positions if you sent me explicit questions that still need clarifications I'd gladly get back to you.

This may be the best route:

[email protected]

Cheers,

-tv