Workflow question with Undo History

I have been playing with the harmonic tool in SL13, and inverting to retain harmonics and then erase some of the “cruft” that appears in between them within a testing source to see if I can get some perceived clarity out of that source.

This is where the question starts, how do you effectively A/B the original state of the file without loosing everything through the undo history? Is the obvious answer that you just make a duplicate layer and use that as the reference point? That seems a bit presumptive as I often would like to A/B some before and after work on layers as well (rabbit hole work), but going back through the undo history to find the source point to audition, ostensibly removes all of my work because if I make a selection, that now becomes the current undo history…

So…how do you audition your work, or compare the results of a little labor without loosing it all to a time selection trying to listen back with some context? Is there a “suspend History” mode/feature needed so that you can go back and forth through the processes made without selections getting logged as recallable events?

Thanks

I always make safeties and mute them- always

and I work non-destructively
I use a layer for Good Nz
and another layer for unwanteds (typically muted)

Yes, if I want to work beyond a certain accepted point, I duplicate finished work and mute one of the groups and continue editing.

I also always make a safety duplicate save of redundant slp jobs. I have lost weeks of work in the past and it’s just not worth it is so corruption appears out of nowhere (which has happened to me)

Thanks for the workflow insight Todd!

Making safeties is something I am getting used to in SL, but I don’t intuitively gravitate, or remember to do so all the time.

I guess what I am looking for is how others are making comparisons for evaluation. Easy use case is running some processing and then wanting to audition it against either the unaltered original, or at some point in the restoration/alteration process.

I think my main issue with how I currently work, is that that selections being entered into the generalized undo history is locking me out of my processing. So, say I blow a selection out into some layers, do some spot processing, and then want to evaluate “is that too far, or is that feeling better?” I would have had to have made a copy of the selection before processing to properly A/B, because If I jump back through the undo history to hit play on where I just came from, the next selection I make will become the current “undo” process and wipe everything I just did…

Maybe it is a bit niche, but this is definitely a big part of helping me evaluate what direction is working with varied source materials. In RX, selections are not entered as process/undo history items, so I can always select the initial source (or other juncture) in the history hierarchy to audition, and spot select between the two to audition without loosing work (in SL you can do this, but making a selection removes the history). There is also a comparison window (which I have been trying to get the izotope team to expand on…), but basically it allows to you run whatever modules you like on a selection source, then gradually audition them in their own isolated compare log (like an independent history window) before choosing which to actually render to your audio, rather then hitting process, undo, change parameter, process again etc etc. This is immensely helpful for understanding module settings, where you can see and compare something like say pixel settings beside each other, on your current material, and how they will process before needing to commit them.

I am kind of adverse to gaining extra layers where not needed, so the safety workflow, which is valuable, is not intuitive to me (I’m working on it), and the layer management cognitive load is also distractive from other workflows of trying to isolate, or achieve some results with the content. I guess I am just used to being able to go down a bit of a rabbit hole, then coming out and being able to evaluate “did that work the way I want?” without loosing work, and without screwing my self over cause I didn’t save copies of layers at critical junctures.

John,
Indeed workflow modifications are the name of the game. A .slp job with too many layers can become a cumbersome, sluggish nightmare.

For me, I seldom use the SLP undos beyond about 10. That doesn’t mean I don’t have it set to max, cuz I do.

I should clarify, I work on short scenes; usually two mono mics of location recordings. The scenes are never longer than 10 min…most are about 5min. I will work on about five words at a time; more than that and I can be sure of sluggish performance: make my “cut to” edits and move on. So I work in a rather macro focus of time. I never run VSTs in SLP and I seldom use any module other than Unmix Noisy Speech. All my work is rectangular, brush and magic wand selection tools and clean up with clone-stamp tool. For wind noise I use transfer tool and transfer to my UNz layer (unwanted noise) for non-destructive work (Most of the time set to 45%). And I don’t edit a single thing until I set up my safeties and my redundant job.

I had noticed from the get-go that I didn’t have enough memory to select 5min of 30+ harmonics with SLP harmonics selection tool, my machine would just hang. Frankly, I have stopped using the harmonic selection tool some time ago, I find the magic wand keyed to the speech layer is far superior to the harmonic selection tool. (it’s very easy to make a mistake doing this! I lost 4 hrs ruining my key layer on this past thursday) I realized I had to work on very tiny sections to realize the NR I desired. It is very time consuming, but my goodness does UNS save weeks of work. And my colleagues and I are thrilled to retain the massive amount of wanted noise of our recordings that was just being turned into artefact-infected content using overall NR filters.

That said, I have some long form music that I will put thru SLP in the near future. I’ll see how cumbersome the longer form is for selections when I get to working on it. Maybe by the I’ll obtain a more powerful computer spec’d to run SLP.