Hi all, just flagging if this is a bug or inherent to the algorithm, but SL12 introduces a significant resonance at 8340hz on the noise layer that doesn’t seem to exist in the mixed signal.
Cheers!
1 Like
I hadn’t noticed it. On every file for you? All my work is mono. All my audio has been 4824. I’ve never tried to process an MP3. Is this in standalone?
I can try it see if it happens to me as well….
I just ran UNS on a 9 min mono file which was an on location recording outdoors of an outtake….crew and talent talking. Cheap Senn shotgun recorded to Sound Devices 702…4824 and zero processing beforehand…SLP in standalone version 12.0.40 build 433
I neither see nor hear any excessive resonance at 8340Hz, nor any other audible (to me frequency (my max audible is 12KHz)
what I do notice is that SLP seems to pick out some human voices as noise (me for instance as I was the recordist)….which I hadn’t seen that much before in the past 2 years and would expect that UNS would simply remove speech from noise…sadly, it’s still a starting point with inhales and exhales, chuckles and sighs and various other wanted components sent to noise layer.
Yeah it seems to happen to every file I’ve been processing, at least on this current project. (Dialogue editing, 48kHz, 24-bit, Mono, from Pro Tools Via the plugin)
If I can attach screen grabs, I will… (seems like I can…) I can only attach one so here’s the noise layer, take my word it’s not like this in the non-unmixed file 
So, does the same thing happen in standalone? I only use SLP in standalone and have not seen this resonance as in your screen shot. I’m not paying for PT to test it
Hahahah that’s totally fair… I think I’m in standalone? I don’t use ARA, and I launch standalone first so that Pro Tools flies it into that one instead of opening a second editor (which seems to have all the menu bar items missing when it does?) I’ll do a bunch more testing and report back. It’s entirely possible it is just highlighting some processing further up the chain like Absentia DX or Hush or something…
1 Like
Did some more testing and it seems to be an artefact of the unmix noisy speech algorithm itself. I generated white noise in SL12, then unmixed noisy speech on the white noise, and you can see the effects of the filter in the algorithm at that frequency. It’s usually not much of an issue except in instances where there is a lot of energy in that region (sibilant consonants) and it’s easy enough to clean up, but not having to clean it would be of course easier
1 Like
Glad you logged this! This has been on my to do list cause I have been noticing as well, pretty much on all source material I throw at UNS. I will say that the instances I have run the module and this did not occur are the outstanding exceptions, but run the module enough times and it will do this on all types of material
1 Like
again, please clarify if standalone or ARA
nevertheless, shall we tag @Robin_Lobel
apologies! Non-ARA, standalone. I -did- initially start from a pro tools clip but used the signal generator in SL to replace the content with white noise.
1 Like
J_harry: Yeah.. for my part it’s noticeable mostly because I’m using it to repair dialogue and the resonance is particularly noticeable when there is a lot of energy in the region (‘S’ consonants)
1 Like
I will have to test more with ARA, but most of my isolated tests are standalone. I will try and see if I can get a read on if the frequency is similar to Dustins
1 Like