Hi, I was wondering if SpectraLayers 10 will also be able to split vocal harmonies of different voices into seperate parts (e.g. for a SATB vocal piece or a song with background singers)
To some extent, it might. Try Process > Voice Denoiser > Background : Other Voice.
Well, I don’t have SpectraLayers yet, but knowing that it could split those SATB/background vocals into separate parts would help with a possible purchasing decision.
Unmix - Song: I just used this on a job where a video editor grabbed a VO that was in the reference video which also had music, in mono! So my only VO at that point was poorly recorded VO with mono music embedded. No clean sources can be found by the people working on the project now so we’re stuck with this crap. But SL10 separated voice from music quite well and then I slapped the mono-to-stereo plugin on the music and now I have something that “works” sonically and am able to satisfy network delivery requirements for separation of VO and music.
Unmix - Multiple Voices: I’ve tried this about three times so far with no luck. In all cases one person begins speaking, another joins and the first person then stops while the second continues. In all cases the separation happens at the point the first person stops speaking, in other words it doesn’t separate when they’re talking at the same time. If you guys have any recommendations on how to make it work better I’m all ears. Getting this to work would be massive for me.
@MattiasNYC if you have trouble with Unmix Multiple Voices (this feature will keep improving over the next patches), you can force voice separation by duplicating the layer, selecting segments where voices overlap, then click Process > Voice Denoiser > Background : Other Voice. And on the duplicated layer do the same, except set “Reduce Signal” instead of “Reduce Background”.
1 Like
Thank you, I will try that.
A big problem at Steinberg: Why do you always choose such horrible voice actors for your promo videos? I couldn’t finish watching this video, the voice-over is unbearable for me. 
5 Likes
This technique works fairly well, not absolutely perfect, but well enough, that further simple editing will help clean things up. Thanks, Robin
1 Like
I had the exact same results for unmix multiple voices. Tried it on 4 or 5 different recordings.
Why not volunteer to do you own version?
It’s probably an American influence type of thing on the music software industry. American’s love sales and will almost never pay “full price” for anything. But put it on sale and they stop looking at the sale price and focus only on “how much they saved”. It’s silly and stupid, but it’s what the culture is I’m afraid.
Bought SL 10 at 23rd and it seems to be a nice piece of software. But it seems to have some major issues.
-
When i drag and drop the Tracks from SL to Cubase, i have to render them, because otherwise the tracks gets muted randomly after some time. Deleting SL from the Extensions doesn´t help at all.
The Audiofile are shown as Wavesforms, but are muted for some reason. -
I usually work on large Voice over projects and had issues with large files like 1h our so. Everytime i run SL on a lager file the entire project gets wobblie and unstable. I already lost a project, where the issue seems affect the entire project.
-
maybe not a Steinberg issue, but i did some Speedtests with SL 10 on my M2 Mac and My older AMD. I took a file for Speecht ot text with SL 10 and the M2 takes around 3 times longer than my AMD. But maybe that has other reasons…
But please!!! Could you make SL more stable and mabye faster in processing time ? There are onlineservices like Auphonic that are faster. Even with upload time included.
Thanks!!!
I have the same problems as you with 1 & 2 in SL10 & Cubase.
But I love that I now can unmix time selected areas!
-
It could be a Cubase issue. I don’t see a reason why SL would suddenly mute a layer.
-
I didn’t see such issue yet, what’s the exact computer configuration where that happens ? Would you be able to record the issue ?
-
The issue is rather probably that Cubase is running in x86 (Rosetta 2) mode instead of native Apple Silicon mode. That’s the default behavior for Cubase 12 on M1/M2 machine, and that’s a massive loss of performance for SpectraLayers. Make sure that Cubase 12 runs in Apple Silicon mode and try again (in the Finder > Applications, check and uncheck Rosetta in Cubase 12’s properties). You can check what mode you’re in by showing the About screen of Cubase (or in SpectraLayers, Help > About).
1 Like
Loving the update. I’m going to use this a lot for sound design, podcast editing, and more. Thanks to those who worked on this.
Thanks for your reply!
-
It´s not only that Tracks gets muted. I also crackles and works not as smooth as without SL. And i only mean, that this happens, when there are lots of SL Action is into place.
I use Cubase 12 since it came out. No major problems at all untill i start using SL. -
Yes…i can record this in Action…will send a videolink soon.
-
Thanks!!! Didn´t know that!
Cubase 12.0.70 has just arrived in the Download manager ,maybe this will address the issue ?
Nothing in the Cubase 12.0.70 release notes mentions ARA2 or SpectraLayers. (The same also applies to the Nuendo 12.0.70 release notes).
Correct , but i’m pretty sure the .70 release would be to coincide with the SL’s release and Ara being tested , Steinberg are not the best at acknowledging under the hood issues but secretly fix them ,so unless you update you don’t know
1 Like
I agree - it is always better to be running the latest version unless there are good reasons why you need to stick with an older version. However, the timing of the Cubendo 12.0.70 releases in relation to SpectraLayers 10 might just be a coincidence. I think the starting point should be to expect the same ARA2 behaviour in 12.0.70 as in 12.0.60 unless @Robin_Lobel or someone else from Steinberg states otherwise, or there is a demonstrable difference in the behaviour.