I bought the Nuendo 13 today just for a well paid voice and cleaning separation job.
I’m not impressed by the results at all. I’ve ended up buying the "Supertone Clear " plugin who did a tremendous job in one pass .
As a side note, I hate the new Nuendo look. Rubish effort in my book. I’m so sorry to be such a douchBag but flat design is BS.
Post sriptum : there is a nasty bug in the voice separator plugin where when you bypass it, Nuendo goes into christmas tree mode with interrupted audio.
That must be one hell of a voice cleanup job if it makes you buy Nuendo just for it.
That said, yeah, not impressed by Voice Separator at all. I hope they improve it some, otherwise it will never see use here. There are much better, inexpensive options around.
We’ve debated that in a separate thread. Definitely some improvements needed. Also look out for damaged first few frames if there are no handles on the clip. It needs a few frames to warm up.
If money is less of an issue, Cedar also just came out with one that’s pretty good based on my testing. Similar to clear, with maybe a slight edge up.
Cedar VoicEX? Doesn’t it cost like over 1000 dollars and was shown to be worse than Goyo / Clarity, etc?
Yes on that price.
I haven’t seen many comparison. In my own testing (2 week trial) it did faired well to slightly better than clear on those tracks. Not $1K better.
As I mentioned in the other thread: There’s more to denoising a voice than getting rid of all non-dialog sounds. In a lot of cases what’s needed for dialog is just getting rid of some of the background noises. When I work on lifestyle or docs or sports-follow type programs all I need a lot of the time is to just bring down the offending broadband background noise by 20-30%. Cedar, if I understand correctly, has been the go-to processor on cinema mix stages because it does this very well. And there are others of course.
For me within Nuendo the problem has been that the iZotope RX modules I’ve used have been notoriously unreliable as far as automation goes. So I would absolutely love a plugin that does even just 30-40% reduction with minimal change to the tonal quality of the voice, with little delay and with full stability and predictability. Having that be a Steinberg plugin is great because I don’t have to worry as much about compatibility.
While I agree with your view, I didn’t like the sound of this plugin even with very modest amounts of reduction. Let’s see how they improve it in time.
I think another reason though that that is true is because Cedar was the only good option for a very long time until pretty recently when all these other options popped up using neural networks and AI and other things I don’t understand.
But yeah, spot on with sometimes only needing to remove a portion of the noise. All these new ones, at least to me, seem to have no problems with that while keeping the integrity of the voice. It’s why it blows my mind Cedar is asking over a thousand dollars for theirs. Not to mention it’s one knob (*edit - 2 knobs: voice and noise)…you can’t even split bands like in Clarity VX Pro and some of the others.
I haven’t really messed around with the new stock Nuendo one so can’t comment on it! Even if its not amazing yet as some have said, its great theres a stock option and hopefully will keep improving!
Is there no easy way to implement a delta function in these plugins? I find that quite useful while cleaning. Accentize Voicegate has it, but DxRevive doesn’t, not sure why. Clear/Goyo lets you solo or mute. Voice Separator too is missing this option. Most of the Acon plugins allow you to solo the noise.
I don’t have 13 but did you compare it with Spectralayers pro supplied unmix and native dialogue denoiser. Its clarity is decent compared to RX dialogue isolate in 12 though none of them support realtime playback.
Yes, SpectraLayers gives excellent results. And so does Acon Digital Extract:Dialogue plugin.
I briefly spoke to the Cedar guys at AES a few weeks ago. This is a bit of new territory for them. But I also gathered that they have a lot riding on this, both in terms of brand and price point, and expect them to improve that reasonably quickly, which is what he told me as well. He also knew exactly how many samples of latency they had right now, and that this would get better over time.
Players like GoYo/Clear are incentivized to keep racing ahead, since this is their only product at the moment.
SB on the other hand, has to balance a lot of priorities. I’m sure they’ll get to it eventually again, but it may take some time, and it may also be good enough for now. And yes, the SpectraLayers version has a lot more focus in several ML tools and nice advances over the last few releases.
So if you have to put your money somewhere, put it on a player whose business is to make this excellent. Having SB’s tool built-in is a nice, essentially free option, for jobs that are simple and where it works just fine. You just need an alternative when it doesn’t work. But no reason not start with it and see what it does. A lot of this ML stuff is very material dependent. An EQ works the same regardless of what you throw at it, not the case with ML algorithms.
And when you compare, worth doing a blind A/B test, so you’re not biased by knowing which plugin you’re listening to, and which one you might prefer to win the race subconsciously. Nugen has a cool plugin to help with that.
Yes machine learning is material dependent but where separating voice is concerned has the potential to be vastly superior to other techniques. The results from SpectraLayers and Acon Digital Extract:Dialogue are already impressive. I’d imagine Steinberg will gradually improve Voice Separator going forward.
I didn’t do a direct comparison, but have used a lot of Spectralayers and specially Acon’s Extract dialogue, that in my opinion but to no one’s surprise, give much better results.
Goyo is also very good, particularly for reverb reduction.
I did a comparison a few weeks ago between GoYo, VoiceEx, Clarity, Acon, SpectraLayers, and Rx in a local post group. Different people preferred different results.
But in order to test it properly you have to do more what ProTools Expert did what about a year ago where they ran it on various materials in different scenarios and then did blind comparisons across a larger group. The take away back then was that RX was coming up way short of it’s long-term default status. Not surprising since they overall seemed to have stalled under the NI group umbrella.
Also it depends on how much time you spend tweaking the parameters for best results - particularly in those with complex UIs like Clarity and Rx.
Maybe they can repeat it with the latest set of tools. That’s better than what anyone of us here can do quickly on our own.
For me where RX shines is in it other tools. The spectral editor is great and very fast to use, and tools like spectral repair, de-wind, de-crackle, de-rustle and others form an extremely powerful package, at least for dialog editing. For reverb removal, stem separation and NR I prefer others like Acon, Spectralayers and now Goyo.
Spectral Repair is the RX tool I use most (95% of the time). For everything else, I use tools from other companies.
I think this is a very important function. I have occasionally used the signal from this “solo” function for some special effects. 
what @allklier may I ask your personal conclusion about those tests regarding the best possible artifact-free separation?
Let’s say when we talk about dialogue recording with too much backgruond noise and/or hiss?
From my experience GOYO does a great job when it comes to separating.
Unfortunately lots of issues with Direct Offline Processing (also with the official CLEAR release.)
From what I saw in youtube videos ACON extractDialogue is doing a goo job after a little tweaking - in one video it seemed to be as good as CEDAR - which I also only know from youtube.
From my personal experience RX 10 advanced (dia isolate) doesn’t convince at all. RX has a lot of other strenghts when i comes to dialogue editing/cleaning.
yes spectrallayer pro 10 is a safe value
Any reason you’re not using SpectraLayers for this?
I had RX long before I had SpectraLayers, and some of it is just habit and established workflow, and not being fluent in SpectraLayers. But especially with the ongoing DOP mess with RX, the ARA2 integration of SpectraLayers is seamless, it seems almost weird not to use it.
I do like in particular that in any spectral editing in SpectraLayers you don’t actually just nuke the selected area, but you move it to another layer, where you can mute/solo, mix it back to taste. Or do a surgical undo, instead of just undo the whole thing. In many ways superior to the RX spectral.
I’m trying to use SpectraLayers more than RX, but it’s slow transition at the moment.