Voice DeCrosstalk: does it work?

SpectraLayers
1

Hi there!

I’m not yet a SL user but I’m very interested in the Voice DeCrosstalk feature of SL13.

I want to use it for podcasting. In video podcasts I often have the problem of people sitting too close to each other, so I get a lot of bleed/crosstalk wich I want to reduce in post production. I do get good results with RX DeBleed, but it’s a tedious process especially when having more than two mic-tracks. I tested the DeBleed feature in a SL12 Demo a few years ago wich was a lot easier to perform but with quite some artifacts in the results, so I choose to keep using RX.
I wonder if SL13 can do the job better? There seems to be no demo version of SL13 yet, so I’m asking you lovely people, if you were able to test it already?

And is it only a pro feature or does it come with SL Elements too? The Steinberg Website says “both” on the Features-page and pro-only on the comparison chart. Which one is true?

Cheers,
Alex

2

No one used it already? In the meantime I’m also open for any general tips for avoiding cross talk in the first place. Especially with speakers NOT wearing headphones and not caring too much about keeping close distance to the microphone. :melting_face:

3

For me it does’nt work, unfortunately.

Or I mean, it seldom works, so I just don’t use it cause it is usually a waste of time.

4

I don’t know exactly why this difference, but I have a project here where Voice Decrosstalk is working wonderfully.

It is separating very consistently two women talking right on top of each other and the quality of the results is astounding!

5

what is the source? what kind of recording? Mic, etc

6

The source is an interview in open air, in the countryside, fairly good recording but not very clean. A Boom and two lavs. No idea what models.

The two women, the interviewees, are the same age and talk over each other sometimes. Im using it to clean some of the meaningless crosstalks that just pollute the main discourse (meaning removing one of the voices completely from all three channels), and also clean each lav from the other person’s voice so I get better balance and tonal stability in sections where both of them are saying something of interest.

7

But I must have jinxed it in the last post, cause today it hasn’t worked at all! =P

8

But when it does… It’s like magic!

9

In a recent film work, there was a lot of overlapping dialogue and I used this a lot. Actually I used the Unmix 2 Voices process mostly, I got very similar results with that and this one though. I used it to generate fills, and also seperate lavs. I got very good results, meanwhile it was not a one click process for me. I generally got some artifacts that needed manual cleanup, but not much, like 20-30 seconds of spectral healing. The director I worked with is super happy.