Any tips for improving extraction of live vocals from a crowd

I’m working on some very challenging audio from an iPhone capture of a gig. The band need me to clean it up and add some underlayers to enhance it. There was no desk audio and the crowd were noisy. The unmix crowd noise has done a decent job but theres a good bit of the needed lead vocals in the crowd layer. I’ve been drawing areas and moving them to the foreground layer but its very time consuming.

I was also trying the frequencty select tool and to less success the harmonics tool. I’m not sure of best ways to get useful results from that.

The layers are fairly modulated/watery sounding but still can work wtih it

Anytips for this kind of job with Spectralayers?

Maybe an Unmix Mid/Side of the crowd layer could help you?

And perhaps another round with Unmix Crowd on each of those two resulting layers?

I would start with Unmix Noisy Speech and see what that yields

I find the tools that are tied to amplitude like Freq/Harm selection, Magic Wand tend to make too narrow of selections and you need to adjust parameters to get something usable

I find the free selection tools give better results for my uses
I also find the harmonic selection tool can wander if you have a lot of harmonics being selected

yep, it is very, very time consuming.
If you have the lead vocal somewhat separated, then use that as a key for selection rather than free-handing (mousing) in the noise layer. I typically use magic wand and brush tool on the separated speech/ voice layer to build up a selection and then switch to the noise/ crowd layer to edit from. When selecting on the noise layer, you can get lost in the noise.

Also, remember ctrl+shift+D (win) for re-selection of previous selection.

Just trying help, good luck

Try the demo of waves Clarity VX. I’ve used it to help get cleaner vocals from audio provided by video company to client. Was able to tune the resulting vocals and put back into the video. In this case there was an incredible amount of spill from music onto the vocal mic.

All good practical tips thank you

I do have Clarity and it is very good yes. Its a pretty challenging source

or tough live recordings in SpectraLayers, try starting with Unmix Noisy Speech instead of relying on freehand selections or frequency/harmonic tools—they’re too narrow and inconsistent. Use any partially separated lead vocal as a key layer: build selections with the Magic Wand and Brush Tool on that layer, then edit the crowd/noise layer from it. Ctrl + Shift + D (Win) helps recall previous selections. You can also experiment with Unmix Mid/Side on the crowd layer, then run Unmix Crowd again on the resulting layers.

If you want faster results, consider external tools like Waves Clarity VX, which can isolate vocals more efficiently from heavy spill.

I would also try Unmix Song with just Vocals and Other selected.