Hi all!
I’ve been using the Unmix Noisy Speech module on some impossible material, with lots of reverberating murmur on top of a low level distant speech, and it is almost there, but not quite.
What I’ve found is that the speech is quite clear in a certain region of the spectrum (where the murmurs don’t reach), but the Unmixing keep getting confused with the other voices, because they get quite tangled up in the low mids.
So I was thinking that maybe we could benefit greatly from being able to set limits to the “detection circuit” of the module, so I could tell it like “look for the voice in this region, and use this as a guide to isolate it in the broad spectrum”. Like a sidechain circuit of a compressor.
Could this be done @Robin_Lobel ?
P.s.: before anyone asks, I’ve tried all unmix modules, and the other related ones don’t even get close. Voice Denoise also gets me nowhere.
Thanks!