Is there any advantage to normalising a track prior to unmixing song?

SpectraLayers
1

Does running the normalise function on a song before running the unmix song function on it, yield better results, or is there no difference?

1 Like
2

NEVER do that!!! In my experience :slight_smile: For all sorts of reasons.

UNLESS, the song initially ONLY contains two things…only. For example a vocal and upright bass ONLY.

imo

2 Likes
3

Care to elaborate?

4

I use SL to remove bleed, mainly from drum feeds. I take a drum feed and unmix song for drums then disregard other.

So I’m wondering whether this yields a higher fidelity output if the feed is first normalised to 0dB by SL?

Since I batch process up to 16 drum feeds, if normalisation doesn’t add anything, then skipping it will save a bit of time and processing power.

I could normalise in the DAW beforehand, but then I’d have to incorporate an additional render stage to inscribed the normalisation into the media file, so easier to just normalise in SL… but only if doing so is advantageous.

5

In general better dynamics will get you better unmix results.
But normalization won’t make the dynamics any better. It will just raise the volume of everything, including the noise floor, with unchanged dynamics.
So I highly doubt that normalization will have any effect on unmixing.

2 Likes
6

Doing a simple test with a file that peaks around -1dBFS:

  • Original file → Unmix Song (high quality, all instruments) → results in pretty clean stems.
  • Copy of original file → -24dB gain → Unmix Song (high quality, all instruments) → +24dB gain on resulting group → stems are with less details/less accurate separation and more noise/artifacts.
  • Both stem groups sum up to the same result and null each other out, when brought back to the same level.

So yes, there might be a benefit, if you normalize/amplify the audio to a higher level before the unmixing process.

2 Likes
7

Thanks, that was also my experience.