I noticed when using Unmix Soundtrack, the resulting layers had lower volumes than the original. This, however, wasn’t consistent. It didn’t happen when I repeated the same selection. The second time it was normal, with no change in volume. Then I tried it a third time to see if maybe it was a fluke, but this time the result was mixed, with choppy volume, from low to the original volume. All very strange.
Even better: It does not null.
-
Open song and duplicate
-
Unmix Soundtrack
-
Merge the unmixed back together
-
Compare the result with inverted phase of first (original) track.
-
The result should null, right?
But it does not! (Why am I not surprised anymore?)
Is there ANY quality testing at all before delivering the software?
Guess not…
Pandoras box of bugs… all over the place
Wow, that’s not right. I know they had problems with this module before in version 12. But it was sorted out. Or so I thought.
Thanks for reporting, this will be fixed in patch 1.
I’m still encountering strange volume fluctuations in this module when using the latest update 13.0.10. Haven’t experienced it with the last version of SLP 12.
Claude and I have been working on similar issues the last two days. Here’s our summary after multiple null tests and Level Check comparisons in SL 13.0.10 Build 449 (x86_64):
Unmix Drums also has a level discrepancy — quantified data (related to the Unmix Soundtrack bug above)
Following on from the Unmix Soundtrack volume bug reported above — I’ve found the same category of issue in Unmix Drums, SpectraLayers Pro 13. Unlike the reports above, mine is consistent and repeatable, not intermittent, which may point to a related but distinct mechanism.
Methodology:
- Source: 12/8 drum loop, unwarped bounce from Ableton (verified null-testing against the original recording before import — not a Consolidate-derived file, to rule out DAW-side artifacts).
- Measured with SpectraLayers’ Level Check module, full selection each time, AES17 RMS enabled throughout.
- Compared pre-Unmix (source layer) against post-Unmix (all 6 drum layers, “All Layers” checked) — same file, same session, no export/reimport round-trip involved in this comparison.
- Confirmed selection state (select-all vs. no-selection) has zero effect on readings (verified both before and after Unmix) — ruling that out as a variable.
- Confirmed FFT Size is not a factor — Unmix Drums is categorized as a Music Unmixer per the “Importance of FFT Size” docs, which explicitly don’t depend on FFT Size.
Results:
| Metric | Pre-Unmix (L) | Pre-Unmix (R) | Post-Unmix (L) | Post-Unmix (R) | Δ (L) | Δ (R) |
|---|---|---|---|---|---|---|
| True peak | -1.50 dB | -3.19/-3.20 dB | -1.72 dB | -3.36/-3.37 dB | ~0.22 dB | ~0.17 dB |
| Sample peak | -1.50 dB | -3.27/-3.28 dB | -1.72 dB | -3.43/-3.44 dB | ~0.22 dB | ~0.16 dB |
| Max RMS | -13.16/-13.17 dB | -13.70 dB | -13.43 dB | -13.92 dB | ~0.26 dB | ~0.22 dB |
| Total RMS | -24.10/-24.19 dB | -24.49/-24.59 dB | -24.28/-24.37 dB | -24.80/-24.91 dB | ~0.18 dB | ~0.32 dB |
| Integrated loudness | -24.74/-24.77 LUFS | — | -24.97/-25.00 LUFS | — | ~0.22 LU | — |
| Loudness range (LRA) | 2.16–2.18 LU | — | 2.08–2.10 LU | — | ~0.08 LU | — |
Every metric shifted in the same direction, by a small but clearly non-zero and repeatable amount — not noise, not a display/selection artifact, not FFT Size, and not a DAW-side gain-staging issue (ruled out separately via a full null-tested source file).
This directly contradicts the documented behavior for Non-Destructive Unmix modes (“the sum of all layers will be equal to the original”), and given the Unmix Soundtrack bug reported above appears related but unresolved even after the “patch 1” fix, this may be a broader issue across the v13 AI unmixing pipeline rather than isolated to one module.
Happy to provide the project file or further detail if useful for reproducing this.
I’ve just updated to SL Pro13.
Just unmixed first song on it. Tried the test of 180 phase on orignal track vs the unmixed tracks. They don’t NUL. Some kick and hi hat is not passing the test.
Is this related to the levels bug reported above?
No effects in use. All faders etc at 0db.
Also the original track wouldn’t play AT ALL after the unmix. I tried track mute, event mute, fader set etc etc.
Nothing would make it play. So without further ado I saved the project and restarted Cubase 15 Pro. It played perfectly.
So there is a bug to do with what SL does to mute the original track. This was post dragging the unmixed SL tracks back to CB.
Never had this problem with previous versions. Is it an acknowledged bug?
Are you using ARA or standalone?
I’ve had trouble with the way levels behave inside ARA, meaning levels would be different when there are extensions active. Not exclusive to SL though, it happens with AAP2 as well.
This was ARA.
Ok! Perhaps to evaluate this issue specifically it’s better to do it on standalone, just to isolate it from ARA issues.
To clarify, my experience with this issue was with the Unmix Soundtrack Module, using the standalone app. The volume level would noticeably jump up or down. Unlike the original material. It was very obvious, and it occurs in both 13.0.0 and 13.0.10. I haven’t had it occur in the last version of SLP 12.