🚀 go-pugleaf

      Turntable rumble is actually 180 degrees out of phase from right channel
      to left.




I don't think that it always is - it rather depends on what causes the rumble. For instance, not all pressings are rumble-free - and anything introduced during the cutting phase is very likely to be in-phase on both channels. If that happens, then you only have NR or EQ to use to get rid of it. But as you say, if it's out of phase, you have another option...

Since you have AA1.5 (I presume) then you have a couple of things that might help you; in MV you can use the frequency band splitter to extract just the part of the frequency range that you want to experiment with. When you've split the track into frequency bands (which turn up as separate tracks in MV, which is why this effect is there, rather than in EV), you can open up the band in EV and apply one of the Channel Mixer effects to it to create a mid/side version of the band. You are looking for the preset that says LR to Mid-Side - this will convert your signal to a sum and difference signal. If in your case there is much more rumble on the difference signal, then simply throw it out and leave the sub-bass as mono. It will make no difference to the resultant sound, because just about all of the sub-bass there is on vinyl has been reduced to mono anyway, simply to stop your stylus flying sideways out of the groove every time it hits any of it.

And I wouldn't worry about phase effects at this sort of frequency either. The bass is either going to work, or not work. Ultimately it's all down to taste - you either like the result, or you do it again until you do.

Steve, am I wrong to use the FFT filter as a HPF? Are we killing the rest of the spectrum phase-wise? I've never been able to see any evidence of it, but in theory it might be a problem. Usually I'll do logarithmic / log scale, 24K FFT size, Blackman, with a steep slope from 35 or so down to 25. Anything lower follows along as set.

Tigerdog, unless you're working with pipe organ or unusual electronic music, you're not likely to be encountering any sound at 25Hz or lower. Noise Reduction is likely to be a hit-and-miss proposition because the amplitude of the rumble can vary so much. Steve's method is very practical, actually, and I once discovered pipe organ bass buried in out-of-phase rumble via the method he described. The only time that method will work against you is if the recording was made Decca-tree style or some other spaced omni arrangement whereby some of your bass ends up cancelling out.

      Steve, am I wrong to use the FFT filter as a HPF? Are we killing the rest
      of the spectrum phase-wise?




It's all a bit moot. I looked out (and modified slightly) a post I made about this six years ago about phase errors and filters on the old Synt forum archive:

      FIR filters can be constructed to be linear phase or not - usually they
      are, but it's not obligatory. IIR filters (ie using feedback techniques)
      can be made to ring, and there are also techniques for reducing the amount
      of non-linearity in them.





      Calling them phase 'errors' is a little misleading. All filters will have
      a phase shift, or delay associated with them - they wouldn't work otherwise.
      The critical part is the linearity of the shift, and that is where FIR
      filters are more likely to give a good 'musical' result.





      AFAIK, the CEP parametric EQ filters are definitely IIR - I can make them
      ring.




This would seem to indicate that using the Scientific Filters to remove rumble would be a perfectly fine idea - but ultimately, if it sounds okay - then it is! And quite frankly, we are only talking about a very low frequency effect, and even many degrees of shift are going to be subjectively less noticeable than they would be at, say, a speaker crossover frequency, and that's why I think it's a moot point.

Thanks for this discussion - it's quite helpful. Sadly, I'm ignorant of even the grosser points of FIR v IIR filters, but I especially appreciate the practical pointers.

As it turns out, I *do* work with a bit of unusual electronic music (a holdover of my youth spent collecting early Tangerine Dream, etc) but I'm just as interested in preserving the gut-feel of acoustic bass drum in jazz and rock recordings.

With a little further experimenting, I've discovered the signature rumble (approx 12Hz, -49dB) from my SOTA Vacuum Star turntable is not purely out of phase; summing this signal only provides about -5dB reduction of the signal. Perhaps if I do this as a first step then apply a 7th or 8th order butterworth filter I'll get the best results with least disruption of the signal?

Seems perfectly reasonable to me. That bass drum would be unusual if it had much oomph below 35 or 40 Hz, but even if it's that rare drum with a reach to 25, it sounds like you'd be OK with that technique.

<tigerdog@adobeforums.com> wrote in message 
news:59b82102.-1@webcrossing.la2eafNXanI...

>I use Audition mostly to create digital archives of vinyl LPs. While I have 
>a very good turntable, preamp and cartridge, there is still enough residual 
>low-frequency information (25Hz and below) that I would like to eliminate 
>it. I do NOT want to lose what LF info is present, only eliminate the 
>residual rumble. I understand there are several ways of going about this in 
>Audition:

> - equalizer / parametric equalizer reducing low frequencies

Best tool for general LF noise  is probably one of the scientific high-pass 
filters.

If there is a discrete LF tone then the DTMF filters are the better choice.

> - noise reduction, getting noise profile from silent passage (lead-out 
> groove)

Low frequency noise from LPs does not seem to always have the necessary 
properties for this to work well.

> The EQ option seems too heavy-handed and I can't seem to find any 
> indication of the phase impact.

Don't worry about phase as long as you do the same thing to both channels.

> FFT filtering can also work but again, I'm concerned about phase.

Actually, the FFT filter has near-zero phase shift.

> I would also like to try an old analogue audiophile trick. Turntable 
> rumble is actually 180 degrees out of phase from right channel to left. I 
> would love to find a way to sum frequencies below 25Hz as (at least) a 
> first step in this process.

Edit Mix/Paste (overlap) with the invert boxes checked, after you've 
filtered and swapped the channels will let you experiment with this.

<tigerdog@adobeforums.com> wrote in message 
news:59b82102.3@webcrossing.la2eafNXanI...

> With a little further experimenting, I've discovered the signature rumble 
> (approx 12Hz, -49dB) from my SOTA Vacuum Star turntable is not purely out 
> of phase; summing this signal only provides about -5dB reduction of the 
> signal.

Not bad.

Perhaps if I do this as a first step then apply a 7th or 8th order 
butterworth filter I'll get the best results with least disruption of the 
signal?

IME, you've found a good direction to go in. Experiment with fiilter types, 
(butterworth, Bessel, Chebychev), order, and corner frequencies.

IME the worst thing about the FFT filter is that it can't be controlled well 
at very low frequencies.