I'm processing my second album recording with Audition 2.0. After a hardware subsonic filter (17 Hz) during recording and subsequent click removal, I normalized to 97%. Now I've noticed this low level waveform between tracks. I'm guessing it's record warp since it occurs approximately every two seconds. But please take a look.



My first idea was to apply a high pass filter to the entire album using a cutoff frequency of 35 Hz. That actually did reduce the amplitude. Then I had the idea that maybe the eloquent way to remove this was to first try to determine what the frequency was, create a notch filter removing this frequency and then apply that to the entire album. Is this smart or overly complicated? How would you handle this glitch?

The other thing I don't understand involves a spectral frequency display of this same in-between track segment. The window zooming is the same. Although most of the frequency content is at the very bottom, there are still these faint vertical "brush strokes" that mar the image. I'm not sure that I hear them, but I'd like to learn what caused them. Is there a different processing technique that might remove them or should I just let them be?



Sorry if the pictures are too big. I wanted you to see detail and I did reduce them in size a little.

They do look a bit like the result of a warp, or possibly a pressing "dimple". If they are, then they should be 1.8 secs apart. The question is, can you hear them? If there are audible, I'd guess they sound like a dull "plop". If they're not audible, leave well alone.


My guess is that if they are audible, they will only be so in very quiet sections. Therefore my inclination would be to try and remove them as isolated glitches, one by one. Plops like this are a classic example of the sort of thing that automatic de-glitchers have no chance with - you can only fix them manually.

Since you have Audition, the first thing to try to fix each glitch would be "Fill Single Click" (or whatever Audition now calls that facility). If that doesn't work, try reducing the amplitude by about 6dB, or perhaps pasting over a copy of the preceding fragment of waveform.

If you could upload a short section of WAV file I will have a go to see what I can do with it. (And presumably Andy might do the same).


I don't see a clear correlation between the disturbances in the waveform view and the vertical lines in the spectral view. So my guess is they are not related. The faint verticals you see in the spectral view are probably just remnants of more generalised surface noise.

As with all apparent glitches, if you can't hear it, don't fret about being able to see it - that doesn't matter. I've seen countless examples of alarming-looking waveforms that sound fine. Leave them alone.

Based on the time scale the fundamental looks to be somewhere around 20hz, which is just above your highpass. That's pretty high for a resonant frequency, and combined with your highpass, I'd just suspect that it's the remainder of the rumble that naturally exists in your system. Rumble can exist all the way to 40hz depending on the state of the system and how you measure it.

I agree with Clive - it's probably not worth worrying about, unless you're interested in upgrading your turntable.

I agree that things that don't matter don't matter. If you can't hear it, and it isn't driving your woofer cone into a tizzy, it won't ever be noticed. Those particular low frequency disturbances are quite common, but rarely a problem.

While it fortunately isn't a consideration on very many albums, I've fixed a fair number of low frequency problems with high pass filters, usually a Butterworth filter in the Scientific Filters, but sometimes the FFT filter. The cutoff frequency I need to use has gone as high as 300Hz, perhaps 350Hz. Third order is the most common setting I use for this purpose.

I do not find Preview very useful. I apply the filter to a selection, listen to it, undo, etc. until things are good, then undo from that selection and apply as widely as necessary.

On some music such filtering sounds terrible and isn't a possibility, regardless of the problem. Often enough however, the only audible effect is elimination of the problem. There have been a few cases where I had to choose between making the music sound different than the original, but still be interesting music, and not getting a product I could listen to at all.

Another approach, often enough successful, is NR applied only to the lower frequencies. That is often better than a notch filter. A notch filter for 60Hz is often quite good, but notch filtering applied to higher order harmonics (120, 180, etc.) is sometimes disastrous. Selective NR may work much better.

Those vertical lines, indications of low level impulse noise, can be from intrinsic surface noise (or wear, especially when played dirty). Very light disk surface scuffing can also be a factor. When audible, these can be significantly improved through the Younglove process. Noise Reduction (obviously not limited to lower frequencies) can also be very useful.

The waveform you see does not have my 35 Hz high pass filter applied, just the preamp subsonic filter at 17 Hz. That portion of the waveform was auto declicked on a "light" setting and then normalized to greater amplitude. I think that is why I noticed it after normalization. I didn't notice this same undulation on my first recording, so I'm hoping it was record specific and not system wide. Who knows if my turntable is as good as it was supposed to be when I purchased it many years ago. But at the time, I knew records were on their way out and I wanted to own a high end turntable that I could use "forever". I ended up getting a $2000 SOTA Star Saphire, a $900 SME 309 tonearm and the most expensive Shure moving magnet cartridge they sold. I now cringe thinking about what I spent. Anyway, I'm stuck with this junk now, so hopefully every recording won't have weird rumble.

No one liked the idea of a carefully designed notch filter? I thought I was being so smart.

Edit: I just saw AndyH-ha's comment about notching.

Interesting.

If you have a reasonable noise reduction plug-in that supports noise profiling (such as Waves' X-Noise), you could use a representative hit as a noise profile in an attempt to isolate an attenuate any instances of them. Granted, the noise reduction will be most effective when the signal isn't bearing smeared by the introduction of other material, but it may be more effective than just attempting to notch the frequency out (notching is just very destructive by nature). Just don't be too aggressive when you're setting the level of reduction. X-Noise can be a little finicky about how it handles its update envelope, so you might not have much luck with such a short profile. And, finally, if you don't have such a plug-in, well, head for the EQ, but you'll inevitably end up taking more out than you'd wish.

Personally, I'd leave it alone. If it's not causing any audible issues during the program, it seems inconsequential. If you want to keep the residual noise during silent passages to retain that vinyl "feel", but without the nasty hits, you can isolate a range between hits and build a quasi room tone with it, pasting over the regions containing hits. I think I'd be more inclined to keep the noise, at least in some form.