Low end rubbish

[Stig Ø]

I posted this in the Mac forum,but thought maybe it was more appropriate here. Mods, please move or tell me off if this isn’t how I should be going about it. Anyway:

I have just recorded a couple of acoustic ditties - solo steel string arrangements. I’m working with eq and compression to give it some polish. In the multimeter in Logic I see some low end rubbish going all the way down to 16Hz! I can’t figure out how to get rid of it - I’ve high-pass filtered the track, both with the channel eq and with a one band on the main buss. What am I missing? I’ve made sure that the meter is placed after the eq.

Edit to add that it’s low in volume, below -60dBFS. I’m still curious as to how to handle it. Let’s say I were to have it pressed on vinyl (wishful thinking), I would prefer to hand over files that were free from stray low end noise, as an example.

- Stig

[The Elf]

Meters don't always show the truth - especially at LF. Can you *hear* any of this low end?

[RichardT]

Yes, at the low end, meters can be misleading sometimes. What filter cutoff frequency and slope are you using? With that instrumentation you should be able to start quite high up with a gentle filter, which reduces artefacts.

[The Elf]

A technique I use for assessing low end is to place a finger *gently* against the edge of a speaker cone. You can learn a lot about the low end using this technique.

[resistorman]

A technique I use for assessing low end is to place a finger *gently* against the edge of a speaker cone. You can learn a lot about the low end using this technique.

Nice, hadn't thought of that!

[Hugh Robjohns]

The guys are right in that FFT displays attached to EQs can be very misleading at the low end because they often have very poor resolution to keep the processing overhead down. Some FFTs allow you to increase the block length and select alternate windowing options which can radically improve LF detail.

You haven't said if these acoustic guitars are being recorded via mics or bugs/pickups. The latter are notorious for generating subsonic mush. Its unusual to suffer lots of VLF from mics — especially directional mics — unless its mechanical vibrstions through the stand (foot tapping?) Or breezes from air con or open windows. Good shockmounts and wind gags are the antidote there.

But removal will need a steep high pass filter ( or two at staggered frequencies) . Ideally 18dB/oct from as high as you can without losing wanted tone and as the very first step before any compression. You might need more after compression, depending on what you're doing/using.

[The Elf]

A technique I use for assessing low end is to place a finger *gently* against the edge of a speaker cone. You can learn a lot about the low end using this technique.

Nice, hadn't thought of that!

I started doing it way back in the late 70s, but as time has gone on I've had other engineers tell me they do it too! Just shows ya...

I believe the last time we discussed this is was dubbed the 'Elven finger'!

[Stig Ø]

Thanks, everyone - wasn't aware of the properties of FFT at low frequencies. I cannot hear it on my small KRK monitors - it's below -60dBFS, and goes from around 30 down to 16Hz. As to how the guitar was recorded, it was miked with a Røde (can't remember the model name, but it's 20-odd years old) LDC, pointed at the 12th fret from 20-30cm distance. The guitar in question is a Dreadnought, so the neck meets the body at the 14th fret. I'm positive that neither foot-tapping nor wind noise caused the offending sub content. The filters had center frequencies at approx. 100Hz, and the one-band HP on the master buss has a 24bB/octave slope. The lowest frequencies rear their ugly heads when I make a somewhat forceful strum - kind of a rasguedo - at certain points in the tune. I'll try the fingertip-on-surround technique mentioned. The funny thing is, I just purchased Mike Senior's book, and suddenly remembered reading the exact same thing there, yesterday or the day before.

[RichardT]

Thanks, everyone - wasn't aware of the properties of FFT at low frequencies. I cannot hear it on my small KRK monitors - it's below -60dBFS, and goes from around 30 down to 16Hz. As to how the guitar was recorded, it was miked with a Røde (can't remember the model name, but it's 20-odd years old) LDC, pointed at the 12th fret from 20-30cm distance. The guitar in question is a Dreadnought, so the neck meets the body at the 14th fret. I'm positive that neither foot-tapping nor wind noise caused the offending sub content. The filters had center frequencies at approx. 100Hz, and the one-band HP on the master buss has a 24bB/octave slope. The lowest frequencies rear their ugly heads when I make a somewhat forceful strum - kind of a rasguedo - at certain points in the tune. I'll try the fingertip-on-surround technique mentioned. The funny thing is, I just purchased Mike Senior's book, and suddenly remembered reading the exact same thing there, yesterday or the day before.

24dB is pretty steep for an HP filter - but with little or no wanted material <100Hz, it probably doesn’t matter much! At 30Hz that filter is already about 40dB down, almost two octaves, so your noise is about 100dB down. I would say you don’t need your second filter!

[Tim Gillett]

The low E note on a guitar is in the 80 Hz region but that doesnt mean other lower freq. energy cant be produced playing the guitar in certain ways.

Maybe you're actually making that sound when you the "rasguedo" thing.

As so often, an audio file example would probably speed up diagnosis.

[Hugh Robjohns]

The filters had center frequencies at approx. 100Hz, and the one-band HP on the master buss has a 24bB/octave slope.

The master buss is waaaayyyy too late for filtering out subsonics. You've got to get rid of them at the earliest point in the chain. Otherwise all your other processing -- especially compression -- is likely to make the situation far worse than it need be.

So if you have a subsonic filter on the mic, use it. If that's not doing enough and you have a high-pass filter on the preamp, use that too... or instead.

And once the signal is in the digital domain, the very first plugin should be a third-order (18dB/oct) high-pass filter. Some filters are steeper -- analogue synth filters are commonly 24dB/octave and digital filter algorithms can often provide some seriously steep filter slopes -- but 18dB/oct is about the best compromise for effective subsonic removeable with minimal audible side-effects.

If your plugins have a linear-phase option, use that too because steep filters do inherently mess up the low-end phase response and you might not like the audible effect (it can start to sound slightly unnatural). Experiment with the high-pass filter's corner frequency (centre frequency applies only to band-pass filters) to find a setting that minimises unwanted subsonics without affecting tone.

[RichardT]

The low E note on a guitar is in the 80 Hz region but that doesnt mean other lower freq. energy cant be produced playing the guitar in certain ways.

Maybe you're actually making that sound when you the "rasguedo" thing.

As so often, an audio file example would probably speed up diagnosis.

That sounds highly likely to me based on what the OP is saying. Rasgueado does create a lot of noise as well as tonal sounds.

[RichardT]

The filters had center frequencies at approx. 100Hz, and the one-band HP on the master buss has a 24bB/octave slope.

The master buss is waaaayyyy too late for filtering out subsonics. You've got to get rid of them at the earliest point in the chain. Otherwise all your other processing -- especially compression -- is likely to make the situation far worse than it need be.

So if you have a subsonic filter on the mic, use it. If that's not doing enough and you have a high-pass filter on the preamp, use that too... or instead.

And once the signal is in the digital domain, the very first plugin should be a third-order (18dB/oct) high-pass filter. Some filters are steeper -- analogue synth filters are commonly 24dB/octave and digital filter algorithms can often provide some seriously steep filter slopes -- but 18dB/oct is about the best compromise for effective subsonic removeable with minimal audible side-effects.

If your plugins have a linear-phase option, use that too because steep filters do inherently mess up the low-end phase response and you might not like the audible effect (it can start to sound slightly unnatural). Experiment with the high-pass filter's corner frequency (centre frequency applies only to band-pass filters) to find a setting that minimises unwanted subsonics without affecting tone.

This is good advice. If the noise really only occurs during the rasgueados, I would consider automating the EQ so that it only kicks in at those points. Because rasgueados have a large noise component, you won’t notice the filters kicking in and out. That way you avoid any negative effects of HP filtering on the rest of the track, even though they are likely to be small anyway.

It’s common for recordings to have unwanted low frequency components - are you sure everything is OK on the rest of the recording? If not, that would change the picture and leaving the EQ in place would be the best option.

[Martin Walker]

A technique I use for assessing low end is to place a finger *gently* against the edge of a speaker cone. You can learn a lot about the low end using this technique.

Damn - The Elf beat me to it with this tip.

I can however add that you can often see really low end unwanted cone wobble visually without using your fingers. I tend to spot it on instrumental (and particularly synth) performances that I receive from elsewhere, that haven't yet been mastered, and occasionally on Bandcamp releases where the musician has smaller monitors, so hasn't heard or spotted the problem before release.

Sound libraries can sometimes be the worst offenders, particularly with kick and other explosive sounds, so watch out!

[James Perrett]

That's the one good thing about the white cones on the NS10 - it makes it easier to see the cone moving which is the best way to judge bass on them.

[The Elf]

That's the one good thing about the white cones on the NS10 - it makes it easier to see the cone moving which is the best way to judge bass on them.

Bass on an NS10?

[Stig Ø]

Thanks again, everyone. Hugh, I see your point about placement of the filter. I do have a HP on a regular channel eq before the compressor, but I’ll try changing the placement and type of the steep filter and use a linear phase eq for that part. I’ll report back.

[Stig Ø]

So I followed Hugh’s advice and swapped out the main buss one-band eq with a linear-phase low cut in the first channel effects slot. In addition, I disabled the two lowest bands in the regular channel eq that I use to shape the rest of the guitar sound. The sub artefacts are now almost completely gone, except a tiny flicker in one of the rasguedos. Next time I’m doing a solo acoustic recording I’ll be sure to enable low cut on the mic itself, or the preamp. Speaking of which, are there important arguments for choosing one over the other, or is it a hear for yourself kind of deal? Thanks to everyone who helped out - much appreciated.

-Stig

[Hugh Robjohns]

So I followed Hugh’s advice and swapped out the main buss one-band eq with a linear-phase low cut in the first channel effects slot....The sub artefacts are now almost completely gone...

Yay! I'm still a useful old fool...

Next time I’ll be sure to enable low cut on the mic itself, or the preamp. Speaking of which, are there important arguments for choosing one over the other...

Low-cut filters on preamps are usually second (12dB/oct) or third (18dB/oct) filters typically with a corner frequency between 30 and 80Hz. These are ideal for helping to reduce unwanted LF and subsonic noise.

Mics that have LF filters can have one or other (or both) of two types of low-cut filter. You need to know which type is provided to use it appropriately.

One type is intended to remove unwanted LF and subsonics, so has a low corner frequency between 30 and 80Hz and a reasonably steep slope of 12 or 18dB/oct.

The other type is intended to correct for proximity effect (bass tip-up from close placement of directional mics) and usually has a very gentle slope of 6dB/oct with a much higher corner frequency, typically somwhere between 100 and 250Hz.

Because the slope is so gentle and the corner frequency so high, this proximity correction filter is of no use for removing subsonics.

So you need to identify the type(s) of filtering provided on your mic, and use the correct one if available.

There's no problem using the low-cut filter in the mic and the low-cut filter in the preamp at the same time if you really need to. Using both together effectively makes the slope even steeper and so gets rid of more subsonic rubbish.... but it will also introduce more LF phase shift which may be noticeable.

[Zukan]

Use the band solo feature in your eq to isolate and hear what is allowed through the lead in on the slope. I tend to go for higher slopes for cutting low end nonsense but you really do need to be wary of how the slope affects the attack transients on low end sounds. On some sub basses if you use a steep slope you can end up with a clicky attack.

Pay attention to the type of eq you are using as well.

[Stig Ø]

Use the band solo feature in your eq to isolate and hear what is allowed through the lead in on the slope. I tend to go for higher slopes for cutting low end nonsense but you really do need to be wary of how the slope affects the attack transients on low end sounds. On some sub basses if you use a steep slope you can end up with a clicky attack.

Pay attention to the type of eq you are using as well.

Thanks - good advice. I’m not sure that Logic’s eq has a band solo function, but will look into it.

[Stig Ø]

So I followed Hugh’s advice and swapped out the main buss one-band eq with a linear-phase low cut in the first channel effects slot....The sub artefacts are now almost completely gone...

Yay! I'm still a useful old fool...

Next time I’ll be sure to enable low cut on the mic itself, or the preamp. Speaking of which, are there important arguments for choosing one over the other...

Low-cut filters on preamps are usually second (12dB/oct) or third (18dB/oct) filters typically with a corner frequency between 30 and 80Hz. These are ideal for helping to reduce unwanted LF and subsonic noise.

Mics that have LF filters can have one or other (or both) of two types of low-cut filter. You need to know which type is provided to use it appropriately.

One type is intended to remove unwanted LF and subsonics, so has a low corner frequency between 30 and 80Hz and a reasonably steep slope of 12 or 18dB/oct.

The other type is intended to correct for proximity effect (bass tip-up from close placement of directional mics) and usually has a very gentle slope of 6dB/oct with a much higher corner frequency, typically somwhere between 100 and 250Hz.

Because the slope is so gentle and the corner frequency so high, this proximity correction filter is of no use for removing subsonics.

So you need to identify the type(s) of filtering provided on your mic, and use the correct one if available.

There's no problem using the low-cut filter in the mic and the low-cut filter in the preamp at the same time if you really need to. Using both together effectively makes the slope even steeper and so gets rid of more subsonic rubbish.... but it will also introduce more LF phase shift which may be noticeable.

Thanks, Hugh. When recording acoustics I use the UA 4-710D preamp. It has a corner frequency of 75Hz, with a slope of 12dB/octave, if memory serves. The frequency on the Røde mic hasn’t been easy to find, but I’ll stick with the preamp’s filter in the future.