Why aren't digital hum eliminators a thing?

[Jarick]

I've been thinking about this a while...we have technology to replicate massive multi-amp rigs with tons of studio quality effects in the size of a pedal, but we still rely on fairly primitive noise gates and reverse wound pickup coils. That's technology going back to when we were trying to figure out how to send a rocket to the moon.

Why isn't there a common and effective method of eliminating hum right at the source of the guitar without dummy pickups? I'd think you could build something like that into a wireless transmitter, or have a programmable chip built into a 5-way switch that you could dial in the exact hum canceling by position. Or at the least something better than a standard ducking function built into these really complex DSP units.

Seems like a fairly basic quality of life improvement that should be low hanging fruit right? Or do you need a super computer to cancel out 60 cycle hum in real time?

 

[ian_dissonance]

Gates and noise reduction are definitely an area where guitar-specific effects shouldn’t be trying to emulate anything. And how some flagship modelers don’t have an input trigger on a gate you can place anywhere in the chain is kinda silly.

 

[newholland]

i definitely wouldnt want such athing given the feel penalty of dsp. i mean, it exists in plugins- waves has had one forever- but the processing takes a few ms. i think just a well made pickup pretty much negotiates the amount of noise offset at gain levels i need without the lagginess.

there are newer quiet pickups with loads of gain and less noise- but i guess it depends how and why theres so much noise in your signal as to how nuclear youd need go go in fixing it!

 

[Jarick]

i definitely wouldnt want such athing given the feel penalty of dsp. i mean, it exists in plugins- waves has had one forever- but the processing takes a few ms. i think just a well made pickup pretty much negotiates the amount of noise offset at gain levels i need without the lagginess.

there are newer quiet pickups with loads of gain and less noise- but i guess it depends how and why theres so much noise in your signal as to how nuclear youd need go go in fixing it!

The tradeoffs with mechanical solutions seems pretty high for the tone. You have stacked humbuckers, rail pickups, even traditional humbuckers, all of which sound and feel different than true single coils. The best mechanical solution involves either routing a guitar for an Illitch coil or getting an aftermarket pickguard with one built in. But there's guitars you don't want to modify like vintage Fenders.

Meanwhile AI now lets you split out a full band song into individual tracks that you can remix, and noise canceling headphones have been around for a decade or longer...it's crazy that you can't just zero in on the 60 cycle hum without completely neutering the sound and feel.

 

[Desertdweller]

Gates and noise reduction are definitely an area where guitar-specific effects shouldn’t be trying to emulate anything. And how some flagship modelers don’t have an input trigger on a gate you can place anywhere in the chain is kinda silly.

Yeah the best one I've heard so far is really the Boss one on the MS3 and the GT1000. It uses the input trigger but lets you put the gate anywhere. It's perfect, as gates go. I've complained that the incredible VP4 from Fractal makes you use 25% of your preset if you want to run in 4CM and have a gate in your amp's loop.

 

[newholland]

The tradeoffs with mechanical solutions seems pretty high for the tone. You have stacked humbuckers, rail pickups, even traditional humbuckers, all of which sound and feel different than true single coils. The best mechanical solution involves either routing a guitar for an Illitch coil or getting an aftermarket pickguard with one built in. But there's guitars you don't want to modify like vintage Fenders.

Meanwhile AI now lets you split out a full band song into individual tracks that you can remix, and noise canceling headphones have been around for a decade or longer...it's crazy that you can't just zero in on the 60 cycle hum without completely neutering the sound and feel.

haha.. yeah, i just negotiate it currently with stacked buckers, p90s, or straight single coils

 

[AFKAEjay(retired)]

Of something AI ish could identify the hum, and reproduce it in counterphase…done! #justlikenoisecancellationheadpnones.

 

[laxu]

i definitely wouldnt want such athing given the feel penalty of dsp. i mean, it exists in plugins- waves has had one forever- but the processing takes a few ms. i think just a well made pickup pretty much negotiates the amount of noise offset at gain levels i need without the lagginess.

there are newer quiet pickups with loads of gain and less noise- but i guess it depends how and why theres so much noise in your signal as to how nuclear youd need go go in fixing it!

Gates have typically slower open/close times than it would take for digital processing. There's no reason why it can't be digitally controlled, e.g it samples the sound (by making a copy of the signal), determines "is this guitar or just noise?" and opens/closes the gate accordingly, but the signal remains analog at all times.

I don't think I've seen a guitar product where you have e.g automatic thresholds either, so whether you connect a single coil (more noise) or humbucker guitar (less noise) it would determine the "perfect" threshold for the gate based on pressing some "calibrate" button.

Granted, gates are not exactly a big selling point, but I think they could use some extra love in digital modelers. Key input features are a fairly recent thing too.

 

[jay mitchell]

Gonna throw some cold water on the fantasies being expressed here.

The sound called "hum" contains the AC power frequency and harmonics thereof. Note the second part; you can't just notch-filter 60Hz (50Hz in most countries) to get rid of it. In many cases, second, third, and higher harmonics are present at higher levels than the fundamental. When it comes from a guitar, it is magnetically induced in the pickups and is due to one or more current loops in the AC wiring.

Detecting hum and applying a reverse-phase signal in an attempt to cancel it only works to the extent that the level and harmonic content of the hum remains constant. Try this to demonstrate: connect a single-coil guitar to your rig, select a pickup, turn guitar controls all the way up, and listen to the hum. Turn up your rig volume until the hum is easily heard. Now, rotate your body and the guitar with it. In almost every case, the level of hum will change. Now, move to a different position. What did the hum do? Next, while standing still, select a different pickup. What did the hum do? Is the light coming on yet? Whatever you do to cancel the hum in one position/orientation/pickup selection won't work when one or more of those things changes.

Trying to use a "smart" algorithm to identify and eliminate hum won't work for similar reasons. If you try to detect hum frequencies and then cancel them, you'll remove desirable signal, and you won't ever fully get rid of the hum.

If your environment induces enough hum that gating becomes necessary, the effectiveness of a gate will vary with the style of music and the degree of subtlety you wish to preserve. The threshold that is required to just kill hum when you're not playing may not be enough when you move or change pickups, and too fast a decay time will cut off the tails of notes you allow to ring. The gates in the modelers I use are downward expanders, which makes them easier to use, but I do my best to avoid using them. In one local bar, the wiring around the stage is so bad that I'll only ever play a humbucker guitar there.

 

[AFKAEjay(retired)]

Gonna throw some cold water on the fantasies being expressed here.

The sound called "hum" contains the AC power frequency and harmonics thereof. Note the second part; you can't just notch-filter 60Hz (50Hz in most countries) to get rid of it. In many cases, second, third, and higher harmonics are present at higher levels than the fundamental. When it comes from a guitar, it is magnetically induced in the pickups and is due to one or more current loops in the AC wiring.

Detecting hum and applying a reverse-phase signal in an attempt to cancel it only works to the extent that the level and harmonic content of the hum remains constant. Try this to demonstrate: connect a single-coil guitar to your rig, select a pickup, turn guitar controls all the way up, and listen to the hum. Turn up your rig volume until the hum is easily heard. Now, rotate your body and the guitar with it. In almost every case, the level of hum will change. Now, move to a different position. What did the hum do? Next, while standing still, select a different pickup. What did the hum do? Is the light coming on yet? Whatever you do to cancel the hum in one position/orientation/pickup selection won't work when one or more of those things changes.

Trying to use a "smart" algorithm to identify and eliminate hum won't work for similar reasons. If you try to detect hum frequencies and then cancel them, you'll remove desirable signal, and you won't ever fully get rid of the hum.

If your environment induces enough hum that gating becomes necessary, the effectiveness of a gate will vary with the style of music and the degree of subtlety you wish to preserve. The threshold that is required to just kill hum when you're not playing may not be enough when you move or change pickups, and too fast a decay time will cut off the tails of notes you allow to ring. The gates in the modelers I use are downward expanders, which makes them easier to use, but I do my best to avoid using them. In one local bars, the wiring around the stage is so bad that I'll only ever play a humbucker guitar there.

Check what melodyne can do. For example: tune 1 note in a guitar chord (in a stereo file), remove vocals, identify all pitches in a stereofile…it’s crazy. If that’s possible, I can imagine identifying hum in real time isn’t so for fetched tbh.

 

[MirrorProfiles]

RX can do incredible stuff for removing unwanted noise, but for anything real-time I think the trade offs would be so much worse than just attempting to minimise the problem at the source. To do it well probably needs a lot of CPU and latency (and ideally to “learn” the specific noise). Anything real-time is likely going to be destructive to the main tone

 

[Dom Mcsweeney]

Bertom Audio's Denoiser (free/paywhatyouwant) plugin, but in a pedal would be a dream for non-studio situations.
*edit* or rather - non-plugin situations

 

[GuitarBilly74]

Eliminate nose is what active pickups do. Whether you like their tone or not is a personal thing but that's a solution available.

The dummy coil system that Suhr uses on some guitars is also perfectly fine. Or some stacked pickups.... and of course humbuckers.

There are plenty of solutions out there. I don't see how someone thinking a dummy coil is too radical a mod would think that adding digital circuitry to the guitar would be more acceptable.

 

[hblu22]

UA’s c-axe plugin for real-time noise reduction for guitars does this - obviously not a pedal but it is in real time and no major artifacts. Then the decimator humeliminator decimator pedal does this as well - used it with single coils and works very well.

 

[jay mitchell]

Check what melodyne can do.

I did, several years ago. It has substantial limitations that, even though the company fully discloses them, most folks tend to overlook. It works by identifying and isolating notes, not instruments. Furthermore, it does not include harmonics of notes it isolates. Ergo, harmonic-rich instruments won't be isolated in full, only the fundamentals of individual notes.

For example: tune 1 note in a guitar chord (in a stereo file),

It can't remove an entire guitar part, though.

remove vocals,

That's not terribly remarkable. There's been a way to remove vocals from most stereo recordings for decades. It requires no DSP.

If that’s possible, I can imagine identifying hum in real time isn’t so for fetched tbh.

Emphasis mine. Not arguing against melodyne as a useful tool, but realtime hum removal is not among its capabilities.

 

[newholland]

Gates have typically slower open/close times than it would take for digital processing. There's no reason why it can't be digitally controlled, e.g it samples the sound (by making a copy of the signal), determines "is this guitar or just noise?" and opens/closes the gate accordingly, but the signal remains analog at all times.

I don't think I've seen a guitar product where you have e.g automatic thresholds either, so whether you connect a single coil (more noise) or humbucker guitar (less noise) it would determine the "perfect" threshold for the gate based on pressing some "calibrate" button.

Granted, gates are not exactly a big selling point, but I think they could use some extra love in digital modelers. Key input features are a fairly recent thing too.

im 100% certain theres progress to be made in a bunch of horrible power instances like bars with rheostats and neon lighting and 'uncle bob did the wiring on the stage' scenarios. those are places im not even sure id CARE what the technology was that did it- if it worked, id be onboard im genuinely surprised that given how much better a lot of stuff has gotten that theres not more sorting out of that issue- but i think higher gain cats suffer worse than myself. a few years ago i recorded in brooklyn with a p90 guitar, and i had to stand in exactly ONE position to avoid ungodly noise, and even in the perfect position.. in quiet parts, it had to be filtered post recording cause it was terrible and basically unavoidable in the room. but thats been ONE very critical time in forever... and now its on record.

 

[Jarick]

I don't buy that you can't do it with technology today.

I have a pair of Apple earbuds that can drop the background noise by 80-90 percent, including voices and songs. The latency has to be extremely small to make it work in real time. These earbuds are $250 so the DSP chips can't be super expensive. And the chips are extremely small to fit in an earbud.

I think you could have a setting too where you train it, basically get a noisy guitar and listen for five seconds and then it knows what to look for.

The upside here is you could have something that works with all your guitars rather than spending $300 per guitar to replace pickups that you like the sound of with something that doesn't sound as good but is quieter.

 

[HotRats]

Gates have typically slower open/close times than it would take for digital processing. There's no reason why it can't be digitally controlled, e.g it samples the sound (by making a copy of the signal), determines "is this guitar or just noise?" and opens/closes the gate accordingly, but the signal remains analog at all times.

I don't think I've seen a guitar product where you have e.g automatic thresholds either, so whether you connect a single coil (more noise) or humbucker guitar (less noise) it would determine the "perfect" threshold for the gate based on pressing some "calibrate" button.

Granted, gates are not exactly a big selling point, but I think they could use some extra love in digital modelers. Key input features are a fairly recent thing too.

When tc electronic Sentry came out I thought it was like that, I immediately bought it ans by the time it arrived I discovered that was full digital. I immediately returned it because at the time I didn't want digital conversions in my path.

Never opened the box. Fastest turnaround of my life. Lol.

It was a big let down (my fault for sure) because I imagined it being a digitally controlled multi band analog gate.

 

[Jarick]

Gates don't help much while you're playing and notes are ringing out though.

 

[newholland]

I don't buy that you can't do it with technology today.

I have a pair of Apple earbuds that can drop the background noise by 80-90 percent, including voices and songs. The latency has to be extremely small to make it work in real time. These earbuds are $250 so the DSP chips can't be super expensive. And the chips are extremely small to fit in an earbud.

I think you could have a setting too where you train it, basically get a noisy guitar and listen for five seconds and then it knows what to look for.

The upside here is you could have something that works with all your guitars rather than spending $300 per guitar to replace pickups that you like the sound of with something that doesn't sound as good but is quieter.

the only thing i can think though is that its NOT quite the same thing. a guitar signal is ONE signal. a headphone in your instance is TWO signals, because the ambient noise feeds a microphone separate from the musical signal. its not parsing noise from the musical signal and reinterpolating NO noise, if you see what im saying. its two discrete functions, and the noise doesnt mess up the music if the reaction time isnt perfect. its a lot harder to do in real time to filter and reconstruct a musical signal than just react to noise, and your brain isnt going to react weirdly to a mic suppressing ambient noise in the same way.