Why aren't digital hum eliminators a thing?

[dronerstone]

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

That's why I prefer placing them earlier in the chain, if at all, and rather gate in minimalist fashion.

Younger me always tried putting gate after drive and thought "well, noise gates are shit!"

 

[jay mitchell]

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

Think about this for a minute. FYI, noise reduction has been a goal of DSP programming since at least the 1980s, likely earlier. I know, because I worked with an EE who had programmed DSP adaptive filters for machine-induced noise reduction during that time. He was at the time working on another use for adaptive filtering; I told him and our employer that it wouldn't end up working. I also told them why it wouldn't work and described the ultimate behavior of the system he was developing. It behaved exactly as I had predicted, and the project was eventually abandoned. Would have been great for everyone if it had worked, but there was no way it could.

Hum reduction/elimination has been a thing - a very high priority thing - ever since audio electronics were first powered by AC lines. (That's a lot longer than either of us has been alive, BTW.) Just ask all the engineers who worked on hum cancellation designs back in the day. Actually, you can't ask them, because they've all been dead for a long time. If "technology today" could indeed fulfil your fantasy, it would already be doing so and selling millions of units. It can't, and it ain't.

 

[ian_dissonance]

I don’t know about all the listen/learn stuff, but a more refined version of whatever is going on in the Precision Drive gate (Horizon Drive in HX) with input triggering and free signal chain placement would probably do what I want.

 

[ragingplatypi]

Hum:

 

[ChordInversion]

This would just be code, not "AI." It would be a marketing term and nothing else in this context (as it is in many others).

And yes, BOSS does this with the GT-1000 (and the NS-1X, for that matter).

 

[[Nathan]]

To say it can never be done is an ambitious statement considering the exponential technology leaps the world has had in the last 20-30-40 years. I’m sure it’s an area where more natural solutions can be made.

I saw this a while ago and thought it was pretty nifty for mics. No idea if the concept is applicable to guitars but they seem to have solved an extremely common issue with mics and PA’s

 

[jay mitchell]

To say it can never be done is an ambitious statement

That I haven't made. It is a fact that present technology is incapable of removing hum in real time. For that matter, removal of hum from recorded material is, at best, only partially effective.

considering the exponential technology leaps the world has had in the last 20-30-40 years. I’m sure it’s an area where more natural solutions can be made.

There are many ways to prevent hum. Competent electrical engineers have been well aware of those techniques for decades. Just as with aliasing - discussed elsewhere on this forum - the only cure is prevention.

I saw this a while ago and thought it was pretty nifty for mics.

There's no way I'm going to watch a fluff video that is almost entirely free of content. FYI, my graduate advisor patented an effective feedback detection and reduction device ca. 1979. My biggest customer has patented and implemented a DSP algorithm that almost entirely eliminates the possibility of feedback in systems he designs and commissions. Those systems (see Electro-Acoustic Solutions) incorporate large numbers of loudspeakers - more than 100 in a system we are presently building speakers for - and similar numbers of signal-processing and power amp channels. His life's work is the study of natural architectural acoustics.

No idea if the concept is applicable to guitars but they seem to have solved an extremely common issue with mics and PA’s

The phrase in bold is doing a lot more work than you realize. Anyone can make a youtube video with unsupported claims.

 

[Jarick]

This is a real time VST plugin that basically eliminates background noise. Again, I don't understand why we have technology that can cancel noise in real time for everything EXCEPT 50/60 cycle hum in guitars.

 

[Jarick]

Think about this for a minute. FYI, noise reduction has been a goal of DSP programming since at least the 1980s, likely earlier. I know, because I worked with an EE who had programmed DSP adaptive filters for machine-induced noise reduction during that time. He was at the time working on another use for adaptive filtering; I told him and our employer that it wouldn't end up working. I also told them why it wouldn't work and described the ultimate behavior of the system he was developing. It behaved exactly as I had predicted, and the project was eventually abandoned. Would have been great for everyone if it had worked, but there was no way it could.

Hum reduction/elimination has been a thing - a very high priority thing - ever since audio electronics were first powered by AC lines. (That's a lot longer than either of us has been alive, BTW.) Just ask all the engineers who worked on hum cancellation designs back in the day. Actually, you can't ask them, because they've all been dead for a long time. If "technology today" could indeed fulfil your fantasy, it would already be doing so and selling millions of units. It can't, and it ain't.

Okay but you're talking about work being done 50 years ago. We are doing this in other applications today, except in guitar.

EHX released a pedal that I think is trying to do this digitally about 20 years ago. We've come a long way in digital technology in 20 years.

 

[jay mitchell]

Okay but you're talking about work being done 50 years ago.

No. I'm talking about ongoing work that began more than 50 years ago. It's at least as high a priority now as it ever was. If there were a viable DSP-based solution, there would already be products that implement that solution, and they would be selling in the millions.

We are doing this in other applications today, except in guitar.

No. Audio signals are audio signals. If a realtime solution existed, it would be well-known and would work as well for guitar as for any other signal source. There is no "guitar universe" that operates on different principles than every other electronic device.

EHX released a pedal that I think is trying to do this digitally about 20 years ago.

How'd that work out for them? Where's yours?

 

[TJontheRoad]

Any kind of digital hum removal devise will always suffer from processing latency and thereby cause phasing issues that will effect your tone.

 

[jay mitchell]

This is a real time VST plugin that basically eliminates background noise.

Not exactly. You're missing two caveats: the noise in the demo is constant in level, and its content lies in the low-frequency end of the spectrum. That type of noise lends itself to effective reduction. Broadband noise, not so much. Note that the character of his voice is altered in a very obvious way by the noise reduction algorithm. Maybe you could acquire the plugin and try it out on AC line-induced noise. Who knows? You may have found your solution.

Again, I don't understand why we have technology that can cancel noise in real time for everything EXCEPT 50/60 cycle hum in guitars.

First, what you don't understand is that technology cannot "cancel noise in real time for everything EXCEPT 50/60 cycle hum in guitars." It can reduce noise of specific limited types at the expense of altering the tone of the remaining signal. That is worthwhile for speech-only applications, but most folks wouldn't accept the tonal alteration for music.

 

[Sascha Franck]

There's no way I'm going to watch a fluff video that is almost entirely free of content.

So you know a video that you don't watch is free of content? Cassandra Mitchell?

 

[jay mitchell]

So you know a video that you don't watch is free of content? Cassandra Mitchell?

The brief demo is unconvincing. What's needed is a quantified measurement of gain before feedback. The effectiveness of any feedback suppression scheme - including optimized aiming/placement of loudspeakers and equalization - is directly measurable by the increase it makes in GBF. This fact is well-known and -understood by sound system design professionals, generally much less so by musicians.

Lots of algorithms can, under very controlled conditions, suppress feedback and reverberant tails. "AI" is not a cure-all, although there are lots of folks who want you to think it is.

 

[Sascha Franck]

The brief demo is unconvincing.

Did you try their plugin?

 

[Sascha Franck]

Fwiw, I just tried that Defeedback plugin. It's using an incredible amount of CPU juice, but it's working way, way better than what I've imagined. So there.

 

[Sascha Franck]

Woooah, I forgot my audio track was set to monitoring and kept the mic used for my test next to my monitors when I removed the plugin. Instant feedback mayhem, almost demolishing my ears! Goes to show how well that thing works. Really little coloration of the source sound, too.
Quite amazing, really.

 

[jay mitchell]

Fwiw, I just tried that Defeedback plugin. It's using an incredible amount of CPU juice, but it's working way, way better than what I've imagined. So there.

Here's a definitive test you can perform that will quantify the benefit:

1. Set up a microphone in a reasonable position (i.e., not directly in front of a speaker).
2. Gradually increase the system gain until you just begin to hear ringing. That's your baseline maximum acoustic gain.
3. Deploy the plugin and let it kill the ringing.
4. Gradually increase gain beyond your baseline until you just begin to hear ringing again. Don't push the system into screaming feedback, or the plugin will just reduce gain further.

The amount by which you were able to increase system gain without going into feedback is the amount of gain before feedback the app has bought you. To confirm, you should be able to easily hear the increased volume the plugin has made possible.

For reference, existing devices can increase gain before feedback by 4-6dB under typical circumstances. There's a hard limit on how much you can increase acoustic gain before feedback imposed by physical principles that can't be violated, AI or not.

 

[Sascha Franck]

Here's a definitive test you can perform that will quantify the benefit:

My test looked pretty similar. The amount of additional gain I can get with the plugin is pretty incredible.

 

[jay mitchell]

My test looked pretty similar. The amount of additional gain I can get with the plugin is pretty incredible.

How much is "pretty incredible?" I thought you were German.