Feedback control

[Manukatche]

Hello friends

I have a great challenge for you




I have a speaker and a microphone placed back to back.

The challenge is to increase speaker volume as high possible before feedback.

No file is played, the sound come from quiet ambiance.

3 rules:
- No DSP. !!! EDIT: actually none attenuation is allowed, the deal is to prevent feedback starting !!!
- the microphone sensibility is locked.
- The anti feedback have to be smaller than the speaker's back.

Would you take it up this great challenge?

 

[kemmler3D]

I imagine you have put together a trick question of some kind here... but... I'll bite anyway.

You can use an analog notch filter to reduce gain at the main feedback frequency in the room. IME this tends to work pretty well and you can add quite a lot of gain after doing it.

edit: My trick question answer is to put the microphone inside a thermos and seal it. The vacuum layer in a thermos has around 100dB of attenuation. Then you can turn up the speaker as much as you want.

 

[Speedskater]

Pros have been (or attempting to) deal with this for decades in the analog world.

 

[Manukatche]

Thx friends

I imagine you have put together a trick question of some kind here... but... I'll bite anyway.

You can use an analog notch filter to reduce gain at the main feedback frequency in the room. IME this tends to work pretty well and you can add quite a lot of gain after doing it.

edit: My trick question answer is to put the microphone inside a thermos and seal it. The vacuum layer in a thermos has around 100dB of attenuation. Then you can turn up the speaker as much as you want.

No trick question, kemmler3D, I'm novice in audio and I have this challenge to solve, so I turn the question in a fun way

Notch filter is prohibited, the deal is to prevent feedback starting. It wasn't explicited in the first post, sorry ^^

Vacuum layer inside a thermos sealed ? It's a great trick answer but no the microphone has to take the ambient sound.

However, the idea to modify the air flow between speaker and microphone seems to be a good reflexion starting. Actually, I'm thinking for something like that, airflow in place of vacuum.

I've another idea, you can laugh if I said bullshit, but if we replace the microphone circular membrane by a triangular membrane, does this trick can prevent more the feedback starting, regarding the circular membrane?

Pros have been (or attempting to) deal with this for decades in the analog world.

Not the way I want. Maybe there is a reason

 

[Old_School_Brad]

If you clarify your goal, you might get more practical and focused answers. -What exactly are you trying to do?

 

[Manukatche]

If you clarify your goal, you might get more practical and focused answers. -What exactly are you trying to do?

I want to be the anti feedback champion

 

[Multicore]

Would you take it up this great challenge?

That sounds to me like a performance of The Wolfman by Robert Ashley. I would certainly take up that challenge.

the deal is to prevent feedback starting

If we are trying to be serious here, the transfer function f() from speaker to mic is indeterminate in the problem statement as you have given it. Avoiding runaway positive feedback to oscillation means keeping the loop gain g() below 1 (one) at all frequencies. But f is a factor in g. To maximize g under g < 1 you need a priori the maximum of f, which is indeterminate.

You gotta measure it. The usual way is to increase the amp gain until oscillation (g > 1) and then reduce it a certain amount from there and hope that f doesn't change too much during the operation (which is exactly where the performer in The Wolfman does her or his magic).

 

[Manukatche]

That sounds to me like a performance of The Wolfman by Robert Ashley. I would certainly take up that challenge.


If we are trying to be serious here, the transfer function f() from speaker to mic is indeterminate in the problem statement as you have given it. Avoiding runaway positive feedback to oscillation means keeping the loop gain g() below 1 (one) at all frequencies. But f is a factor in g. To maximize g under g < 1 you need a priori the maximum of f, which is indeterminate.

You gotta measure it. The usual way is to increase the amp gain until oscillation (g > 1) and then reduce it a certain amount from there and hope that f doesn't change too much during the operation (which is exactly where the performer in The Wolfman does her or his magic).

Hello Multicore

the goal isn't artistic, it's pure technique, to push the feedback starting as far ( in level sound ) as possible.

If I have to decrease the global volume, I won't reach the goal. Cause I'm actually doing that I'll tell you soon what's I'm doing, I just want, please, your enlightenments before.

Speaking of enlightenments, may you please explain with some details me what are doing the 2 musicians in the The Wolfman by Robert Ashley video? Especially that is at right?

EDIT: I've learned more about echo cancellation, it's actually quite interesting and accurate for this challenge, but not for my project which doesn't allow me to model the transfer function. That's why I wait before tell more about this project, if not you probably didn't teach me this trick

 

[Barrelhouse Solly]

I've played amplified harmonica for a long time. Feedback is always a problem. with a mic and a speaker as described, there are two things that spring to mind. One is a notch filter/equalizer. The other is a unidirectional mic.

 

[Manukatche]

The other is a unidirectional mic.

Hello Barellhouse Solly

Yeah sure ^^ what about a cardoid microphone which is triangular? Can it push the feedback further? I'haven't enough money to acquire this microphone to test

Throw an eye at 7:14, the sound engineer explains in 10 seconds how triangular membrane works

 

[Multicore]

Speaking of enlightenments, may you please explain with some details me what are doing the 2 musicians in the The Wolfman by Robert Ashley video? Especially that is at right?

I cannot account for what those two are up to but it seems compatible with the description of the version of The Wolfman that is described in the notes to the Lovely CD. http://www.lovely.com/titles/cdalgamarghen2.html

I read a more evocative account of Ashley's about a time he visited a church somewhere and the priest was intoning something using a PA system that was adjusted right at the threshold of oscillating and Ashley witnessed this priest continually adjusting his position and voice to control the oscillations that he stimulated and he realized that this priest was, without knowing it, performing The Wolfman.

Speaking of echo cancellation, such filters use a known training signal in order to learn the transfer function f. You could try to do this quickly, nearly inaudibly with a fast Farina sweep, use the result to configure a compensating filter and adjust the gain. That seems worth a try. You could try to build a device that does that automatically, that the mic plugs into, that the voice talent could activate by pressing a button. Maybe the device could also have a "margin of safety" knob. Automatic echo cancellation in communications systems uses the user's voice signal as the known training signal. That might also be possible with feedback control although the sweep might be an easier place to get started.

 

[kemmler3D]

what about a cardoid microphone which is triangular? Can it push the feedback further?

I don't think so, doesn't feedback mostly depend on resonances in the room / speaker? The mic will just have straight gain, any resonances in the mic / diaphragm won't matter, right?

 

[Multicore]

I don't think so, doesn't feedback mostly depend on resonances in the room / speaker? The mic will just have straight gain, any resonances in the mic / diaphragm won't matter, right?

I think you're right. The OP's objective is to maximize gain without oscillation. So changing mic type might select a different room mode as the biggest peak in the transfer function from the speaker but other than that the problems remain the same.

 

[Multicore]

Hello Barellhouse Solly

Yeah sure ^^ what about a cardoid microphone which is triangular? Can it push the feedback further? I'haven't enough money to acquire this microphone to test

Throw an eye at 7:14, the sound engineer explains in 10 seconds how triangular membrane works

That's a funny video. Thanks! I don't know French but lemmy guess: Dude with home studio carefully adheres to the Youtube-required scene staging, lighting, editing, and talk-to-camera stylings to explains his beliefs about a mic he's proud of? "It's amazing how much these guys look like these guys." Even the French!

For your purposes, I don't think that mic will make much difference.

If you refuse to use signal processing and want to maximize gain then your next best choice is to modify the acoustics, i.e. move the speaker, mix, and modify or treat the room. For example, put thick, heavy absorbers on all surfaces and between the mic and speaker and you'll likely be able to turn the amp up a bit. Idk if that helps or not because you haven't explained what you are trying to accomplish. Sometimes referred to as the XY Problem.

About that mic with triangular diaphragm... Idk if boundary reflection resonances are a significant concern in LDCs but if they are and you want to change them by choosing a different shape then why not use a Kepler triangle instead of equilateral? Or maybe one of the new aperiodic monotiles?

 

[Manukatche]

any resonances in the mic / diaphragm won't matter, right?

I think you're right.

It seems you are not sure

 

[Multicore]

It seems you are not sure

I believe you.

 

[Manukatche]

Idk if that helps or not because you haven't explained what you are trying to accomplish. Sometimes referred to as the XY Problem.

Yes, you're absolutely right... I just wanted you focus on feedback control without give attention to surrounding because it's changing. At home, at car, at wild, anywhere...

Anywhere, even in a anéchoïque chamber I have feedbacks. I have persistant feedback at 2.5kHz, no matters what is surrounding my set. If I try to notch it, I DSP. I mustn't modify my EQ et dynamic compression already in place. Moreover, if 2.5kHz is notched, I generate 2 bit weaker feedbacks one at 2.2 Khz and one at 2.8 kHz ( nearly ). And if I notch 2.2 and 2.8 kHz, I generate 2 bit weaker feedbacks at etc, etc, etc... That's double sentence, I notch some frequencies AND I decrease level sound.

So I have to find a mechanical way to prevent feedback starting. No matters what's surrounding the set, stereo microphones and speakers mustn't communicate.

I'm making hearing aid. Microphones are at listening position ( one micro each ear ), near to multidrivers IEMs ( 5 drivers each ear ) whiches deliver a great heavy sound. Yeah, I'm hard of hearing ( -90dB on mediums at left, that's here I struggle to prevent feedbacks ) and that's why I cannot modify my EQ/Compression, that's why I cannot use echo cancellation with Farina sweep.

The idea is to set this set ( micro + IEMs ) with specific material or geometry. I guess is the best way to explore. ( I hope is not an another XY problem ).

About that mic with triangular diaphragm... Idk if boundary reflection resonances are a significant concern in LDCs but if they are and you want to change them by choosing a different shape then why not use a Kepler triangle instead of equilateral? Or maybe one of the new aperiodic monotiles?

Ok I know now the Kepler triangle thx May you please tell me where I can found some documentation about audio application with aperiodic monotiles?

 

[Multicore]

May you please tell me where I can found some documentation about audio application with aperiodic monotiles?

Certainly not. I made it up as a joke.

 

[Multicore]

I'm making hearing aid. Microphones are at listening position ( one micro each ear ), near to multidrivers IEMs ( 5 drivers each ear ) whiches deliver a great heavy sound.

I see! That changes everything. I'm this case the room likely isn't first priority. High gain and close proximity and/or mechanical connection from mic to driver explains the consistent oscillation effect you experience.

That's a specialist area that I have no expertise in. I could imagine that the adaptive (learning) filter techniques used in active noise and echo cancellation might have some use but they are tricky to do well and engineering them takes a lot of experiments.

I'm sorry to learn you're having to deal with this.

 

[Manukatche]

I'm sorry to learn you're having to deal with this.

I'm not I work on it from 2-3 years and this 2.5kHz feedback is the last challenge I have to solve to accomplish my dream: ear the life without distortion
I feel why listen to music is so emotive, voices are transcendent, I feel very comfortable to listen at restaurant full of people. And balanced stereophonie is just marvellous, I feel my brain is balanced

I've just to wait for another advices, I posted this thread few time ago. Thx for sharing Robert Ashley video, it give an interesting auditive approach about feedbacks.