I’m currently using the Dwarf to handle both guitar and vocals in our heavy metal band.
mic ----1----**** DWARF**** --------- mixer ----- PA speakers
gtr -----2---**************
The rehearsal room isn’t really treated for the loud volumes we produce and the vocal mic starts feedbacking. (with or without DWARF btw)
I was wondering; I’ve seen Feedback killer/controllers as hardware before, would something similar be feasable on the Dwarf? Could a plug-in detect and suppress feedback somehow?
I would suggest getting the feedback suppressed outside the Dwarf. Or else you will turn up to a gig and need to be plugging a computer in to adjust settings to adjust for the different room.
I don’t know so much about how feedback suppressors work but part of my job here is to inspire.
I’m curious to learn whether it is possible or not and what the steps towards realisation could be if it is.
It might be better to do that outside the Dwarf.
But that answer could be the answer to a dozen threads here as well.
What is the challenge and what can be done to tackle the challenge?
It is because of the “learning” that is necessary?
In an implicit fashion, that was part of the question.
Could the learning part be handled in in the Dwarf?
Could it learn and adapt on the fly?
Do you need a “listening period” first so it can learn (bind a button, start listening when pressed, dwarf identifies problem frequency/frequencies, press again to stop) while another bound button turns on/off the suppressor, using what he learned?
If not, could there be another way?
(example: a listerne that identifies a problem range and you manually set the freq interval with 2 turning knobs?) etc.
I don`t know what mic you are using but I would start with that.
We kind of had the same problem. Not feedback but I record most of our live sessions and wanted to isolate the vocals more from the drums. The SE V7 does a good of that - its constructed a bit different than a SM58 but reduces the background quite a bit. It has a supercardioid polar pattern.
And Its pretty inexpensive and sounds good.
Another suggestion but a completely different ballpark is to use inear system. Before we used our X32 i`ll always had low budget Version with a headphone amp for tight spaces. That way the vocals or stereo mix go straight out of the mixer and at least the singer could’d hear themselves without blasting the monitors.
A very complex AI- or sample-based system that will match a suddenly rising (spike) frequency at a specific frequency or set of frequencies while also differentiating them from other sounds at those same frequencies – therefore, acting like a spectrum analyser, which involves a considerable latency because it would need in the lowest possible estimate some 2048 samples – a 42.6 ms just to kick in, plus processing time.
It would most likely require that feedback noise be sampled in advance (which is what many of the current noise removal tools employ since CoolEdit Pro in the 90s, currently Adobe Audition.)
And one such tool would have an extremely high CPU consumption. Current software does that mostly in post-production, and it takes 100% CPU for a long period.
Plus, they are expensive. Think Izotope Rx prices.
That too, but also – and most importantly – because it will change according a multitude of factors: walls, the material of which they are made, the angle at which they react to the speaker, the phase of both speakers, the distance of the microphone to the speakers and wall, the diameter of the speaker cones, the diameter of the microphone’s diaphragm, … (it’s a long list.)
Feedback should not be removed with a specific frequency-cancelling filtyer (such as a notch filter) for a number of reasons. It rises due to very different and to some extent unpredictable factors, such as the singer walking on stage and inadvertently entering the range of a speaker.
But most importantly, feedback occurs in frequencies that are within the range of many instruments. So if one employs a graphic EQ to tame a 1KHz “howling” feedback, when the guitar player hits the B5 or C6 notes in the neck, they will get muffled.
The EQ removal technique is a last resource, when all else fails.
If you need, you can use one of the parametric EQs in the Dwarf, since they also offer Q factor (the “width” of the centre-frequency) and with that you can adjust it to act in a narrow band, so to speak.
It is absolutely, decidedly better to remove that before any mixer, amp, multi-effects, etc.
The best advice has already been given by @Ant above: set yourself up in the room so that you have no feedback.
Or else, take @spunktsch’s advice: first, dynamic micrphones only – highly directional even better (thanks @Kevin). Never condenser, ribbon, etc. Then, if at all possible, run your rehearsal in headphones. Snarky Puppy recorded a full album live in a large room with audience, entirely with headphones (even for the audience.) Obviously not practical in a bar with 500 people, but for rehearsals it’s a lifesaver – especially for your neighbours.
Many good insight guys, some things to try here.
Thanks for answering my questions because I’m a curious fella.
I completely get why this is very hard to achieve in realtime processing
I’m the singer in the band.
I’m not new to singing in a microphone and I use a AKG D5 dynamic with a supercardioid pattern.
I tried several different locations and angles vs the speakers.
Turning down the band is kinda hard with heavy metal drumming going on
We have little we can do in the room; it’s not ours, it has flat wall and besides the sofa, there isn’t much to do. Flat hard walls and ceilings.
Cutting out everything unneeded frequencies in EQ seems like a good idea. I already applied low and high pass filters for the reason but I guess I can be more selective.
I’ll keep in mind to keep incoming gain as low as possible
‘rinsing the room’ with a graphic EQ.
Intrigued by this passage; will read more on it later
I was considering a little monitor for myself,
ok;
You guys inspired me to experiment some more with a few things, I’ll let you know what progress I made next rehearsal!
Cardioid is better for rejecting the rear, I’d recommend a trusty sm58.
Are you trying to balance vocal to unmic’d drums? if so and you can’t without feedback you will need a more powerful PA in a bigger room - or simply ask the drummer to play a little quieter - all drummers think they have to play at the very top end of their dynamic range - they don’t, even metal drummers!
