Page 1 of 2

Question for signal processing gurus

Posted: Sun Feb 10, 2019 4:04 pm
by himhimself
Is there a way to duck just a portion of the freq bandwidth for voice? I hate the "monster truck" type ads on the radio that overuse ducking, but wondered how it would work if you could just drop the frequencies where the voice was. Like during events where music is playing but you want to make announcements or commentary and keep the music solid but not make the announcement painfully loud to get over top of the music. I have a pa2 but didn't see any easy way to do this. Any ideas? Would what I'm proposing do what I want?

Re: Question for signal processing gurus

Posted: Sun Feb 10, 2019 4:37 pm
by Bill Fitzmaurice
I can't say how it's done but I recall seeing somewhere that it can, so keep searching.

Re: Question for signal processing gurus

Posted: Sun Feb 10, 2019 5:24 pm
by Strange Kevin
What mixer are you using, I think little digital mixers like the X-air can do it... I haven't tried though, so done quote me on that...

Re: Question for signal processing gurus

Posted: Sun Feb 10, 2019 9:11 pm
by CoronaOperator
What you want to do is use the mic channel to "sidechain" trigger a "multiband compressor" or a "dynamic EQ". How you get there affordably is another problem.

The waves f6 plugin does exactly what you want. Setting up a daw for live use isn't cheap.

You may be able to set up an x-air/x32/m32 digital mixer to do what you want, not sure if you can sidechain those effects though, I don't have one of those to play with.

Cheapest way would be to use any mixer and just manually take down the mid eq dial on the music channel strip when you speak. Use lots of compression and a bit of EQ on your vocal channel to cut through as well.

Re: Question for signal processing gurus

Posted: Sun Feb 10, 2019 9:20 pm
by Bruce Weldy
Geez.....maybe I'm old school, but why not just turn down the music to talk, then back up again?

There are commercial units that have automated ducking - used in bar and permanent installs so that the mic at the bar can turn down the music that comes from another location in the venue.....but, all that has been mentioned seems like a lot of trouble when just pulling the fader does what you want.

Re: Question for signal processing gurus

Posted: Sun Feb 10, 2019 9:32 pm
by CoronaOperator
Bruce Weldy wrote: Sun Feb 10, 2019 9:20 pm Geez.....maybe I'm old school, but why not just turn down the music to talk, then back up again?

There are commercial units that have automated ducking - used in bar and permanent installs so that the mic at the bar can turn down the music that comes from another location in the venue.....but, all that has been mentioned seems like a lot of trouble when just pulling the fader does what you want.
He wants to keep the bass line pumping and just duck the mids to hear his voice.

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 12:38 am
by CoronaOperator
Strange Kevin wrote: Sun Feb 10, 2019 5:24 pm What mixer are you using, I think little digital mixers like the X-air can do it... I haven't tried though, so done quote me on that...
I think you would have to get creative to do selective frequency ducking on an X-air/x-32/m32. I would split your incoming L/R music channels into 3 duplicate channels each. So your left input would plug into channels 1, 2, and 3, your right input channels, 4, 5, and 6. Then using the channel EQ's, set up high and low pass filters at your frequency of choice for the 3 channels - channel 1 and 4 is bass, channels 2 and 5 is mids, channels 3 and 6 are highs. Use channel 7 for your mic. Then use your channel 7 mic as a sidechain to trigger gates on your bass, mid, high channels. You could have different amount of ducking for each pass band. Then throw channels 1-6 into a subgroup so you can eq and adjust music volume level all in one place.

FYI the ducking on ALL the musicgroup mixers has a couple firmware flaws. Attack and release knobs are reversed (attack is release and vice versa), the attack fader (which is actually release) only goes up to 120ms which is pretty abrupt, it should have an option for much much longer times than that. 3rd flaw is that even when set to sidechaining, the original channel source can trigger the ducker if it exceeds the threshold you set. You have to run those channels low in level and run the mic channel hot to avoid the music from ducking itself. :noob: Why this hasn't been fixed in a firmware update is anybody's guess.

It all sounds complex but once set up and tuned to your liking and saved you never have to mess with it again.

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 2:35 am
by Grant Bunter
That's a lot of work, and a lot of channels CO, but it would get the result.

I'm going to wait until I hear about the mixer first...

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 6:50 am
by CoronaOperator
himhimself wrote: Sun Feb 10, 2019 4:04 pm I have a pa2 but didn't see any easy way to do this. Any ideas? Would what I'm proposing do what I want?
Just to add, the pa2 is a system processor that controls the speakers, those units can't do what you want. You need a high end digital mixer that can do it, or a work-around on a cheaper digital mixer like I suggested, or a computer based audio console with a plugin that can do it. Stand alone hardware units to do that don't exist anymore and the used ones that do are broadcast/studio units costing thousands or tens of thousands of dollars.

Yes, I think what you are proposing would work. You would also need to highpass, eq, compress and limit the vocal channel to get good results. All easy to do in a digital mixer. If you duck your mids by 15dB's, you need to squash the dynamic range of the vocal down to about 10 dB's to fit in the gap between the ducked response and the rest of the music volume. The difference between the music level and the crowd noise (dynamic range of music) would also dictate how much compression you need. A roaring crowd would need more compression than a quiet critical listening crowd. I think you would get best results by daisy chaining a compressor and a limiter in the vocal channel. Compressor to lower the dynamic range, limiter to stop any peaks getting through.

If you are using a DJ mixer to play music you can do the same thing (less the vocal effects) by manually turning down the mids outlined in red while you speak. If you are speaking on behalf of another DJ and don't want to reach over him while he plays, then put the outputs of his mixer into the inputs another mixer off to the side and do the same thing.

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 7:02 am
by Chris_Allen
tc electronics do the c300 which has fixed bands in which to apply the compression.

I would run compression on a subgroup of everything you wanted ducking, leaving the bass/kick on an unaffected subgroup. If you want to reduce some channels more than others, add a proportional low mix to the other subgroup.

EDIT: Failed to spot this was DJ type application for which this won't apply.

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 7:19 am
by CoronaOperator
Chris_Allen wrote: Mon Feb 11, 2019 7:02 am tc electronics do the c300 which has fixed bands in which to apply the compression.

I would run compression on a subgroup of everything you wanted ducking, leaving the bass/kick on an unaffected subgroup. If you want to reduce some channels more than others, add a proportional low mix to the other subgroup.

EDIT: Failed to spot this was DJ type application for which this won't apply.
You may be on to something though! How about you scoop out the mids on the DJ inputs (channels 1 and 2), assign them to a subgroup, EQ that subgroup to cut out highs and lows. Adjust volume level on sub group so that they (hopefully) blend in like the original, then just compress/duck the subgroup sidechained to the vocal mic? Then use a dca to control your overall volume level if need be, although for DJ work your master outs can do that. That way you only use 2 channels instead of 6

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 8:22 am
by Bruce Weldy
CoronaOperator wrote: Sun Feb 10, 2019 9:32 pm
Bruce Weldy wrote: Sun Feb 10, 2019 9:20 pm Geez.....maybe I'm old school, but why not just turn down the music to talk, then back up again?

There are commercial units that have automated ducking - used in bar and permanent installs so that the mic at the bar can turn down the music that comes from another location in the venue.....but, all that has been mentioned seems like a lot of trouble when just pulling the fader does what you want.
He wants to keep the bass line pumping and just duck the mids to hear his voice.
Aux-fed subs......just pull back the mains and let the bass keep going. Not as precise as just pulling mids, but would get you part way there....maybe good enough for what you want.

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 8:59 am
by himhimself
Thanks for all the replies. Unfortunately my mixers are both analog. Music is coming from a dedicated laptop fed into 2 channels. The mixers do have decent eq, so maybe just spinning the mids all the way back would create a hole for the mic. I'll try that, although finding a way to do it automatically would be better, especially when someone else is doing to commentary. I've considered running with a daw, just cuz that would allow lots of fun stuff with plugins too, and I've wanted to play with waves maxxbass for a while.

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 9:56 am
by jimbo7
All my dj mixers have a "talkover" button that pretty much pulls the mids and some highs and lows. Yeah, it does lower the the overall sound some, but it's the same thing you hear when listening to the radio.

Re: Question for signal processing gurus

Posted: Mon Feb 11, 2019 10:08 am
by CoronaOperator
So I did a little experiment on my UI-16. It doesn't have sidechain ducking like the x-air's do but I just muted the second channel manually to prove the concept.

Here was my setup: Headphones, mic, UI-16, and cellphone for music off of youtube
thumbnail_20190211_072131.jpg
I played with EQ cuts to the music channel to find the freq slot that matched my voice. Then played with vocal mic compression, attack, release rates, high cut and level to fill the hole I just made in the music.

I then duplicated the music channel and carved out the opposite of what I cut and added that back in. Then I just muted that channel when I spoke on the mic to simulate a sidechain ducking compressor.

Here is my mic high pass:
thumbnail_20190211_072048_0 (1).jpg
...and as predicted, I had to use HEAVY compression to be heard without raising the overall sound level:
thumbnail_20190211_072027_0.jpg
Here was the cut to the music channel to make a slot that fit my voice:
thumbnail_20190211_072106_0.jpg
Here was the duplicate channel to replace that cut mixed over top:
thumbnail_20190211_072114_0.jpg
Now I just played music and when I spoke I just muted manually the duplicate channel with the mid boost. A ducker would be MUCH Much smoother as you can adjust the attack/release times (fadeout and back in) to be much more seamless.

Here is the result:

http://s000.tinyupload.com/index.php?fi ... 4106894758

It recorded very quiet so I used audacity to give a 30db boost to the recording so that might affect the noisefloor some but I think the results were good and the concept sound.

The X-air 12 will automate this process and is only $299 delivered :shock: . Lots of power for the price although if you think you might get into bigger shows I'd get the 18 channel one.