What's new

Ducking music volume to make way for voice narration - is there a better way than side-chaining ?

ManicMiner

in the Skylab landing bay
I have some voice narration over a music bed.
I'd like to lower the volume a touch when the voice is active and have it return to its normal Db when its not active.

I know that you can do this using side-chaining (as they do on radio stations), but is there any other way, or is there a plugin that can look ahead and say, "voice is coming in 2 seconds, lower volume a fraction of a second before it comes in..."

(I have Reaper, I am sure there might be an action around somewhere that might do it, but can't find it... anyway...)
rtexqzq.jpg
 
Last edited:
Aside from a sidechained dynamic EQ or compressor, I'm not aware of any shortcut method. You need the sidechain to inform the music track when the dialog is active. If I recall correctly, you can slap Fabfilter ProQ on both tracks, and then simply tell it to use one of them as a sidechain to a dynamic EQ node... But maybe I'm dreaming (not in front of my rig right now).

If you don't want to sidechain, then another approach would be to aggressively EQ the music track, carving out the primary frequencies in the dialog. That way, both can co-exist reasonably well, and provided you set the levels right at the outset, you won't need to adjust the volume of the music track.

But that's really a workaround. I'd love to hear about a plugin that can do it without sidechaining.
 
Similar to sidechaining by the product description, but the Trackspacer plugin sounds really useful to carve out room for a voiceover or whatever else you need to stay on top.

https://www.wavesfactory.com/trackspacer/

This! Great plugin. Instead of compression it uses eq curves to duck those frequencies that might interfere with the vocal frequency and therefore keeps all the other frequencies of the composition in tact. I would say, however, that if you have drums overpowering the vocal then you would still want some type of volume automation to duck the low frequencies a bit "if" they are distracting from the vocal even though they don't overlap in frequency. Just use your ears to decide.

Edit: Since this is Voice narration, if I understand correctly, then I think tradition side chain ducking with a compressor might be the best approach. If you're trying to get a second of ducking in early, before speaking, I think you'll need to manually do some volume automation at those point before the speaking, perhaps in combination with side chaining so it's not automatic but I always do what I think is best for the track and not always the easiest/quickest.
 
Last edited:
Stagger the tracks a bit, so the voice actually is a second or two ahead. Do your sidechain thing and print the music that way. Realign the v/o track and Bob's yer' uncle. At least I always thought he was...
Was going to suggest similar. Duplicate VO track, shift it earlier two seconds, bring the fader all the way down, and use that as the side chain. Turn the attack and release up a good bit so you get gradual fades. Im not a mixer, so not sure if this works. I like your idea though - might be a good middle ground between a "real" mix and auto ducking the way they do on the radio.
 
Was going to suggest similar. Duplicate VO track, shift it earlier two seconds, bring the fader all the way down, and use that as the side chain. Turn the attack and release up a good bit so you get gradual fades. Im not a mixer, so not sure if this works. I like your idea though - might be a good middle ground between a "real" mix and auto ducking the way they do on the radio.
@SchnookyPants & brek- yes this is a great solution. I never thought of this, but it makes perfect sense.
I'll also check out the Trackspacer , there's a demo of it.
 
Last edited:
maybe

mid volume automation
trackspacer
sidechain

since you take a lot of mid signal and carve out some frequencies you may even sidechain less (more volume yay)

transient designer with negative attack, perhaps?
 
Usually, dimming the freqs around 2khz in the music makes room for clarity of speech to come through. That way, you don't need to duck the music so much. (If you have a way of sidechain ducking just that band - I never tried but I'd be curious to know how it goes)
 
Last edited:
As someone suggested, you can move the audio forward if you want it to trigger the sidechaining early. The simplest thing to do is copy and shift the audio track, and sum it and the original audio to different track, and use that track to sidechain the music.

Check out this video for details on how to duck using ReaComp.

As someone else suggested, you can duck using EQ - see this video for details of how to do it using ReaEQ.

Obviously, it's not an either/or thing - you can duck using the compressor and the EQ if that works best for you.
 
Adding to the list of suggestions:
  • Dyn Eq in sidechain instead of a compressor
  • Parallel compression on the music bus (this restores some of the prowess the music looses when mixed at low volume)
  • When ducking, don't use autogain ;)
 
Just for hits and giggles:




Set the music and VO tracks out different busses. Duplicated the VO track and shifted the duplicate ahead 1 second. Sent that to the side chain on a compressor on the music bus. Set the attack at 100ms (the max on the default Cubase one), hold at 1500ms, and release at 1000ms. I think I'd like this more if I could make the attack longer.
Sorry for not preparing a better script or working with finished music. :)
 
Just for hits and giggles:
Set the music and VO tracks out different busses. Duplicated the VO track and shifted the duplicate ahead 1 second. Sent that to the side chain on a compressor on the music bus. Set the attack at 100ms (the max on the default Cubase one), hold at 1500ms, and release at 1000ms. I think I'd like this more if I could make the attack longer.
Sorry for not preparing a better script or working with finished music. :)
This sounds great, and it was over music that was fairly busy too. The 1 sec release was good. thanks.
 
You know, it can become very obtrusive if you duck the music every time someone talks and then raise it up again in between sentences. Often, it's more subtle if you only bring the music up during long pauses in the speech, and then sneak it in at the beginnings of phrases so it's not so obvious.

Of course I'm not saying *always*, just often. Or maybe usually.

And then there's the other big part - writing music that works under VO.
 
You know, it can become very obtrusive if you duck the music every time someone talks and then raise it up again in between sentences.
And then there's the other big part - writing music that works under VO.
true. the long attack and release times help smooth any jumps out. And the trick might also to be not to duck the volume by a huge amount.
I posted another thread yesterday on what Spitfire Evolution - type libraries would work as a music bed, so I also agree that the music itself needs to be appropriate.
Some of the Evo type libs are screechy / edgy, I am looking for something a bit softer and warmer. I thin I have a few examples in eDNA in A1... Looking at Symphonic strings Evo as well as Olafur Arnolds..
 
This! Great plugin. Instead of compression it uses eq curves to duck those frequencies that might interfere with the vocal frequency and therefore keeps all the other frequencies of the composition in tact. I would say, however, that if you have drums overpowering the vocal then you would still want some type of volume automation to duck the low frequencies a bit "if" they are distracting from the vocal even though they don't overlap in frequency. Just use your ears to decide.
Couldn't the same thing be done with multi-band compression, which is built into most DAWs?
 
true. the long attack and release times help smooth any jumps out. And the trick might also to be not to duck the volume by a huge amount

Slow release is okay, because the maximum settings are measured in seconds (I think my hardware comp's max is 3 seconds). But attacks are measured in milliseconds.

If I were automating brek's passage, I'd only automate the ducking a small amount, so you're probably not aware of it unless you focus on that (which you wouldn't be doing if you're listening to the voice). Then I'd do longer-term changes manually.
 
Top Bottom