Don't Just Compress your Audio - Try This First for Dialogue Audio

Curtis Judd
1.4K followers
1.4K Video Views·Jan 5, 2025
Now, to get the loudness that you want for your videos for YouTube or other places online,
we're going to demonstrate a way to do that without using compression as our only tool.
Let me explain.
First, turn off stable volume in the gear menu so you can hear what we're doing without
YouTube over-compressing our sound for this video.
Now, to get your spoken word audio loud enough for your videos on YouTube, you normally have
to boost the levels after the recording.
And here, for example, if I wanted to boost it to the levels that YouTube identifies as
their max target, look what happens.
If I just boost it like that, it's going to be all clipped and distorted and sound awful.
So we're going to Command Z.
We can't just do that.
Normally what you would do is you would do some compression.
You would compress down some of these transients that are sticking out so that your overall
audio level is more level.
But the problem is with that approach is that sometimes it ends up sounding a little bit
unnatural.
So what I'm going to demonstrate here today is potentially more transparent, and it allows
you to still boost your dialogue audio without clipping and distorting.
Now, of course, over here in DaVinci Resolve in Fairlight, there is a Dialogue Leveler.
And I can just pop that open.
If you have the Studio version, this comes with it.
This one is OK, but not great.
So we're not going to demonstrate this one.
But if you do use this one, let me just give you a couple of quick tips.
Number one, make sure you're using Optimize Moderate Levels and turn off the Lift Soft
Dialogue option.
All that does is it basically boosts the noise and the breaths in between phrases.
And that's not very helpful.
So there are a couple of tips there for the Dialogue Leveler in Resolve.
What I'm going to demonstrate is not a free plugin.
So being completely open and honest about that, I'm using iZotope Nectar 4 Advanced
with its Auto Level plugin.
I'll just pull that up right here.
And also, I don't have any sort of affiliation with iZotope, so they're not paying me to
do this.
I don't get paid if you buy iZotope Nectar 4 Advanced or RX or anything else.
All these were bought with my own money.
So I'm going to demonstrate this using the Nectar 4 Auto Level plugin within iZotope
RX Audio Editor.
You don't have to use iZotope RX Audio Editor.
I'm just doing it here because this is a better way to teach it because you can visualize
it a lot better here.
You could, of course, use the Nectar 4 Auto Level plugin in DaVinci Resolve, in Final
Cut, in Premiere, and anything else along those lines.
So this is what I typically do.
First of all, notice that our overall loudness range is pretty varied.
So we have these parts here, these transients, that are kind of sticking way out.
And overall, our loudness range, which is essentially a measure of perceptual loudness
differences, so perceptually quiet to perceptually loud things, are different.
It's not quite the same as dynamic range.
Dynamic range is measured differently than loudness range, but normally, when I finish
my videos, and I get them up to, normally I aim for minus 17 LUFS for YouTube.
And the reason I do that is YouTube actually allows a maximum of minus 14 LUFS.
And minus 14 LUFS is really good for pop music.
It's not really good for dialogue, in my opinion, and this is a personal opinion, because it
generally, if it's mainly dialogue in your videos, it tends to be kind of fatiguing to
listen to very, very low dynamic range or low loudness range audio for a long period
of time.
So I generally tend to go to minus 17, which is plenty loud, but also retains some of that
loudness range and some dynamic range.
So normally, I'm looking for my loudness range to be around six when I'm targeting minus
17 LUFS.
So if it's larger than that, you know you probably need to do some leveling or some
compression or some of both.
Let's start with leveling.
Now I'm not saying that you should never use compression.
I'm just saying that leveling actually may be a better first step, and let me show you
why.
I'm going to go ahead and open our plug-in again.
What I'm going to do is show you how to set this up to level out this audio.
And you can see, obviously, again, these transients sticking out.
The audio, the dialogue is actually fairly consistent aside from these bits that stick
way out.
So what I'm going to do is I'm going to go ahead and just use the top of this plug-in
window here and identify most of my dialogue is falling in under here, and then I just
have these few spiky things that are sticking out.
So what I'm going to do with that information, as I can see now over here on my meter, that's
at minus nine.
Okay, so I'm going to set my target dB level right here to minus nine.
So that's what I want the loudness meter to kind of target.
Now you could just use this learn button, but I haven't found it to be very useful.
In fact, it's okay, but it doesn't generally result in the type of sound that I want.
So I'd be careful about using it.
Let me just tell you how I set it up and you can decide for yourself.
The next thing we want to set is range, and the default setting for that is six dB.
What range is it?
It tells the plug-in, the Auto Level plug-in, how much it's allowed to change the audio
levels at any point within the audio file.
And in our case, I would get away from six.
I'd go down to three dB.
I find that if you leave it at six, it becomes pretty obvious that something's been done
to the audio.
It doesn't sound transparent and neutral.
So three is usually what I target, and usually three is enough for dialogue.
Also for strength, strength is essentially the sensitivity of the detection algorithm.
If you go to a really low level, it's not going to change things a lot.
If you go to a really high level, it's going to get kind of twitchy.
I find that around 40 is a pretty good place to be, so we'll go ahead and start with that
at 40.
Tame noise.
I turned this off.
I found that it actually creates more problems than it fixes.
Maybe that'll get better over time.
Okay, so with those settings, let's go ahead and play through this one portion here.
I'm going to play without the plug-in, and then I'm going to preview the plug-in.
So here's without for making this happen.
Now of course it's up to Peter and the rest of the RØDE team to actually make this happen
Okay, now here's the preview with the plug-in for making this happen.
Now of course it's up to Peter and the...
Now of course it's up to Peter and the rest of the RØDE team to actually make this happen.
Okay, you can see what it's doing here is it's showing essentially the waveform behind,
and what it's doing is it's showing that for this transient right here, it actually dipped
it.
It actually reduced its overall level by almost exactly 3 dB, it looks like.
So that's actually sounding a little bit better to me.
Let's go ahead and render.
There's still some problems.
Don't get me wrong.
There's plenty of work to do here, but you can see now that we've applied that, overall
our audio is looking a lot more even, a lot more leveled, and in fact our loudness range
is now 6.4 instead of it was up above 7, almost 8, and our loudness has come up.
It was at minus 26.9, it's now at minus 24.6, so we're already closer to our target.
So thank you Nectar for auto-level, that was helpful.
Now, I still need to get this to minus 20, and our ultimate goal is to get to minus 17.
You need to just kind of trust me for a moment here.
We need to get this to minus 20 on our mono audio, and then once we convert it to stereo,
it'll get to minus 17.
So to get to 20, we have about 4.5 dB that we need to shave off, plus I need another
1.5 dB of true peak headroom.
I'll explain that true peak headroom in a little bit, but 4.5 plus 1.5, that means 6,
so I need 6 dB of headroom.
Right now I only have 5, that means I need to come in here and manage some of these transients
that are still sticking out.
So this is where I'm going to pull out my compressor now.
I'm going to go ahead and pull out the Nectar 4, you could use any compressor, but I've
got Nectar 4 compressor here available for us, and what I want to do is I just want to
compress these peaks that are sticking out maybe above here.
Don't want to affect the overall dialogue too much, just these things that are sticking
out to give us that additional headroom.
So if I use the top of the window here, that's minus 8, so we'll set our threshold at minus
8.
So anything above minus 8 is going to get compressed down just a touch.
I'm going to change my ratio, in this case I'm okay going a little bit more aggressive
at 4 to 1.
I want my attack to be very fast at 0.1 milliseconds, and the release, just to keep it fairly transparent
sounding again, 139 milliseconds.
Turn off the auto-gain, and let's go ahead and render.
Okay, it pulled down those peaks nicely for us, go ahead and close that.
We now have 7 dB of headroom, so we have plenty of headroom.
So I can now boost our audio overall up to minus 20.
So I'll just put that in right here, minus 20, and we'll render that.
There we go.
Takes it a moment here.
Now once we've rendered that, of course we'll have less peak, true peak headroom.
We still have 2.4 dB of true peak headroom.
Now let me explain why that's important.
Also incidentally, we are now at minus 20 LUFS, so we're doing great here.
All right, 2.4 dB of true peak headroom.
The reason that's important is that once I take this audio and put it into my video
editing app, and then I'm done editing my video, when I export it, I'm going to export
it to a format that has a lossy audio codec.
Specifically, it's going to be AAC, or if you're going old school, it might be MP3.
But what happens when audio is encoded into one of those lossy formats is that some of
this headroom gets eaten up.
That's how it saves space, file space.
So we need to make sure that we keep at least minus 1.5, in our case we have minus 2.4,
so we're fine.
All right, now I'm at minus 20, how do I get to minus 17, Curtis?
How, what do I do?
How come we're not there yet?
I'm going to Command-A to select all, Command-C. I'm going to make a new file and take and
copy this mono file into a stereo file.
So we've got a stereo file, we're going to change our sample rate to 48 kHz.
I can Command-V to paste it right in here, now I have the exact same thing on the left
channel and on the right channel, and notice now that our overall loudness is minus 17
LUFS.
OK, let's go ahead and play back that part that we were working on before.
I've got to warn you, turn down your volume just a little bit.
This is much louder than we've been so far, so once you have that turned down, here's
what it sounds like.
Now of course it's up to Peter and the rest of the RØDE team to actually make this happen
over the next little bit here, and we're looking forward to seeing you release new
products that will provide us with years of service.
OK, so there's still some problems I need to address with this audio overall.
I need to take care of those breaths.
There's still a little bit of sweetening EQ that I might do.
Still sounds a little bit kind of in your face and a little harsh, but overall this
is how I can level those levels out and boost to my target loudness.
Hope that was helpful for you.
Get out there and make some great sound, and we'll talk to you soon.
