hpr-knowledge-base/hpr_transcripts/hpr1349.txt

Episode: 1349
Title: HPR1349: Melissa Dupreast helps me with Audio Compression
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr1349/hpr1349.mp3
Transcribed: 2025-10-17 23:57:05

---

All right, so this is John Culpin, Lafayette, Louisiana, and I'm sitting here with a graduate
of our university.
This is Melissa Dupreest.
Say hi.
Hi.
And she's with me here because she's going to help me learn how to do something that I'm
terrible at.
And that is to make my audio sound better without uploading it to one of my new favorite
services is this place called Alphonic, where you just upload your audio file and they
will fix it for you.
Oh, nice.
Yeah, but you have less control over it.
I mean, you're basically, you know, hoping for the best realizing it.
I'm terrible at this.
They're probably pretty good at it.
I'm just going to let them do it and see how it goes.
And I got to say it's been pretty impressive so far.
They can take something that has like very short little wave forms and when it comes back
it all seems evened out and the quiet parts are louder and the loud parts are, they still
kind of sound loud.
But anyway, it sounds pretty good, but I'd like to know how to do this stuff myself.
Okay.
You know, the nerd in me wants to understand it better.
And so I have here a lecture that I just recorded and I'm guessing that you're going
to look at that and say, man, you set those levels way too low.
Looking at a waveform like that, this meter here is going to actually tell you a little
bit better because it's telling you where, like that's, you know, that's unity.
Seeing waveforms like this, you can always adjust the sizes of those visually and so just
looking at the waveform, I don't rely on that.
I look at the consistency of the waveform, how this is a big spike compared to this,
you know, just to get an idea, but just looking at the waveform, I don't judge that.
Okay.
That's probably the part where I just said something really loud to check like what was
going to be my peak level, I guess, and if I play this, I'm going to speak loud now.
It's a teensy bit distorted, too, isn't it?
That little bit like that, you end up recording like an hour and have it all distorted
and then it's just lost.
So, we'll see how well this works.
I mean, I can hear it.
Okay, it's not like I can't hear, but like if I were to import a pot, like a professionally
done podcast into audacity here, all of those waveforms are like much, much bigger, like
they've used compression or something to even everything out and bring it up to some
kind of broadcast standard or something.
And so, this is where...
Yeah, and there are tons of different things you can use for that, too.
I'm not very familiar with audacity, I don't really use that, so let's take a little snippet
and...
I've got here maybe nearly two minutes, and let me find the compression plug-in that
seems like the...
I know.
So, these are the various things you can adjust.
So, about compression.
The way compression works is...
Sometimes I've actually compared it to medication for bipolar disorder, because people have
the manakais and then they go and have their like depressive lows.
And what the medication will do is bring down the highs and, you know, squish it to where
their high moods aren't so drastically high and their lows aren't, so that's what you're
doing with volume, basically.
But the compressor, it doesn't just, you know, bring the bottom up.
It'll bring the loud stuff down.
It'll squish it and compress it.
Then with your output gain, you can move that window up to make up for that.
So, like, I renounce the differences and then bring the whole thing up to the level you
want.
Yes.
Okay.
Yes, you can do that.
The question ratio is, well, let me explain the threshold first.
Your threshold is that cutoff point where I don't want anything to be louder than this.
So, you set your threshold as your roof.
And then your ratio is how much you're compressing.
So for, you could say, like, if you have a 3-to-1 ratio, for every three decibels, so
that actually goes above your threshold, only one is going to come out.
Okay.
So that's kind of how that works.
If you have a 3-to-1 ratio.
The default here is set at 0.5.
And in the ones that I've tried to do myself earlier, I think I've set it at 1.
But I'm a little bit shooting in the dark here.
And the maximum, is this the threshold you're talking about here, the max amplitude?
Is that a different term for this one?
Yeah, I don't like the, and again, this is, this is audacity.
I don't like the terminology that they're using here.
But do you think it's the same thing?
That's what I'm going to do, I think I'm going to push in, I'm going to push in hard
and this one.
Um, yeah, that's going to be your, your threshold.
Okay.
Um, let's see, let me read this.
So for those of you listening, there are five values that can be changed here.
The first is compress ratio.
The second is compression hardness.
Then there's the floor, which has a negative number to start out with minus 32 decibels.
It looks as the default in audacity.
And there's a noise gate fall off.
I don't really know what it is.
And then the max amplitude.
It looks like they're trying to combine two different signal processors.
They're combining a compressor and a gate and a compressor and a gate are kind of opposite
things.
Um, a gate will, uh, when you set a threshold for a gate, you, um, you're cutting off
anything that's lower in volume.
For example, if you set a gate on your microphone when you're giving your lecture, um,
anytime you stopped talking, it would cut all the sound off.
Mm-hmm.
So you wouldn't hear any.
That could sound a little choppy and, yeah, it, it depends on how smooth you set your
gate.
You can set how quickly the, I mean, and think of it like a gate, a noise has to be loud
enough for the gate to open and sound to come out and you can set how quickly the gate
closes, um, so you can do it to where it kind of like fades and it doesn't even close
all the way before you speak again, um, so that it sounds natural.
Um, in some cases, you can do a hard gate.
For example, if you're doing like a rock and roll band and you have, uh, you typically
set gates on all the toms.
So when he's not hitting any of the toms, they're all, but you can't hear the silence of
them being off because the rest of the band is playing.
But every time he hits the toms, there it is loud and clear and nice and it doesn't interfere
with everything else.
Okay.
So this is a different compressor interface.
Does this look more like the stuff that you're used to here?
Mm-hmm.
Okay.
So let's, let's look at this one and this is, um, this is the built-in compressor effect
in Audacity.
The one I was using before was a plug-in that I added because somebody had recommended
it online.
Wow.
Somebody online recommended it.
So, uh, anyway, this one looks like one that you would be more comfortable with.
Yes.
So here we have threshold noise floor ratio, attack time and decay time.
So what would I, what would you do if you were engineering these two minutes of audio?
What would you do to this?
Um, well, let's see.
I would start with, um, lowering threshold, um, cause your threshold is going to be
kind of low anyway, I think.
And let's just take this clip like where this little section where you have the, um,
loud one right next to the soft one and let's just listen to that a couple of different
ways.
Okay.
So, uh, you want to just select this bit right here?
Yeah.
And then we'll do effect compressor.
Mm-hmm.
And so what should I set the threshold?
Let's, um, I'm just lower it, um, maybe to about here to, sure, we'll try that.
Mm-hmm.
And I mean, we're just, we're just trying this.
So we're going to move this back a little bit so we don't bang it.
There's going to be some bangs in the recording right there, but, uh, whatever, I will compress
them out later.
Okay.
So the noise floor is default set at minus 40 dB, um, we'll leave that there for now.
Okay.
Okay.
Turn the ratio up to, uh, 301.
Mm-hmm.
Okay.
All right.
And, um, we'll leave the attack and decay right there for now.
Okay.
So I'm going to click okay.
And it has dramatically changed the waveform.
Let's listen.
Start recording.
Out of the levels.
Look.
I'm going to speak loud now.
And now soft.
Okay.
I'm going to check levels or else you can end up recording like a power and then have
it all distorted.
And then you have to go.
Your output gain is too much to see how well this works.
Okay.
So what's undue and then so you were saying the output gain is too much.
Yes.
Because you were clipping.
Um, that mean I need to lower the threshold more.
No.
No.
Um, see, they're not giving you an option to change your output gain.
Okay.
So.
Wait.
Click that.
Uncheck the makeup.
Yeah.
Makeup gain for zero.
Do you be after compressing?
Press based on peaks.
Eh.
We'll leave that.
Let's try it.
We're just changing the one thing first.
That's going to.
Oh, yeah.
That looks like it's not so.
Start recording.
Out of the levels.
Look.
I'm going to speak loud now.
And now soft.
Okay.
I think I'll do it.
I've got to check levels or else you can end up recording like a power.
And have it all distorted.
And then it's just lost.
It still seems.
Yeah.
It sounds distorted.
Um, you're squishing it a lot.
Do you hear that it sounds squished?
A little bit.
Yeah.
Okay.
Hearing compression is.
Um, it takes some time.
When it's compressed a lot.
You can hear it.
Raise your threshold.
Raise the threshold.
Raise the threshold.
Yeah.
Yeah.
Okay.
Okay.
Hearing compression is.
Um, it takes some time.
When it's compressed a lot.
You can hear it.
Raise your threshold.
Okay.
We'll try that.
20?
Sure.
Should I have kept the ratio where it was?
Um, three to one.
We'll try that.
Two.
A lot of this is there's no set numbers to go to every time.
Um, I always try a few things and, um, just knowing how compressors work, you know,
where to set the threshold, how to work your ratio.
Um,
I listen.
I mean, that's you.
That's you.
Yeah.
Use your ears.
Start recording.
Out of the levels.
Look.
I'm going to speak loud now.
And now soft.
Okay.
Thank you.
That's a lot better.
I've got to check levels or else you can end up recording like an hour.
Granted, we're listening on really crappy little tiny computer speakers.
Yes.
So, uh, if I were to use my synth hyzer headphones here, it probably would help.
But, uh, you know, the audience will not be able to hear that.
So, let's check one more time.
Um, I wonder if, uh, well, what, I mean, what would you do at this point?
Did you like that?
I think what's, to me, that sounded the best so far.
Yes.
Yes.
Um, I'm not sure what they mean by this noise floor.
Um, let's, let's move it dramatically and see what happens.
Way down here.
Yep.
7 minus 70 dB.
Oh, I plugged in the headphones.
Oh, okay.
My bad.
I've got to check levels or else you can end up recording like an hour and have it all distorted.
And then it's just lost.
So, we will see how well.
I'm not really sure what this works.
It almost seems like I hear more background noise.
I need to check and see what this dude had as work.
Probably you do.
If the noise floor is set lower, then that would allow more noise in.
Yeah, I would think, yeah.
Hmm.
That's a neat little feature they have on there.
Yeah, I would put that back.
The, um, raise the noise floor back.
Okay.
So raise it back up.
Where was that minus 40?
About 40.
Yeah.
But I might like to have the noise in there.
I don't know.
I can hear my students rustling around.
Spread the difference.
Hmm.
Yeah, let's make it.
We'll put it at 60 instead of minus 80.
Yeah.
All right.
Well, what about?
Let's try.
On the, my previous attempts at compression, I've used a 1.0 ratio.
I won't even go down that low will it?
Um, 1.5.
Let's just try that.
So it doesn't do as much.
Right.
Start recording.
Out of the levels, look.
I'm going to speak loud now.
And now soft.
That was excessively loud.
I think that's the loudest thing I said the horn.
Yeah.
Just because I was being obnoxiously loud for that one moment.
But if we were to take a,
you're going to have to go here in a sec, aren't you?
Yeah.
If we were going to take a more normal looking segment of it,
and then compress that,
maybe then the 2.0.
Ah.
Threshold, or 2.1, 2 to 1 ratio would be fine.
Don't make up gain.
What are this two dad is working?
Get this new gadget.
Ah, sweet.
Okay.
Yeah, check this out.
See, I've got this little keyboard.
It's like a mini keyboard, a little mouse.
I've shown my students a new gadget I got.
This is a very compelling part of the class right here.
Before classes actually started, I mean.
Well, uh, Missy, I know you've got a teacher class here
in, what, 16 minutes.
Anyway, thank you very much for talking about compression.
You're welcome.
I might have some follow-up questions.
Oh, yeah.
Feel free any time.
Um, and I'll have to familiarize myself with audacity.
It's a good thing to know, yeah.
Because it is.
It is.
A lot of people have it.
Um, I don't though.
We'll see.
It's free so you can get it.
That's right.
I have no excuses.
And that's, especially as someone who's going to,
who teaches this kind of stuff.
I mean, you work professionally in radio and in live sound
and all kinds of things.
And so you use the standard professional tools.
Right.
But if you end up teaching or even casually helping somebody,
they're not going to pay a lot of money for pro tools or something.
That's right.
And so if you have some basic knowledge of audacity,
you can just.
Oh, yeah.
And, and a lot of the basic knowledge is the same.
Like all compressors work the same way.
Yeah.
It's just every program has their own little things, you know.
They call things different names.
Yeah.
You just got to poke around with it.
That's right.
Figure it out.
Well, thank you so much.
I really appreciate it.
You're very welcome.
And I hope the HPR audience will take some instruction from it.
Bye, guys.
You have been listening to Hacker Public Radio at Hacker Public Radio,
does our.
We are a community podcast network that releases shows every weekday
on day through Friday.
Today's show, like all our shows,
was contributed by an HPR listener by yourself.
If you ever consider recording a podcast,
then visit our website to find out how easy it really is.
Hacker Public Radio was founded by the Digital.Pound
and the International Computer Cloud.
HPR is funded by the binary revolution at binref.com.
All binref projects are crowd-responsive by linear pages.
From shared hosting to custom-private clouds,
go to lunarpages.com for all your hosting needs.
Unless otherwise stasis,
today's show is released under creative comments,
attribution, share a line,
lead us our license.