July 15, 2025 • 41 mins
Confused about which audio sample rate to choose for your music? Whether you're recording your first track or mastering for Dolby Atmos, knowing the difference between 44.1kHz, 48kHz, and 96kHz is more than technical—it’s essential to how your music sounds and where it ends up.
In this episode of Inside the Mix, mastering engineer and Berklee professor Ian Stewart returns to demystify one of the most misunderstood elements in music production: digital audio sample rates.
We answer common questions like:
- What is a sample rate in audio production?
- Should I use 44.1kHz or 48kHz when mixing?
- Does a higher sample rate improve sound quality?
- What sample rate should I use for streaming or mastering?
Ian explains the Nyquist-Shannon theorem, how aliasing impacts your mix, and why 48kHz is the new practical standard, from YouTube to Bluetooth devices to Dolby Atmos. Plus, we discuss how higher sample rates give your plugins more headroom for harmonic generation and cleaner processing.
Don’t miss this one if you’re serious about choosing the best sample rate for your music. And stay tuned, next episode we dive into dither.
Links mentioned in this episode:
Follow Ian Stewart
Ways to connect with Marc:
Book your FREE 20 Minute Discovery Call
Radio-ready mixes start here - get the FREE weekly tips
Grab exclusive access to BONUS content
Follow Marc's Socials:
Instagram | YouTube | Synth Music Mastering
Thanks for listening!!
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Marc Matthews (00:00):
Does sample rate actually make your music sound
better?
You've probably heard argumentsover whether you should record
at 44.1, 48 or even 96, buthere's a real question does it
really matter?
In this episode, we're divingdeep with someone who not only
understands the theory, butlives it every day in the
mastering room.
Ian Stewart (00:20):
You're listening to the Inside the Mix podcast with
your host, mark Matthews.
Marc Matthews (00:23):
Welcome to Inside the Mix podcast with your host,
mark Matthews.
Welcome to Inside the Mix, yourgo-to podcast for music
creation and production.
Whether you're crafting yourfirst track or refining your
mixing skills, join me each weekfor expert interviews,
practical tutorials and insightsto help you level up your music
and smash it in the musicindustry.
(00:43):
Let's dive in.
Hello folks, and welcome toInside the Mix.
Today's episode is part one of aspecial two-part conversation.
This might be the first time.
No, I tell you what it's thesecond time I've done a
two-parter in all 200-plusepisodes now.
So we're with mastering,engineer, educator and all-round
audio wizard, ian Stewart.
(01:04):
Ian is a returning guest.
Check out the listener.
Favorite episode, episode 165what is midside eq?
One of the most popularepisodes of the podcast, might I
add.
He's the founder of flow townmastering and an assistant
professor at berkeley college ofmusic, where he teaches the
science and art of audiomastering.
His resume includes work withlegends like krs1 and he's a
(01:26):
regular contributor to iZotope'sLearn blog and Wave Labs as
well, recently on YouTube.
In this episode we are talkingall things sample rate, what is
it, what it isn't, and why.
Getting it right could make orbreak your project.
That's enough from me, ian, howare you?
And welcome back.
Ian Stewart (01:43):
I am excellent, Mark.
Thank you so much.
I'm excited to be here.
I am in lovely New England,where today it is a balmy
90-something degrees, which forthe rest of the world is
30-some-odd.
I don't know.
So, it's sticky up here.
I've got the windows open.
If you hear a little extraneousnoise, that's probably from my
(02:04):
end.
So I I do apologize for any ofthat, but I am great, it's
summer, we're doing it, life'sgood working on cool music, so
yeah love it.
Marc Matthews (02:12):
Yeah, that's the attitude man.
I find, like when it I thinkthe audience, when they listen
to like the back catalog of thepodcast, they could tell the
seasons, because there's certainlike episodes or series of
episodes and then you just hearlike animals, like birds,
seagulls, all this stuff, so youcan kind of like the podcast
when you're listening to it.
You move with the seasons, whichis quite cool yeah I appreciate
(02:35):
it, but I know it's been hothere in the uk, as we were
saying off episodes, so it's uhhot, hot time, especially when
you're in a studio as well, likeit's, and you haven't got that
properly that the hvac going andstuff yeah, we're not.
Ian Stewart (02:47):
We're not quite it was.
It was, like you know, cool andrainy until a week ago and then
, all of a sudden, bam summer.
Yeah, so yeah, it's caught me alittle off guard yeah, yeah, it
always doesn't.
Marc Matthews (03:00):
It always catches us off guard, to be honest.
Yeah, it doesn't take much foreveryone to start getting the
barbecues out and then theshorts and t-shirts, even though
it's probably.
I mean, it has been hot here,but sometimes it's in the late
teens and you still people arelike I don't think it's that hot
to be like that.
But there we go, so in thisepisode, folks, we're peeling
back the layers of confusionaround sample rates.
This is kind of almost like aback to basics, really.
(03:21):
From 44.1 to 96 and everythingin between, Ian's going to help
us separate technical truth fromaudio folklore.
Think of this as your go-toguide for demystifying sample
rates.
But before we do that, if youwant weekly tools and tips to
make your music playlist worthy,click the link in the episode
description and get my freeweekly tips direct to your inbox
(03:41):
as well.
No spam, no fluff.
Just once a week I'll send youweekly tips and tricks, not just
for me, but from alsoindividuals that have been on
the podcast and stuff that Ifind out about as well.
So enough for me.
Going back to sample rate, Ithink to start our discussion,
it's probably quite importantjust to outline what sample rate
actually is.
What are we referring to inaudio?
Ian Stewart (04:01):
Yeah, so when we record digital audio, right,
when we want to get somethingfrom a microphone for those of
you that are just listening, I'mpointing to my microphone and
we want to get it into a file,we have to do a bunch of
conversion.
And part of the process is, youknow if back in the old days,
when we use tape or othermediums, right, that it was just
a voltage that came out of themic, went through a preamp and
(04:23):
that voltage would get stored,mad magnetically on a tape.
But when we go digital, we haveto measure that voltage and
store it as a value.
And because we can't do thatinfinitely fast and continuously
, we kind of take a sample, we,we measure that voltage a bunch
(04:44):
of times every second and savethat.
Then we say, okay, the voltageis this, and we save it as a
value, as a digital number, andwe move on to the next sample
and say, oh, now the voltage isthis and we save that, and so
the sample rate just tells youhow many times every second
you're taking that sample.
So for 44.1 kilohertz, that's44,100 times every second.
(05:07):
We're measuring the audio.
So kilohertz is thousands ofhertz.
A hertz is just a measure ofcycles per second.
So it's just how many thousandsof times every second?
Really, we're measuring thevoltage of the audio and storing
it as a digital number, andthat's pretty much the root of
the audio.
And storing it as a digitalnumber, and that's pretty much
the root of it.
Marc Matthews (05:26):
Yeah, yeah, yeah, definitely.
So whilst we're on that, justunpacking the 44.1.
So if we were to dive into sortof the frequency that we can
hear from 20, let's say, 20hertz to 20,000 hertz, can you
maybe just explain to theaudience why it is that we're
using the number 44.1?
Ian Stewart (05:46):
explain to the audience why it is that we're
using the number 44.1?
Yeah, absolutely so.
Um, if, if you've ever heard ofthe shannon sampling theorem or
the nyquist sampling theorem orthe shannon nyquist sampling
theorem or anything, any kind ofcombination?
Uh, these were two engineersback in the oh, I should know a
time range, but I I want to saymaybe even the 30s, 1930s, I
think Nyquist certainly.
I should have been moreprepared for this.
(06:07):
I think Nyquist worked for BellLabs, but he kind of pioneered
this digital sampling theory waybefore we even had the
technology to do it.
And what he computedmathematically is that if we
want to store any givenfrequency say it's 100 hertz,
pretty low, right.
But if we want to store that,we need two sample points in
(06:29):
that waveform, right.
So if it goes from zero up toits peak, back down to zero,
down to negative one, itsnegative peak and back.
Basically, the idea is we atleast need to store the maximum
and minimum, those positive andnegative peaks, to be able to
accurately describe what thatfrequency is.
So you need two samples forevery period of a waveform.
(06:52):
So commonly the range of humanhearing is given as 20 hertz to
20,000 hertz, 20 kilohertz and,to be honest, that's pretty
generous Really.
Once you get out of your teens,maybe even a little younger,
20k is probably gone and notcoming back, but we kind of take
(07:12):
that as the standard acceptedrange.
And so when digital audio wasreally being developed by Sony
and Philips back in the 70s or80s, they were trying to figure
out what's the ideal sample ratethat we need to capture audio,
and they knew they wanted to putthe data on a disc, right, an
(07:34):
optical disc, cds.
And so they're trying to playthis balance of having good
enough quality that they cancapture everything that human
range of hearing can appreciate,but not storing more data than
is really necessary, right?
They don't want to, becausemore data means you can fit less
music on the disc ultimately.
So they said, okay, well, if wecan hear up to 20 kilohertz,
(07:59):
that means our sampling rateneeds to be at least 40
kilohertz, right?
So if we need two samples forevery period, it's got to be
double.
Whatever the frequency you wantto capture is Sample rate needs
to be double that top frequencyyou want to capture.
So if we're trying to capture upto 20 kilohertz, we need at
least 40 kilohertz sampling rateand at that stage, one of the
(08:23):
things that you have to do isyou have to remove frequencies
above that.
So if you say I'm going tosample at 40 kilohertz, there is
a requirement that you not letany frequencies above 20
kilohertz into the digitalconversion chain.
And if you do, that results insomething called aliasing, and
maybe we can get to that alittle bit later if you want.
(08:45):
If you've heard of aliasing,that's kind of where this idea
comes from.
And so you have to put a reallysteep low-pass filter on your
input audio before it hits theconverter.
All right, stuart, can we talkabout digital audio please?
So back in the 70s and 80s thesefilters like to make.
(09:08):
A really steep filter was tough.
You couldn't.
There are limits to doing itdigitally.
Doing a really steep analogfilter was tough.
So basically they said okay,you know what, we'll have the
filter start rolling out at 20kilohertz, but we need a little
bandwidth for it to get thesignal low enough that it
doesn't create aliasing.
So they add a little bit more.
(09:29):
They go up to 22.05, right, twoand a half kilohertz, or two
and a whatever 2050 kilohertz ofextra bandwidth to allow the
filter to do its job.
Now you double that extrabandwidth to allow the filter to
do its job.
Now you double that, you get44,100.
So that's where that 44.1Ksample rate comes from, is
(09:53):
saying we want the 20 kilohertzbandwidth of audible frequency,
then we need a little extra roomfor the filter to knock out
stuff above it, and then that'swhat our sample rate's going to
be.
Marc Matthews (09:59):
Love it.
This is great because it takesme back to many, many, many
years ago now, when I was doingmy studies and we did a whole
thing on um, sampling andsynthesis and it was, yeah, we
sort of dive, dive deep intothis and hands, and, yeah, a lot
of it I have forgotten becauseit's not something that, oh, I
mean, we play, we, we look atsample rates etc and we're
setting sample rates, but beyondthat I don't really dive too
(10:22):
deep into it.
So this is a.
This is a great refresher for me, yeah you, you usually don't
have to think about it too no,no, exactly, yeah, yes so, um,
but you say, when you say thingslike aliasing nicholas, like oh
yeah, yeah, that does ring abell.
now I remember going throughthis and what you described
there about how, uh, when we'retaking those two, two samples
and then we're doubling it andthen you've got that, basically
(10:43):
that buffer, when we're going upto 44.1.
But it kind of leads me onnicely to my next question,
which is when we start to enterthe realm of 48, 96 and beyond
and everything in between, whatwould you say is the standard
sort of sample rate for musicproduction?
If someone were to ask you okay, well, I'm recording a band,
I'm recording a band, I'mrecording an artist, whatever,
(11:04):
I'm just doing a recording.
Yeah, what would your advice be?
Ian Stewart (11:08):
with regards to sample rate?
That's an excellent question.
I think not that many years agothe answer might have been a
little trickier.
Actually, definitely when CDsdominated.
44.1, you knew you were goingto have to end up there at some
point.
44.1, you knew you were goingto have to end up there at some
point, and sample rateconversion technology was not as
good, so there was maybe anargument to be made for
(11:30):
recording at 44.1 some years ago.
These days, though, I thinkit's pretty clear cut that we
could say a really good standardis 48k.
Um, for a number of reasons.
One, we know atmos is stillbeing pushed, and I think it's
still.
For me, the jury's still out onon how long-lived atmos is
(11:54):
going to be as a format and ifeveryone's really going to start
doing it or whatever.
That's topic for another dayyeah, an interesting one yeah,
but, but Atmos can be at 96k,but it's kind of a pain in the
butt, yeah, and so practicallyAtmos mixes need to be at 48k,
like it's one or the other, andreally the vast, vast majority
(12:16):
of them are at 48k.
Youtube recently updated theiraudio recommendations For the
longest time.
It was funny.
They recommended 44.1, but theyrecently updated their
recommendations.
Recommendations for the longesttime is funny.
They recommended 44 one, butthey recently updated their
recommendations to 48k.
48k has been the standard foraudio that accompanies video for
pretty much ever.
Marc Matthews (12:33):
I was just going to mention that.
Yeah, is that with youtube?
Now, is that for youtube musicwhere they're recommended for?
Ian Stewart (12:38):
yes, for, for all of youtube not just youtube
music, for all of it and eveneven for the main video site.
For a long time theyrecommended 44.1 because they
encode it to other codecs orwhatever, right, but now they're
saying 48.
So there really has been astandardization in the last
(12:59):
handful of years of almost alldeliverables moving to 48k as a
minimum.
Yeah, so I would say 48k is agreat sample rate to record at,
to work at, certainly to deliverat.
I still, as a masteringengineer, I still give people
1644 files for CDs in case theywant to do that, but the main
(13:22):
deliverable I send out is 24-bit48K, yeah.
So, yeah, I think there's abenefit there, just in terms of
standardization and not havingto go through extra rounds of
sample rate conversion.
And then the other thing.
So I think the natural nextquestion is well, if 44.1 is all
(13:42):
we need to capture our range ofhearing, and really a little
bit more, what's the benefit ofgoing up higher than that?
Uh, and there there are a few,and if I'm jumping the gun here,
feel free to bring me in, but Iwas going to say it was going
to be.
Marc Matthews (13:55):
My next line of questioning was with regards to
so we're getting in sync.
Ian Stewart (13:59):
we're starting to know each other, mar, mark,
Please carry on.
So, yeah, the nice thing about48K is that basically it allows
that filter that rolls off thehigh end to be a little gentler,
and while I'm not aparticularly woo-woo kind of guy
(14:24):
, give me facts and good, hardevidence, or it just doesn't
really matter.
I you know and this thisextends to many areas of my life
in audio, though, there's nogood evidence that audio content
above 20 kilohertz makes anydifference to anyone.
Uh, there's never been a testthat has conclusively shown that
there are some people who canhear to 25 kilohertz and it
changes Like it's.
There's just no data on that.
Yeah, it doesn't mean thatthose people don't exist.
(14:45):
It just means, in all thestudies that have been done,
we've never found one right so,starting with that as the
baseline, we know that, however,what filters do is they change
the phase response.
Um, some of these filters canbe linear phase.
A lot of them are not.
They are minimum, becauselinear phase filters add latency
(15:06):
, uh, so a lot of times we'retrying to get fast performance,
so there are minimum phasefilters that that do this, roll
off and those do have a phaseshift that's associated with
them, and a phase shift is notexactly the time the same as a
time shift, but it's.
You can kind of think of itlike that.
And one thing that we do haveevidence to say is that human
(15:30):
ears are unbelievably sensitiveto timing, like like crazy and
there's a good evolutionaryprecedent for this that if
there's a tiger that's half ameter closer to you, you want to
hear that one first and payattention and get away from it,
and not the one that's a littlefurther.
So we're really sensitive totiming and so these phase shifts
.
A lot of times people will heardifferences in different sample
(15:53):
rates and I don't know that wecan say this conclusively, but I
think there's enough strongevidence that points at this as
the conclusion that really whatyou're hearing is the phase
response of those filters.
And by going to 48k or 88.2 or96, you allow that filter to be
a little gentler and notactually get in the way and
(16:13):
impact the stuff that you canhear.
The frequencies you can hear,which may be down at even 16, 18
kilohertz.
That phase shift may stillimpact them a little bit.
So that's one of the benefits,right, it can sound a little
more natural.
So nonlinear processing issomething that's dependent on
the input level, right?
So like a compressor or alimiter or a clipper or a
(16:35):
saturator and all of thesethings.
One of the characteristics thatthey have in common is that
they create overtones.
So if you feed them a onekilohertz tone, depending on the
type of processing, they maycreate overtones at 2k and 3k
and 4k and 5k, or, it may justbe odd, it may be three and five
and seven and nine and so on.
(16:56):
So when you feed a nonlinearprocessor a very high frequency,
there's a chance that it startscreating content that is above
our Nyquist frequency, and theNyquist frequency is that half
the sampling rate.
So if we're at 48K, the numbersare a little easier.
So let's just say, for 48K theNyquist frequency is 24
(17:18):
kilohertz as the highestfrequency that we can capture,
and so if we start generatingcontent that's over that, now
that breaks that band limitingrequirement.
Remember, I said earlier thatwe have to have this filter
because we can't allow stuffinto digital audio that's above
that half the sampling frequencyrate.
So now we have to deal withthat, and you can either not
(17:44):
deal with it, which is what someolder plugins that was kind of
the naive way when we didn'tknow better you didn't deal with
it and it turns into aliasing.
Yeah, the other common way todeal with it is to oversample
the plugin.
So you internally switch to ahigher sample rate in the plugin
, you allow for that higherfrequency content, then you
(18:10):
filter it back out and then youkind of re-inject that into your
main audio stream.
Um, and so there's atheoretically there's an
argument that if you work at 96k, uh, you have even you kind of
have to worry about that alittle less yeah, so.
Marc Matthews (18:19):
So with regards to the overtones that being
created by non-linear processing, but how?
How is that being created?
How are the what?
How are the overtones that arebeing created by nonlinear
processing?
How is that being created?
How are those overtones beingcreated?
Ian Stewart (18:29):
It's basically what's happening is the shape of
the waveform is changing.
So let's think clipping.
Clipping is kind of the mostextreme example and also creates
a lot of overtones.
So say, if you clip a sine wave, right, you've just got your
regular sine wave, and then youpush it up into clipping, the
top of it gets just flattenedoff.
Marc Matthews (18:50):
Yeah.
Ian Stewart (18:50):
And you can.
There are kind of two commonways.
Marc Matthews (18:54):
We're getting a little deep and again call me
back at any point, if you need,but I'll go for it Two common
ways that we think about audio.
Ian Stewart (19:02):
We talk about time domain, we can talk about
frequency domain and they'reinterrelated and you can't
really disentangle them.
So the time domain version of aclipped sine wave is that it
goes up and then it flattens offat the top and then it goes
down, it flattens off at thebottom.
The frequency domain version,if we think about what happens
to the frequency content of thatsine wave, in order to flatten
(19:27):
the top off, you basically haveto add higher frequencies with
different phase relationshipsback into that sine wave.
That is mathematically what'shappening, right, and so
clipping or any other sort ofwave shaping where you you're
really changing the amplitudeshape of of the waveform, it
(19:50):
changes the shape of it.
The change of the shape makesthese extra harmonics and
frequencies pop out.
If you try and take thosefrequencies out, the shape goes
back to what it originally was.
Yeah, so, like a square wave.
If you in a synthesizer, right,a square wave is the
fundamental, and then oddharmonics at a specified kind of
(20:11):
ratio.
That's what a square wave is.
Same thing.
If you start filtering outthose harmonics, it just slowly
turns into a sine wave as youget down to the fundamental.
Yeah, so.
So this is one of these things.
If I were better prepared andthought we were going there, I
would have images for this too,but it's a little complicated,
for sure, and it's a little hardto just wrap your head around
(20:33):
and envision.
But they're really two sides ofthe same coin.
As you start changing thatwaveform shape and doing
clipping or limiting, limiting,does this too?
Limiting will add highfrequency content.
Yeah, any compression,saturation, even any cue that
models saturation, right, allthese things will add these
(20:56):
higher harmonics to whatever.
And so if you're feeding in asignal that has frequencies up
to 20 kilohertz, it's going tostart generating stuff up to 40,
at least, if not more, and soyou need a strategy to deal with
that, otherwise you getaliasing.
Marc Matthews (21:14):
Yeah, interesting .
I think what we'll have to dois I say this every time we chat
.
We'll have to do anotherepisode whereby I prepare,
because I again I didn't realizeit's just when you say these
things I'm like, oh, I'm gonnadive deeper into that and get to
the uh, the nuts and bolts ofit myself.
yeah, I think, yeah, if we pullit back, because maybe we'll
come back, but reallyinteresting stuff.
(21:36):
And it, the, the oversampling,because I that you see that now
more and more in plug-ins, inparticular when you mentioned
the the clipper there and we gotour standard clip myself and
you got the oversampling featurein it for that reason, so
really interesting.
You mentioned there aboutaliasing.
Yes, maybe, if we sort of likechange tact a bit and then just
have a quick overview ofaliasing and basically what that
(21:59):
is.
Ian Stewart (22:00):
Yeah.
So I'm going to make up somenumbers just to try and make
this a little bit easier, if Ican.
Let's say that we're playing a10 kilohertz.
Uh, no, sorry, I'm thinkingthrough this in real time.
You're seeing my seeing thebrain, the gears turning here.
Let's start with sample rate.
(22:22):
Let's say our sample rate is 20kilohertz low.
Right, that's lower than usual.
That means that our nyquistfrequency, the maximum frequency
that we can capture, is whatwould you say it was 20?
kilohertz.
If, if our sample rate is 20kilohertz, half of that, 10
(22:42):
kilohertz, 10k, right, so 10k isthe maximum frequency that we
can capture.
So what aliasing is?
Aliasing is sort of the theeasy ish not easy, but easy ish
way to think of it.
Visually is almost like amirror for frequencies and it's
almost like you put this mirrorat the, the Nyquist frequency,
(23:03):
so half of your sample rate.
Okay, so here our our mirror isat 10 kilohertz, so if we do a
sine sweep, starting down at 20,that ramps up to 20 kilohertz,
right.
So it goes past our Nyquistfrequency, past that halfway
point.
When it hits 10 kilohertz, whenit hits that mirror, it bounces
(23:26):
back down into the audiblerange, down below 10K.
So, up to 10K, everything's asexpected.
When the frequency, once itgoes above 10K, when it gets to
11 kilohertz at the input, thedigital output is nine kilohertz
, so the input has gone onekilohertz above the maximum.
(23:48):
The output ends up 1 kilohertzbelow.
Right, right.
So you start to hear this tonesweeping back down Rather than
continuing on up to 20k.
You start to hear it sweepingback down in the opposite
direction.
Yeah, okay, right.
So if the input gets up to 15know that's 5k above the input or
the above nyquist the outputends up 5k below.
(24:10):
So now the output is back downat 5 kilohertz right, I got you
and as it gets up to 20kilohertz the output goes right
down to dc right.
It goes down through 1K, 500,120.
It turns into this really lowrumble.
So aliasing creates theseinharmonic right.
(24:33):
There's no musical relationship.
The relationship of the inputto the output is a mathematical
relationship between the inputand the Nyquist frequency which
has.
It's not harmonic.
It has nothing to do with themusical timbre or quality of it,
and so it's a very kind of.
It's a very digital type ofdistortion and it just sounds
(24:55):
weird and hashy and usually,since the frequencies are much
higher, it just ends up soundingbrittle.
It kind of adds some high end,but in a very brittle,
unmusically related way.
Yeah yeah, so that's aliasing.
And while we're talking aboutthat, I should just mention
there's also something calledimaging, which is basically the
(25:15):
opposite at the output.
And so, just like we have tofilter the input to make sure
that frequencies above ourNyquist frequency don't make it
in at the output of a converter,we're basically sending out
these pulses every 44,100 timesa second or 48,000 times a
second or whatever, at thespecified value, amplitude,
(25:38):
voltage, whatever and if youdon't then filter those, they
actually create a whole bunch ofextra frequencies above and
below.
So you've got to filter theoutput too, in the same way that
you filter the input.
Same frequency, same roll-off,same phase response.
(25:58):
So you really you're hearing twofilters, right?
You're hearing whatever, thefilter that was used to record
the sound, and then you'rehearing the filter that's being
used to play it back.
And so aliasing and imagingvery similar Again, kind of two
sides of the same coin.
One happens at the input whenyou're going from analog to
digital, and the other happensat the output when you're going
(26:19):
from digital back to analog.
Marc Matthews (26:21):
Yeah, so just to clarify that.
So when we go in aliasing andoutput imaging, imaging, yeah.
So we used the analogy there ofthe 20 kilo sample rate.
Half of that, 10 kilohertz, 10K, 11K then results in Bounces
(26:43):
back to 9K Bounces back to 9K,yeah, and then so forth.
So we go all the way up to 20.
That bounces back to zero.
that's when we get the lowrumble yeah and then when we
have the, the imaging of theoutput, it's another filter on
there.
So we've got two filters, asyou say, that filter going in,
filter going out as well.
Yeah, yeah, it's reallyinteresting stuff, like you say.
I said right at the beginning,it's not we just look at sample
(27:04):
rates but and you just think,well, I'll just go for 44.1 or
go to 48, whatever it may be.
I mean, it's not often that wereally dive deep into okay, well
, why, and it's really why is hedoing that?
Ian Stewart (27:15):
I think it's worth saying here too.
A lot of people hear about thisand all of a sudden, they start
thinking, oh my God, I have toset some high-pass or low-pass
filters.
So I do want to pause andreinforce.
This is not something all thisfiltering that we're talking
about.
It's not something that youneed to take care of.
Yeah, yeah, of course, if yourgear is properly designed which,
to be totally honest, in 2025,even the most kind of consumer
(27:38):
stuff is right.
Yeah, we've gotten pretty goodat this kind of stuff that you
know, even very cheap AD and ADCand DAC chips can do a very
good job of this.
Now, it's not something thatyou need to worry about.
It's just understanding what'shappening in the background and
why the sample rates are whatthey are in relation to the
frequency content that we cancapture.
Marc Matthews (28:00):
Really really interesting stuff.
I realize we're coming towardshalf an hour now with regards to
this this always happens withthe, because it's probably
because I go off on tangentswhen we were talking about the
linear phase and et ceteraearlier.
Yeah, but I think the finalquestion, really, I think for
the audience listening withregards to sample rate and you
touched on it earlier when Iasked you the question about
(28:20):
what sample rate would youadvise?
And we went through 48kilohertz, because the majority
of platforms now are requesting48 kilohertz.
You mentioned YouTube, forexample, so would that be your
advice for sort of streaming aswell?
I don't know how muchexperience you have with
actually uploading it to sort oflike a digital distributor,
(28:41):
something like DistroKid Othersare available.
In that scenario, would youalso recommend 48 if they allow
it?
Ian Stewart (28:51):
yeah.
So usually what I recommendthese days uh, really, for my
clients, right, for my masteringclients I say, um, look, if
they send me mixes at 96k, I'mgonna master at 96k and I will
give them back 96k 24-bit filesas their kind of high res master
and and basically what Irecommend is that they upload
(29:12):
their high res master to distro,kid tune, core, cd, baby,
whatever.
Um, most services have alossless option these days,
right?
Uh, apple music has a losslesstier title, amazon HD, co-buzz,
a lot of them do, so people aregoing to be able to listen at
(29:41):
2496 or whatever in certainscenarios.
The other thing I didn'tmention earlier is that pretty
much any Bluetooth headphonesAirPods, airpod Pros, airpod
Maxes, even other ones 48K is avery standard sample rate for
Bluetooth devices.
I didn't know that.
And when I say standard, youcan't switch it to anything else
.
That's what they run at.
So even if you're listening tosomething higher or lower, it's
(30:02):
going to get converted to thaton playback.
So there's kind of another goodreason to deliver at 48K, right
?
One less conversion step.
But I generally advise myclients to upload 24-bit at
whatever sample rate they sentme their mixes at, which is what
(30:22):
I'm going to master at and kindof conform everything to that
master at and kind of conformeverything to that Um, because
for the few platforms that don'tsupport that Spotify, who for
five years has been saying thata lossless, at least CD quality
tier is coming People keeppromising oh no, it's happening
(30:42):
soon.
I will believe it when I see it.
Um, uh, but my, my feeling isis kind of that if someone is
content to listen on Spotify, orthey don't have the high res
version of whatever service,they're not listening to music
because they want the ultimatein fidelity, they just want some
good music to listen to.
(31:03):
And they probably aren't goingto notice that it's streaming at
some lossy codec, lower bitrate.
Anyway.
There's plenty of other thingsthat are happening to change the
signal a little bit and changethe imaging and high frequency
stuff.
So the sample rate conversionis kind of the least of the
worries at that point.
And then if you do upload your96K 24-bit masters for the
(31:24):
people that have the setup toappreciate it and want to,
they're getting the absolutebest experience that they can.
That said, I think 48k is great.
It kind of shallows out thefilter enough that it sounds a
little bit more open.
You get a little just kind ofmore natural top end and we
(31:44):
can't go there today.
This will have to wait foranother one.
There are other trade-offsnegative trade-offs as you start
to increase your sample rate,certainly above 96 kilohertz, as
you go up to 176 and 192 and384 and crazy shenanigans like
that.
There are actually othernegative trade-offs that most
(32:06):
people in the audiophilecommunity will gladly tell you
you're wrong about and why theythink it's so much better.
Marc Matthews (32:12):
But there there's good science to show that
actually it can create othernasty artifacts yeah, I think
that would be a fantastic uhdiscussion, because it's a very,
very thought conversation,provoking discussion.
I think the negative trade-offof 96k and above, as you say.
I think we'll save that foranother episode and hopefully
(32:32):
get some examples and some otherbits and pieces in there as
well, because I think that wouldbe super interesting.
But touching on superinteresting, I didn't realize
Bluetooth was 48 kilohertz.
I would have gone with theassumption that it was 44.1.
Ian Stewart (32:45):
And it's weird because it's already a lossy
codec, although it's gettingbetter, like aptX, and there are
a few different iterations onBluetooth that have really made
it sound a lot better than itused to.
Yeah, I don't know if I don't.
There may be Bluetooth chipsthat can run at other sample
rates, but usually, for whateverthe device is, it's fixed.
You can't change it.
(33:06):
I know pretty much all ofApple's stuff is 48K.
My older Sony, you knowwhatever Sony model number.
They're really got to loveSony's model numbers Super
memorable.
I can't remember what it is,some string of whatever.
Those were 48K, you couldn'tchange it, so it's pretty
standard and there may be someexceptions, but again it's going
(33:30):
to be set at 44, one orwhatever it is.
Um, so yeah, I really I think 48K is a great deliverable.
If you want to record at 88,two or 96, there's no harm in
that it.
It may make something sound alittle better while you're
mixing right, it gives you alittle more room for those
overtones and the oversampling.
But as a deliverable, 48k isgreat.
(33:55):
And I also should just sayquickly that sample rate
conversion has gottenunbelievably good, like really
really, really good.
And so if you have a masteringengineer who has the right tools
, which anyone with rx, can dothis, um, uh, wave labs,
built-in sample rate conversionis also excellent.
(34:16):
Um, other dawes not so much,but the ones that pay attention
and get it right it can bealmost completely invisible yeah
, so you kind of.
You know, if you need to getdown to 48 K, you can do it, and
if recording at 88, two or 96,you feel like you get more out
of the recording and yourplugins and some things just
work better.
They're great, you know, andand you can easily, at the end
(34:38):
of the mastering stage, comedown to 48, if you want that as
your deliverable.
Marc Matthews (34:42):
So yeah, essentially the options there
isn't it?
Ian Stewart (34:44):
now with the technology.
Marc Matthews (34:45):
we have now the option's there, isn't it?
Ian Stewart (34:48):
Yeah, With the technology we have now.
The option is there, yeah, butI would say 44.1, it's not.
I'm not going to say it's dead,I'm not ready to declare that
yeah, but it's.
There are fewer and fewer caseswhere you really need that
right.
Pretty much the only case thesedays is if you are actually
making CDs, then you've got toget it down to 44.1.
Marc Matthews (35:04):
Yeah, and I suppose as well, while we're
talking about technology, wealso have larger the
accessibility of storage as well, because obviously it's going
to, but in doing having theselarger sample rates is going to
take up more storage.
Ultimately, yeah, now we're ina position whereby we can have
more and more storage.
Ian Stewart (35:22):
Yeah, if you've got it, yeah, why not use it and go
up to that for you.
You can go get a 5-terabytehard drive for $100 or something
, right, yeah?
Marc Matthews (35:32):
yeah.
Ian Stewart (35:33):
These days, why not ?
It's wild.
Marc Matthews (35:35):
It's mad.
I was having a bit of a tangent.
I was discussing the other dayand I used to work for Staples I
think they're still going inthe States.
I don't think it is in the ukanymore, I don't think still
over and I remember selling harddrives.
This was about 20 years ago.
I'm selling the lamb and you wewere selling like memories, and
it was like 128 megabits, yeah,and now, and I, I think I found
(35:57):
one the other day and I waslike man, I remember when people
come into the store and theywould buy like 32 meg, like um
sd cards and stuff, and I waslike, yes, crazy man, crazy yeah
, I had a boss a long time agowho had a story he loved to tell
about his first hard drive thatwas 600 megs, basically a CD's
worth of audio.
Ian Stewart (36:15):
Yeah, that cost him several thousand dollars.
It's the complete oppositerelationship now.
Marc Matthews (36:23):
Yeah, I was watching something the other day
and it was a film and it wasthe early 90s and they
referenced a solid state and Iwas thinking to myself were
solid states I'll get correctedon this were solid states around
in the early 90s?
They may well have been, but asolid state in the early 90s?
Ian Stewart (36:40):
if they were, that would have Insane amount of
money Very esoteric, I think thetechnology and the idea for it
was probably there, but it wasprobably large and very
expensive and difficult to make.
Marc Matthews (36:52):
Yeah, yeah, it was one of those ones where I
was watching the film and Icalled it out and I was like
that's not accurate, being oneof those people.
That's clearly not accuratethat teenager's talking about
having an SSD.
There's no way they couldafford that in the early 90s.
Xst.
Ian Stewart (37:05):
There's no way they could afford that in the early
90s.
Marc Matthews (37:07):
There we go, Ian.
Thank you so much for this thishas been brilliant.
And I appreciate it.
I put you on the spot with afew questions, so I appreciate
your willingness to go intothose subjects.
So I know the audience will getloads out of this, in
particular, because sample rateis one of those ones.
You see questions all the timeonline.
You see discussions and someonesays one thing, someone says
(37:33):
another.
But it's always nice to whenyou're searching for those
questions and someone says, doit at this reason, do it at that
rate, and then explains andexplains why, and this is this
is what this episode is going togive.
We're going to give anexplanation as to why you can
make, and then you can make aninformed choice.
That's the sample rate that youwant to use.
So really, really good stuff.
Before we wrap up and I go intowhat we're going to be going
through in part two, is thereanything you want to share with
(37:54):
the audience?
Alternatively, where can they?
I say alternatively, inaddition, where can they?
Find you if they want to learnmore.
Ian Stewart (38:01):
Yeah, so my website is flotownmasteringcom.
F-l-o-t-o-w-n.
I've got links there.
There's a page that has linksto all the podcasts I've done,
so the past episodes I've donewith you Mark and some other
shows I've been on.
If you want to hear me yammer onabout other things, if you want
to get in touch about mastering, there's a page with kind of a
(38:24):
little about and some of thetools that I like to use, which
you can see some of if you'rewatching the video, and an
intake form where you can justsubmit all your stuff.
Yeah, ian Stewart Music onInstagram.
I don't post a ton, but youknow, there's like pictures of
the dog who left, like if we gohiking, you'll see pictures of
me and the dog or sometimes apost-release stuff.
(38:45):
Yeah, I write for the iope blog.
If you, if you go look formastering articles there, you'll
find some of my stuff.
I just did some videos for wavelab.
It's there.
It's wave labs 30th anniversary, so it's a great time to grab
wave lab if you've been thinkingabout that.
Um, if you are in need of amastering dog certainly not for
everyone yeah, I.
Marc Matthews (39:05):
I don't know.
Ian Stewart (39:06):
But really honestly , mastering is the main thing.
I love mastering, I lovetalking about it and teaching it
.
So if you've got some music youwant to make awesome, I would
love to do that together.
Marc Matthews (39:17):
Amazing stuff Links in the episode description
for the audience listening.
Please do go and check that outand just follow Ian as well.
So whilst we are still in thisepisode, don't forget part two
is going to be the next episode,so it's going to be episode 204
.
So if you're listening to thiswhen this episode drops, you've
(39:38):
got a bit of time, but if you'relistening to this way forward
in the future, make sure youcheck out episode 204, where
we're going to be diving.
I can't get my notes here headfirst.
Why put head first?
Why not?
Yeah, well, head first, I guessnot literally into another
mysterious but crucial topic,dither One.
I don't think I've evermentioned dither on the podcast
(39:59):
before.
No sense.
Ian Stewart (40:00):
Look, if you're going to get involved with
dither, no sense in dipping yourtoes in Just head first.
Marc Matthews (40:05):
Yeah, just go all in All in there, no sense in
dipping your toes in just headfirst.
Yeah, just go all in all inhold your breath yeah fantastic
and uh audience remember as well.
I mentioned it earlier.
If you want to get weekly tools, tips etc and just this week I
included uh ian's wave labs, uhfeatures on youtube as well, so
an example of what you mightfind in that newsletter click
the link in the episodedescription and get those free
(40:27):
weekly tips direct to your inbox.
And until next time, stayinspired, keep creating and
don't be afraid to experimentinside the mix.
Does sample rate actually make your music sound
better?
You've probably heard argumentsover whether you should record
at 44.1, 48 or even 96, buthere's a real question does it
really matter?
In this episode, we're divingdeep with someone who not only
understands the theory, butlives it every day in the
mastering room.
Ian Stewart (00:20):
You're listening to the Inside the Mix podcast with
your host, mark Matthews.
Marc Matthews (00:23):
Welcome to Inside the Mix podcast with your host,
mark Matthews.
Welcome to Inside the Mix, yourgo-to podcast for music
creation and production.
Whether you're crafting yourfirst track or refining your
mixing skills, join me each weekfor expert interviews,
practical tutorials and insightsto help you level up your music
and smash it in the musicindustry.
(00:43):
Let's dive in.
Hello folks, and welcome toInside the Mix.
Today's episode is part one of aspecial two-part conversation.
This might be the first time.
No, I tell you what it's thesecond time I've done a
two-parter in all 200-plusepisodes now.
So we're with mastering,engineer, educator and all-round
audio wizard, ian Stewart.
(01:04):
Ian is a returning guest.
Check out the listener.
Favorite episode, episode 165what is midside eq?
One of the most popularepisodes of the podcast, might I
add.
He's the founder of flow townmastering and an assistant
professor at berkeley college ofmusic, where he teaches the
science and art of audiomastering.
His resume includes work withlegends like krs1 and he's a
(01:26):
regular contributor to iZotope'sLearn blog and Wave Labs as
well, recently on YouTube.
In this episode we are talkingall things sample rate, what is
it, what it isn't, and why.
Getting it right could make orbreak your project.
That's enough from me, ian, howare you?
And welcome back.
Ian Stewart (01:43):
I am excellent, Mark.
Thank you so much.
I'm excited to be here.
I am in lovely New England,where today it is a balmy
90-something degrees, which forthe rest of the world is
30-some-odd.
I don't know.
So, it's sticky up here.
I've got the windows open.
If you hear a little extraneousnoise, that's probably from my
(02:04):
end.
So I I do apologize for any ofthat, but I am great, it's
summer, we're doing it, life'sgood working on cool music, so
yeah love it.
Marc Matthews (02:12):
Yeah, that's the attitude man.
I find, like when it I thinkthe audience, when they listen
to like the back catalog of thepodcast, they could tell the
seasons, because there's certainlike episodes or series of
episodes and then you just hearlike animals, like birds,
seagulls, all this stuff, so youcan kind of like the podcast
when you're listening to it.
You move with the seasons, whichis quite cool yeah I appreciate
(02:35):
it, but I know it's been hothere in the uk, as we were
saying off episodes, so it's uhhot, hot time, especially when
you're in a studio as well, likeit's, and you haven't got that
properly that the hvac going andstuff yeah, we're not.
Ian Stewart (02:47):
We're not quite it was.
It was, like you know, cool andrainy until a week ago and then
, all of a sudden, bam summer.
Yeah, so yeah, it's caught me alittle off guard yeah, yeah, it
always doesn't.
Marc Matthews (03:00):
It always catches us off guard, to be honest.
Yeah, it doesn't take much foreveryone to start getting the
barbecues out and then theshorts and t-shirts, even though
it's probably.
I mean, it has been hot here,but sometimes it's in the late
teens and you still people arelike I don't think it's that hot
to be like that.
But there we go, so in thisepisode, folks, we're peeling
back the layers of confusionaround sample rates.
This is kind of almost like aback to basics, really.
(03:21):
From 44.1 to 96 and everythingin between, Ian's going to help
us separate technical truth fromaudio folklore.
Think of this as your go-toguide for demystifying sample
rates.
But before we do that, if youwant weekly tools and tips to
make your music playlist worthy,click the link in the episode
description and get my freeweekly tips direct to your inbox
(03:41):
as well.
No spam, no fluff.
Just once a week I'll send youweekly tips and tricks, not just
for me, but from alsoindividuals that have been on
the podcast and stuff that Ifind out about as well.
So enough for me.
Going back to sample rate, Ithink to start our discussion,
it's probably quite importantjust to outline what sample rate
actually is.
What are we referring to inaudio?
Ian Stewart (04:01):
Yeah, so when we record digital audio, right,
when we want to get somethingfrom a microphone for those of
you that are just listening, I'mpointing to my microphone and
we want to get it into a file,we have to do a bunch of
conversion.
And part of the process is, youknow if back in the old days,
when we use tape or othermediums, right, that it was just
a voltage that came out of themic, went through a preamp and
(04:23):
that voltage would get stored,mad magnetically on a tape.
But when we go digital, we haveto measure that voltage and
store it as a value.
And because we can't do thatinfinitely fast and continuously
, we kind of take a sample, we,we measure that voltage a bunch
(04:44):
of times every second and savethat.
Then we say, okay, the voltageis this, and we save it as a
value, as a digital number, andwe move on to the next sample
and say, oh, now the voltage isthis and we save that, and so
the sample rate just tells youhow many times every second
you're taking that sample.
So for 44.1 kilohertz, that's44,100 times every second.
(05:07):
We're measuring the audio.
So kilohertz is thousands ofhertz.
A hertz is just a measure ofcycles per second.
So it's just how many thousandsof times every second?
Really, we're measuring thevoltage of the audio and storing
it as a digital number, andthat's pretty much the root of
the audio.
And storing it as a digitalnumber, and that's pretty much
the root of it.
Marc Matthews (05:26):
Yeah, yeah, yeah, definitely.
So whilst we're on that, justunpacking the 44.1.
So if we were to dive into sortof the frequency that we can
hear from 20, let's say, 20hertz to 20,000 hertz, can you
maybe just explain to theaudience why it is that we're
using the number 44.1?
Ian Stewart (05:46):
explain to the audience why it is that we're
using the number 44.1?
Yeah, absolutely so.
Um, if, if you've ever heard ofthe shannon sampling theorem or
the nyquist sampling theorem orthe shannon nyquist sampling
theorem or anything, any kind ofcombination?
Uh, these were two engineersback in the oh, I should know a
time range, but I I want to saymaybe even the 30s, 1930s, I
think Nyquist certainly.
I should have been moreprepared for this.
(06:07):
I think Nyquist worked for BellLabs, but he kind of pioneered
this digital sampling theory waybefore we even had the
technology to do it.
And what he computedmathematically is that if we
want to store any givenfrequency say it's 100 hertz,
pretty low, right.
But if we want to store that,we need two sample points in
(06:29):
that waveform, right.
So if it goes from zero up toits peak, back down to zero,
down to negative one, itsnegative peak and back.
Basically, the idea is we atleast need to store the maximum
and minimum, those positive andnegative peaks, to be able to
accurately describe what thatfrequency is.
So you need two samples forevery period of a waveform.
(06:52):
So commonly the range of humanhearing is given as 20 hertz to
20,000 hertz, 20 kilohertz and,to be honest, that's pretty
generous Really.
Once you get out of your teens,maybe even a little younger,
20k is probably gone and notcoming back, but we kind of take
(07:12):
that as the standard acceptedrange.
And so when digital audio wasreally being developed by Sony
and Philips back in the 70s or80s, they were trying to figure
out what's the ideal sample ratethat we need to capture audio,
and they knew they wanted to putthe data on a disc, right, an
(07:34):
optical disc, cds.
And so they're trying to playthis balance of having good
enough quality that they cancapture everything that human
range of hearing can appreciate,but not storing more data than
is really necessary, right?
They don't want to, becausemore data means you can fit less
music on the disc ultimately.
So they said, okay, well, if wecan hear up to 20 kilohertz,
(07:59):
that means our sampling rateneeds to be at least 40
kilohertz, right?
So if we need two samples forevery period, it's got to be
double.
Whatever the frequency you wantto capture is Sample rate needs
to be double that top frequencyyou want to capture.
So if we're trying to capture upto 20 kilohertz, we need at
least 40 kilohertz sampling rateand at that stage, one of the
(08:23):
things that you have to do isyou have to remove frequencies
above that.
So if you say I'm going tosample at 40 kilohertz, there is
a requirement that you not letany frequencies above 20
kilohertz into the digitalconversion chain.
And if you do, that results insomething called aliasing, and
maybe we can get to that alittle bit later if you want.
(08:45):
If you've heard of aliasing,that's kind of where this idea
comes from.
And so you have to put a reallysteep low-pass filter on your
input audio before it hits theconverter.
All right, stuart, can we talkabout digital audio please?
So back in the 70s and 80s thesefilters like to make.
(09:08):
A really steep filter was tough.
You couldn't.
There are limits to doing itdigitally.
Doing a really steep analogfilter was tough.
So basically they said okay,you know what, we'll have the
filter start rolling out at 20kilohertz, but we need a little
bandwidth for it to get thesignal low enough that it
doesn't create aliasing.
So they add a little bit more.
(09:29):
They go up to 22.05, right, twoand a half kilohertz, or two
and a whatever 2050 kilohertz ofextra bandwidth to allow the
filter to do its job.
Now you double that extrabandwidth to allow the filter to
do its job.
Now you double that, you get44,100.
So that's where that 44.1Ksample rate comes from, is
(09:53):
saying we want the 20 kilohertzbandwidth of audible frequency,
then we need a little extra roomfor the filter to knock out
stuff above it, and then that'swhat our sample rate's going to
be.
Marc Matthews (09:59):
Love it.
This is great because it takesme back to many, many, many
years ago now, when I was doingmy studies and we did a whole
thing on um, sampling andsynthesis and it was, yeah, we
sort of dive, dive deep intothis and hands, and, yeah, a lot
of it I have forgotten becauseit's not something that, oh, I
mean, we play, we, we look atsample rates etc and we're
setting sample rates, but beyondthat I don't really dive too
(10:22):
deep into it.
So this is a.
This is a great refresher for me, yeah you, you usually don't
have to think about it too no,no, exactly, yeah, yes so, um,
but you say, when you say thingslike aliasing nicholas, like oh
yeah, yeah, that does ring abell.
now I remember going throughthis and what you described
there about how, uh, when we'retaking those two, two samples
and then we're doubling it andthen you've got that, basically
(10:43):
that buffer, when we're going upto 44.1.
But it kind of leads me onnicely to my next question,
which is when we start to enterthe realm of 48, 96 and beyond
and everything in between, whatwould you say is the standard
sort of sample rate for musicproduction?
If someone were to ask you okay, well, I'm recording a band,
I'm recording a band, I'mrecording an artist, whatever,
(11:04):
I'm just doing a recording.
Yeah, what would your advice be?
Ian Stewart (11:08):
with regards to sample rate?
That's an excellent question.
I think not that many years agothe answer might have been a
little trickier.
Actually, definitely when CDsdominated.
44.1, you knew you were goingto have to end up there at some
point.
44.1, you knew you were goingto have to end up there at some
point, and sample rateconversion technology was not as
good, so there was maybe anargument to be made for
(11:30):
recording at 44.1 some years ago.
These days, though, I thinkit's pretty clear cut that we
could say a really good standardis 48k.
Um, for a number of reasons.
One, we know atmos is stillbeing pushed, and I think it's
still.
For me, the jury's still out onon how long-lived atmos is
(11:54):
going to be as a format and ifeveryone's really going to start
doing it or whatever.
That's topic for another dayyeah, an interesting one yeah,
but, but Atmos can be at 96k,but it's kind of a pain in the
butt, yeah, and so practicallyAtmos mixes need to be at 48k,
like it's one or the other, andreally the vast, vast majority
(12:16):
of them are at 48k.
Youtube recently updated theiraudio recommendations For the
longest time.
It was funny.
They recommended 44.1, but theyrecently updated their
recommendations.
Recommendations for the longesttime is funny.
They recommended 44 one, butthey recently updated their
recommendations to 48k.
48k has been the standard foraudio that accompanies video for
pretty much ever.
Marc Matthews (12:33):
I was just going to mention that.
Yeah, is that with youtube?
Now, is that for youtube musicwhere they're recommended for?
Ian Stewart (12:38):
yes, for, for all of youtube not just youtube
music, for all of it and eveneven for the main video site.
For a long time theyrecommended 44.1 because they
encode it to other codecs orwhatever, right, but now they're
saying 48.
So there really has been astandardization in the last
(12:59):
handful of years of almost alldeliverables moving to 48k as a
minimum.
Yeah, so I would say 48k is agreat sample rate to record at,
to work at, certainly to deliverat.
I still, as a masteringengineer, I still give people
1644 files for CDs in case theywant to do that, but the main
(13:22):
deliverable I send out is 24-bit48K, yeah.
So, yeah, I think there's abenefit there, just in terms of
standardization and not havingto go through extra rounds of
sample rate conversion.
And then the other thing.
So I think the natural nextquestion is well, if 44.1 is all
(13:42):
we need to capture our range ofhearing, and really a little
bit more, what's the benefit ofgoing up higher than that?
Uh, and there there are a few,and if I'm jumping the gun here,
feel free to bring me in, but Iwas going to say it was going
to be.
Marc Matthews (13:55):
My next line of questioning was with regards to
so we're getting in sync.
Ian Stewart (13:59):
we're starting to know each other, mar, mark,
Please carry on.
So, yeah, the nice thing about48K is that basically it allows
that filter that rolls off thehigh end to be a little gentler,
and while I'm not aparticularly woo-woo kind of guy
(14:24):
, give me facts and good, hardevidence, or it just doesn't
really matter.
I you know and this thisextends to many areas of my life
in audio, though, there's nogood evidence that audio content
above 20 kilohertz makes anydifference to anyone.
Uh, there's never been a testthat has conclusively shown that
there are some people who canhear to 25 kilohertz and it
changes Like it's.
There's just no data on that.
Yeah, it doesn't mean thatthose people don't exist.
(14:45):
It just means, in all thestudies that have been done,
we've never found one right so,starting with that as the
baseline, we know that, however,what filters do is they change
the phase response.
Um, some of these filters canbe linear phase.
A lot of them are not.
They are minimum, becauselinear phase filters add latency
(15:06):
, uh, so a lot of times we'retrying to get fast performance,
so there are minimum phasefilters that that do this, roll
off and those do have a phaseshift that's associated with
them, and a phase shift is notexactly the time the same as a
time shift, but it's.
You can kind of think of itlike that.
And one thing that we do haveevidence to say is that human
(15:30):
ears are unbelievably sensitiveto timing, like like crazy and
there's a good evolutionaryprecedent for this that if
there's a tiger that's half ameter closer to you, you want to
hear that one first and payattention and get away from it,
and not the one that's a littlefurther.
So we're really sensitive totiming and so these phase shifts
.
A lot of times people will heardifferences in different sample
(15:53):
rates and I don't know that wecan say this conclusively, but I
think there's enough strongevidence that points at this as
the conclusion that really whatyou're hearing is the phase
response of those filters.
And by going to 48k or 88.2 or96, you allow that filter to be
a little gentler and notactually get in the way and
(16:13):
impact the stuff that you canhear.
The frequencies you can hear,which may be down at even 16, 18
kilohertz.
That phase shift may stillimpact them a little bit.
So that's one of the benefits,right, it can sound a little
more natural.
So nonlinear processing issomething that's dependent on
the input level, right?
So like a compressor or alimiter or a clipper or a
(16:35):
saturator and all of thesethings.
One of the characteristics thatthey have in common is that
they create overtones.
So if you feed them a onekilohertz tone, depending on the
type of processing, they maycreate overtones at 2k and 3k
and 4k and 5k, or, it may justbe odd, it may be three and five
and seven and nine and so on.
(16:56):
So when you feed a nonlinearprocessor a very high frequency,
there's a chance that it startscreating content that is above
our Nyquist frequency, and theNyquist frequency is that half
the sampling rate.
So if we're at 48K, the numbersare a little easier.
So let's just say, for 48K theNyquist frequency is 24
(17:18):
kilohertz as the highestfrequency that we can capture,
and so if we start generatingcontent that's over that, now
that breaks that band limitingrequirement.
Remember, I said earlier thatwe have to have this filter
because we can't allow stuffinto digital audio that's above
that half the sampling frequencyrate.
So now we have to deal withthat, and you can either not
(17:44):
deal with it, which is what someolder plugins that was kind of
the naive way when we didn'tknow better you didn't deal with
it and it turns into aliasing.
Yeah, the other common way todeal with it is to oversample
the plugin.
So you internally switch to ahigher sample rate in the plugin
, you allow for that higherfrequency content, then you
(18:10):
filter it back out and then youkind of re-inject that into your
main audio stream.
Um, and so there's atheoretically there's an
argument that if you work at 96k, uh, you have even you kind of
have to worry about that alittle less yeah, so.
Marc Matthews (18:19):
So with regards to the overtones that being
created by non-linear processing, but how?
How is that being created?
How are the what?
How are the overtones that arebeing created by nonlinear
processing?
How is that being created?
How are those overtones beingcreated?
Ian Stewart (18:29):
It's basically what's happening is the shape of
the waveform is changing.
So let's think clipping.
Clipping is kind of the mostextreme example and also creates
a lot of overtones.
So say, if you clip a sine wave, right, you've just got your
regular sine wave, and then youpush it up into clipping, the
top of it gets just flattenedoff.
Marc Matthews (18:50):
Yeah.
Ian Stewart (18:50):
And you can.
There are kind of two commonways.
Marc Matthews (18:54):
We're getting a little deep and again call me
back at any point, if you need,but I'll go for it Two common
ways that we think about audio.
Ian Stewart (19:02):
We talk about time domain, we can talk about
frequency domain and they'reinterrelated and you can't
really disentangle them.
So the time domain version of aclipped sine wave is that it
goes up and then it flattens offat the top and then it goes
down, it flattens off at thebottom.
The frequency domain version,if we think about what happens
to the frequency content of thatsine wave, in order to flatten
(19:27):
the top off, you basically haveto add higher frequencies with
different phase relationshipsback into that sine wave.
That is mathematically what'shappening, right, and so
clipping or any other sort ofwave shaping where you you're
really changing the amplitudeshape of of the waveform, it
(19:50):
changes the shape of it.
The change of the shape makesthese extra harmonics and
frequencies pop out.
If you try and take thosefrequencies out, the shape goes
back to what it originally was.
Yeah, so, like a square wave.
If you in a synthesizer, right,a square wave is the
fundamental, and then oddharmonics at a specified kind of
(20:11):
ratio.
That's what a square wave is.
Same thing.
If you start filtering outthose harmonics, it just slowly
turns into a sine wave as youget down to the fundamental.
Yeah, so.
So this is one of these things.
If I were better prepared andthought we were going there, I
would have images for this too,but it's a little complicated,
for sure, and it's a little hardto just wrap your head around
(20:33):
and envision.
But they're really two sides ofthe same coin.
As you start changing thatwaveform shape and doing
clipping or limiting, limiting,does this too?
Limiting will add highfrequency content.
Yeah, any compression,saturation, even any cue that
models saturation, right, allthese things will add these
(20:56):
higher harmonics to whatever.
And so if you're feeding in asignal that has frequencies up
to 20 kilohertz, it's going tostart generating stuff up to 40,
at least, if not more, and soyou need a strategy to deal with
that, otherwise you getaliasing.
Marc Matthews (21:14):
Yeah, interesting .
I think what we'll have to dois I say this every time we chat
.
We'll have to do anotherepisode whereby I prepare,
because I again I didn't realizeit's just when you say these
things I'm like, oh, I'm gonnadive deeper into that and get to
the uh, the nuts and bolts ofit myself.
yeah, I think, yeah, if we pullit back, because maybe we'll
come back, but reallyinteresting stuff.
(21:36):
And it, the, the oversampling,because I that you see that now
more and more in plug-ins, inparticular when you mentioned
the the clipper there and we gotour standard clip myself and
you got the oversampling featurein it for that reason, so
really interesting.
You mentioned there aboutaliasing.
Yes, maybe, if we sort of likechange tact a bit and then just
have a quick overview ofaliasing and basically what that
(21:59):
is.
Ian Stewart (22:00):
Yeah.
So I'm going to make up somenumbers just to try and make
this a little bit easier, if Ican.
Let's say that we're playing a10 kilohertz.
Uh, no, sorry, I'm thinkingthrough this in real time.
You're seeing my seeing thebrain, the gears turning here.
Let's start with sample rate.
(22:22):
Let's say our sample rate is 20kilohertz low.
Right, that's lower than usual.
That means that our nyquistfrequency, the maximum frequency
that we can capture, is whatwould you say it was 20?
kilohertz.
If, if our sample rate is 20kilohertz, half of that, 10
(22:42):
kilohertz, 10k, right, so 10k isthe maximum frequency that we
can capture.
So what aliasing is?
Aliasing is sort of the theeasy ish not easy, but easy ish
way to think of it.
Visually is almost like amirror for frequencies and it's
almost like you put this mirrorat the, the Nyquist frequency,
(23:03):
so half of your sample rate.
Okay, so here our our mirror isat 10 kilohertz, so if we do a
sine sweep, starting down at 20,that ramps up to 20 kilohertz,
right.
So it goes past our Nyquistfrequency, past that halfway
point.
When it hits 10 kilohertz, whenit hits that mirror, it bounces
(23:26):
back down into the audiblerange, down below 10K.
So, up to 10K, everything's asexpected.
When the frequency, once itgoes above 10K, when it gets to
11 kilohertz at the input, thedigital output is nine kilohertz
, so the input has gone onekilohertz above the maximum.
(23:48):
The output ends up 1 kilohertzbelow.
Right, right.
So you start to hear this tonesweeping back down Rather than
continuing on up to 20k.
You start to hear it sweepingback down in the opposite
direction.
Yeah, okay, right.
So if the input gets up to 15know that's 5k above the input or
the above nyquist the outputends up 5k below.
(24:10):
So now the output is back downat 5 kilohertz right, I got you
and as it gets up to 20kilohertz the output goes right
down to dc right.
It goes down through 1K, 500,120.
It turns into this really lowrumble.
So aliasing creates theseinharmonic right.
(24:33):
There's no musical relationship.
The relationship of the inputto the output is a mathematical
relationship between the inputand the Nyquist frequency which
has.
It's not harmonic.
It has nothing to do with themusical timbre or quality of it,
and so it's a very kind of.
It's a very digital type ofdistortion and it just sounds
(24:55):
weird and hashy and usually,since the frequencies are much
higher, it just ends up soundingbrittle.
It kind of adds some high end,but in a very brittle,
unmusically related way.
Yeah yeah, so that's aliasing.
And while we're talking aboutthat, I should just mention
there's also something calledimaging, which is basically the
(25:15):
opposite at the output.
And so, just like we have tofilter the input to make sure
that frequencies above ourNyquist frequency don't make it
in at the output of a converter,we're basically sending out
these pulses every 44,100 timesa second or 48,000 times a
second or whatever, at thespecified value, amplitude,
(25:38):
voltage, whatever and if youdon't then filter those, they
actually create a whole bunch ofextra frequencies above and
below.
So you've got to filter theoutput too, in the same way that
you filter the input.
Same frequency, same roll-off,same phase response.
(25:58):
So you really you're hearing twofilters, right?
You're hearing whatever, thefilter that was used to record
the sound, and then you'rehearing the filter that's being
used to play it back.
And so aliasing and imagingvery similar Again, kind of two
sides of the same coin.
One happens at the input whenyou're going from analog to
digital, and the other happensat the output when you're going
(26:19):
from digital back to analog.
Marc Matthews (26:21):
Yeah, so just to clarify that.
So when we go in aliasing andoutput imaging, imaging, yeah.
So we used the analogy there ofthe 20 kilo sample rate.
Half of that, 10 kilohertz, 10K, 11K then results in Bounces
(26:43):
back to 9K Bounces back to 9K,yeah, and then so forth.
So we go all the way up to 20.
That bounces back to zero.
that's when we get the lowrumble yeah and then when we
have the, the imaging of theoutput, it's another filter on
there.
So we've got two filters, asyou say, that filter going in,
filter going out as well.
Yeah, yeah, it's reallyinteresting stuff, like you say.
I said right at the beginning,it's not we just look at sample
(27:04):
rates but and you just think,well, I'll just go for 44.1 or
go to 48, whatever it may be.
I mean, it's not often that wereally dive deep into okay, well
, why, and it's really why is hedoing that?
Ian Stewart (27:15):
I think it's worth saying here too.
A lot of people hear about thisand all of a sudden, they start
thinking, oh my God, I have toset some high-pass or low-pass
filters.
So I do want to pause andreinforce.
This is not something all thisfiltering that we're talking
about.
It's not something that youneed to take care of.
Yeah, yeah, of course, if yourgear is properly designed which,
to be totally honest, in 2025,even the most kind of consumer
(27:38):
stuff is right.
Yeah, we've gotten pretty goodat this kind of stuff that you
know, even very cheap AD and ADCand DAC chips can do a very
good job of this.
Now, it's not something thatyou need to worry about.
It's just understanding what'shappening in the background and
why the sample rates are whatthey are in relation to the
frequency content that we cancapture.
Marc Matthews (28:00):
Really really interesting stuff.
I realize we're coming towardshalf an hour now with regards to
this this always happens withthe, because it's probably
because I go off on tangentswhen we were talking about the
linear phase and et ceteraearlier.
Yeah, but I think the finalquestion, really, I think for
the audience listening withregards to sample rate and you
touched on it earlier when Iasked you the question about
(28:20):
what sample rate would youadvise?
And we went through 48kilohertz, because the majority
of platforms now are requesting48 kilohertz.
You mentioned YouTube, forexample, so would that be your
advice for sort of streaming aswell?
I don't know how muchexperience you have with
actually uploading it to sort oflike a digital distributor,
(28:41):
something like DistroKid Othersare available.
In that scenario, would youalso recommend 48 if they allow
it?
Ian Stewart (28:51):
yeah.
So usually what I recommendthese days uh, really, for my
clients, right, for my masteringclients I say, um, look, if
they send me mixes at 96k, I'mgonna master at 96k and I will
give them back 96k 24-bit filesas their kind of high res master
and and basically what Irecommend is that they upload
(29:12):
their high res master to distro,kid tune, core, cd, baby,
whatever.
Um, most services have alossless option these days,
right?
Uh, apple music has a losslesstier title, amazon HD, co-buzz,
a lot of them do, so people aregoing to be able to listen at
(29:41):
2496 or whatever in certainscenarios.
The other thing I didn'tmention earlier is that pretty
much any Bluetooth headphonesAirPods, airpod Pros, airpod
Maxes, even other ones 48K is avery standard sample rate for
Bluetooth devices.
I didn't know that.
And when I say standard, youcan't switch it to anything else
.
That's what they run at.
So even if you're listening tosomething higher or lower, it's
(30:02):
going to get converted to thaton playback.
So there's kind of another goodreason to deliver at 48K, right
?
One less conversion step.
But I generally advise myclients to upload 24-bit at
whatever sample rate they sentme their mixes at, which is what
(30:22):
I'm going to master at and kindof conform everything to that
master at and kind of conformeverything to that Um, because
for the few platforms that don'tsupport that Spotify, who for
five years has been saying thata lossless, at least CD quality
tier is coming People keeppromising oh no, it's happening
(30:42):
soon.
I will believe it when I see it.
Um, uh, but my, my feeling isis kind of that if someone is
content to listen on Spotify, orthey don't have the high res
version of whatever service,they're not listening to music
because they want the ultimatein fidelity, they just want some
good music to listen to.
(31:03):
And they probably aren't goingto notice that it's streaming at
some lossy codec, lower bitrate.
Anyway.
There's plenty of other thingsthat are happening to change the
signal a little bit and changethe imaging and high frequency
stuff.
So the sample rate conversionis kind of the least of the
worries at that point.
And then if you do upload your96K 24-bit masters for the
(31:24):
people that have the setup toappreciate it and want to,
they're getting the absolutebest experience that they can.
That said, I think 48k is great.
It kind of shallows out thefilter enough that it sounds a
little bit more open.
You get a little just kind ofmore natural top end and we
(31:44):
can't go there today.
This will have to wait foranother one.
There are other trade-offsnegative trade-offs as you start
to increase your sample rate,certainly above 96 kilohertz, as
you go up to 176 and 192 and384 and crazy shenanigans like
that.
There are actually othernegative trade-offs that most
(32:06):
people in the audiophilecommunity will gladly tell you
you're wrong about and why theythink it's so much better.
Marc Matthews (32:12):
But there there's good science to show that
actually it can create othernasty artifacts yeah, I think
that would be a fantastic uhdiscussion, because it's a very,
very thought conversation,provoking discussion.
I think the negative trade-offof 96k and above, as you say.
I think we'll save that foranother episode and hopefully
(32:32):
get some examples and some otherbits and pieces in there as
well, because I think that wouldbe super interesting.
But touching on superinteresting, I didn't realize
Bluetooth was 48 kilohertz.
I would have gone with theassumption that it was 44.1.
Ian Stewart (32:45):
And it's weird because it's already a lossy
codec, although it's gettingbetter, like aptX, and there are
a few different iterations onBluetooth that have really made
it sound a lot better than itused to.
Yeah, I don't know if I don't.
There may be Bluetooth chipsthat can run at other sample
rates, but usually, for whateverthe device is, it's fixed.
You can't change it.
(33:06):
I know pretty much all ofApple's stuff is 48K.
My older Sony, you knowwhatever Sony model number.
They're really got to loveSony's model numbers Super
memorable.
I can't remember what it is,some string of whatever.
Those were 48K, you couldn'tchange it, so it's pretty
standard and there may be someexceptions, but again it's going
(33:30):
to be set at 44, one orwhatever it is.
Um, so yeah, I really I think 48K is a great deliverable.
If you want to record at 88,two or 96, there's no harm in
that it.
It may make something sound alittle better while you're
mixing right, it gives you alittle more room for those
overtones and the oversampling.
But as a deliverable, 48k isgreat.
(33:55):
And I also should just sayquickly that sample rate
conversion has gottenunbelievably good, like really
really, really good.
And so if you have a masteringengineer who has the right tools
, which anyone with rx, can dothis, um, uh, wave labs,
built-in sample rate conversionis also excellent.
(34:16):
Um, other dawes not so much,but the ones that pay attention
and get it right it can bealmost completely invisible yeah
, so you kind of.
You know, if you need to getdown to 48 K, you can do it, and
if recording at 88, two or 96,you feel like you get more out
of the recording and yourplugins and some things just
work better.
They're great, you know, andand you can easily, at the end
(34:38):
of the mastering stage, comedown to 48, if you want that as
your deliverable.
Marc Matthews (34:42):
So yeah, essentially the options there
isn't it?
Ian Stewart (34:44):
now with the technology.
Marc Matthews (34:45):
we have now the option's there, isn't it?
Ian Stewart (34:48):
Yeah, With the technology we have now.
The option is there, yeah, butI would say 44.1, it's not.
I'm not going to say it's dead,I'm not ready to declare that
yeah, but it's.
There are fewer and fewer caseswhere you really need that
right.
Pretty much the only case thesedays is if you are actually
making CDs, then you've got toget it down to 44.1.
Marc Matthews (35:04):
Yeah, and I suppose as well, while we're
talking about technology, wealso have larger the
accessibility of storage as well, because obviously it's going
to, but in doing having theselarger sample rates is going to
take up more storage.
Ultimately, yeah, now we're ina position whereby we can have
more and more storage.
Ian Stewart (35:22):
Yeah, if you've got it, yeah, why not use it and go
up to that for you.
You can go get a 5-terabytehard drive for $100 or something
, right, yeah?
Marc Matthews (35:32):
yeah.
Ian Stewart (35:33):
These days, why not ?
It's wild.
Marc Matthews (35:35):
It's mad.
I was having a bit of a tangent.
I was discussing the other dayand I used to work for Staples I
think they're still going inthe States.
I don't think it is in the ukanymore, I don't think still
over and I remember selling harddrives.
This was about 20 years ago.
I'm selling the lamb and you wewere selling like memories, and
it was like 128 megabits, yeah,and now, and I, I think I found
(35:57):
one the other day and I waslike man, I remember when people
come into the store and theywould buy like 32 meg, like um
sd cards and stuff, and I waslike, yes, crazy man, crazy yeah
, I had a boss a long time agowho had a story he loved to tell
about his first hard drive thatwas 600 megs, basically a CD's
worth of audio.
Ian Stewart (36:15):
Yeah, that cost him several thousand dollars.
It's the complete oppositerelationship now.
Marc Matthews (36:23):
Yeah, I was watching something the other day
and it was a film and it wasthe early 90s and they
referenced a solid state and Iwas thinking to myself were
solid states I'll get correctedon this were solid states around
in the early 90s?
They may well have been, but asolid state in the early 90s?
Ian Stewart (36:40):
if they were, that would have Insane amount of
money Very esoteric, I think thetechnology and the idea for it
was probably there, but it wasprobably large and very
expensive and difficult to make.
Marc Matthews (36:52):
Yeah, yeah, it was one of those ones where I
was watching the film and Icalled it out and I was like
that's not accurate, being oneof those people.
That's clearly not accuratethat teenager's talking about
having an SSD.
There's no way they couldafford that in the early 90s.
Xst.
Ian Stewart (37:05):
There's no way they could afford that in the early
90s.
Marc Matthews (37:07):
There we go, Ian.
Thank you so much for this thishas been brilliant.
And I appreciate it.
I put you on the spot with afew questions, so I appreciate
your willingness to go intothose subjects.
So I know the audience will getloads out of this, in
particular, because sample rateis one of those ones.
You see questions all the timeonline.
You see discussions and someonesays one thing, someone says
(37:33):
another.
But it's always nice to whenyou're searching for those
questions and someone says, doit at this reason, do it at that
rate, and then explains andexplains why, and this is this
is what this episode is going togive.
We're going to give anexplanation as to why you can
make, and then you can make aninformed choice.
That's the sample rate that youwant to use.
So really, really good stuff.
Before we wrap up and I go intowhat we're going to be going
through in part two, is thereanything you want to share with
(37:54):
the audience?
Alternatively, where can they?
I say alternatively, inaddition, where can they?
Find you if they want to learnmore.
Ian Stewart (38:01):
Yeah, so my website is flotownmasteringcom.
F-l-o-t-o-w-n.
I've got links there.
There's a page that has linksto all the podcasts I've done,
so the past episodes I've donewith you Mark and some other
shows I've been on.
If you want to hear me yammer onabout other things, if you want
to get in touch about mastering, there's a page with kind of a
(38:24):
little about and some of thetools that I like to use, which
you can see some of if you'rewatching the video, and an
intake form where you can justsubmit all your stuff.
Yeah, ian Stewart Music onInstagram.
I don't post a ton, but youknow, there's like pictures of
the dog who left, like if we gohiking, you'll see pictures of
me and the dog or sometimes apost-release stuff.
(38:45):
Yeah, I write for the iope blog.
If you, if you go look formastering articles there, you'll
find some of my stuff.
I just did some videos for wavelab.
It's there.
It's wave labs 30th anniversary, so it's a great time to grab
wave lab if you've been thinkingabout that.
Um, if you are in need of amastering dog certainly not for
everyone yeah, I.
Marc Matthews (39:05):
I don't know.
Ian Stewart (39:06):
But really honestly , mastering is the main thing.
I love mastering, I lovetalking about it and teaching it
.
So if you've got some music youwant to make awesome, I would
love to do that together.
Marc Matthews (39:17):
Amazing stuff Links in the episode description
for the audience listening.
Please do go and check that outand just follow Ian as well.
So whilst we are still in thisepisode, don't forget part two
is going to be the next episode,so it's going to be episode 204
.
So if you're listening to thiswhen this episode drops, you've
(39:38):
got a bit of time, but if you'relistening to this way forward
in the future, make sure youcheck out episode 204, where
we're going to be diving.
I can't get my notes here headfirst.
Why put head first?
Why not?
Yeah, well, head first, I guessnot literally into another
mysterious but crucial topic,dither One.
I don't think I've evermentioned dither on the podcast
(39:59):
before.
No sense.
Ian Stewart (40:00):
Look, if you're going to get involved with
dither, no sense in dipping yourtoes in Just head first.
Marc Matthews (40:05):
Yeah, just go all in All in there, no sense in
dipping your toes in just headfirst.
Yeah, just go all in all inhold your breath yeah fantastic
and uh audience remember as well.
I mentioned it earlier.
If you want to get weekly tools, tips etc and just this week I
included uh ian's wave labs, uhfeatures on youtube as well, so
an example of what you mightfind in that newsletter click
the link in the episodedescription and get those free
(40:27):
weekly tips direct to your inbox.
And until next time, stayinspired, keep creating and
don't be afraid to experimentinside the mix.
Popular Podcasts
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.