00:00:00
everything that's improved my mastering
00:00:01
over the past 10 years each item I pack
00:00:04
with examples walkthroughs and you will
00:00:05
not believe some of the things I've dug
00:00:07
up so strap in and let's go from day dot
00:00:11
as an intern back in 2012 2013 if
00:00:13
something left the studio to anyone at a
00:00:16
label it had to be louder than the
00:00:17
producers mix and we did this mainly
00:00:20
through driving ad converters today
00:00:23
things still need to be loud I do not
00:00:25
care what you read on the internet
00:00:27
the best mastering Engineers compete to
00:00:29
make loud Masters which sound great and
00:00:31
that's a skill in itself if so get ready
00:00:34
and hold on because I have four
00:00:35
different techniques I've developed over
00:00:36
the past 10 years that have helped me
00:00:39
make loud Masters that compete on this
00:00:41
stage the first one you know it I know
00:00:43
it it's clipping and originally I would
00:00:45
clip my ad converter on the TC finalizer
00:00:48
then I eventually got a dangerous
00:00:50
convert ad plus and most recently what
00:00:52
I'm using most of the time is standard
00:00:54
clip okay in the box and it produces the
00:00:57
best results for me now why standard
00:00:59
clip well it's not that this is a better
00:01:01
Clipper than other Clippers it's just
00:01:03
that in the box I have control the
00:01:05
process of clipping a signal with the
00:01:07
aim of getting something exceptionally
00:01:09
loud requires a high degree of fidelity
00:01:13
I mean like to the tenth or 100th of a
00:01:15
decibel you need to be able to set that
00:01:17
that threshold perfectly so that's
00:01:19
Precision analytical tools and I'll show
00:01:22
you exactly how I set this up now so
00:01:24
we've got this track the warden's
00:01:25
warning sign the peak level is 1 dbel
00:01:28
which is all the way over here now that
00:01:30
only means we're at -4.6 LS and we want
00:01:33
to be at least above 10 ls
00:01:36
overall this ceiling here if we bring
00:01:38
down and have the Clipper what did I set
00:01:40
the Clipper on in the session -6 DB so
00:01:43
let's go to -6 you're probably wondering
00:01:46
wow look at all those Peaks you're
00:01:47
shaving off but let me just zoom in and
00:01:49
then we're going to actually listen to
00:01:50
this AB so let's just zoom into this
00:01:52
peak here where are we this -1 DB Peak
00:01:56
which is on the left hand side let's
00:01:58
actually count the the amount of samples
00:02:00
I'm clipping at any given time let's
00:02:03
include these ones here so we go 1 2 3 4
00:02:05
5 6 7 8 plus another nine or so there
00:02:09
let's say maximum there's 20 samples
00:02:12
being clipped at any given instance
00:02:14
there okay 20 samples 48,000 samples a
00:02:18
second okay if we do the maths we're
00:02:21
talking hundreds or thousandths of a
00:02:23
second that the clipping is occurring
00:02:26
which to hear audibly is going to be
00:02:28
almost impossible
00:02:30
[Music]
00:02:32
let me in let me you see
00:02:37
[Music]
00:02:38
feel why do I feel
00:02:41
there you running me look at me got
00:02:46
[Music]
00:02:50
wish I couldn't hear a single difference
00:02:52
now if we go let's just like find a
00:02:54
single hit where it's clipping actually
00:02:56
we'll pick out that one that's on the
00:02:58
exact Peak there
00:03:00
[Music]
00:03:15
identical because this clipping is
00:03:17
happening in hundreds to a thousandth of
00:03:20
a second this is a great tool to use to
00:03:22
get overall loudness with Precision when
00:03:24
you pair it with isotope audio editor
00:03:27
you can actually see how many samples
00:03:29
going to be clipping at any given
00:03:31
instant next is parallel compression and
00:03:33
there's two types we're going to explore
00:03:34
here transparent and color now
00:03:37
transparent parallel compression this is
00:03:38
how I used to set it up I'd set up an
00:03:41
auxiliary bus where are we we'll set up
00:03:43
a stereo Orcs track and we'll call this
00:03:48
parallel comp okay that will go out to
00:03:52
master bus print wait a second parallel
00:03:56
comp so it's going to send out to there
00:03:58
and what we can do is we we can set a
00:04:00
compressor here there we go TDR konic
00:04:03
this is just so slow this is how I used
00:04:05
to do it it's just slow you know set up
00:04:07
a compressor in parallel and then this
00:04:09
way what you can do is you can really
00:04:11
like slam let's go like an 8:1 ratio
00:04:15
here you can really slam the signals in
00:04:17
the loud sections let me just play
00:04:22
that so so it's all slamming it will
00:04:25
Slam in the loud sections but not really
00:04:26
engage in the soft sections so you could
00:04:28
make those soft sections louder and this
00:04:30
is how I used to set it up I went to
00:04:32
like seven eight years ago it was a bit
00:04:33
of a clunky setup okay and I don't and I
00:04:36
didn't really use it all that often
00:04:38
unless I had to use it so it' be a
00:04:39
parallel compressor you set it up like
00:04:41
this and these softer sections become
00:04:43
much louder I'll just turn it on and off
00:04:45
and you'll hear
00:04:47
[Music]
00:04:48
[Applause]
00:04:51
[Music]
00:04:53
[Applause]
00:04:57
that so it make those softer sections
00:05:00
louder and then in these loud sections
00:05:01
you'd be compressing so much you
00:05:02
wouldn't hear it at all but that was
00:05:04
just super clunky now if I want
00:05:06
transparent parallel compression isotope
00:05:08
have made it super easy upward compress
00:05:11
and you can bring up those sections via
00:05:13
this fader here you don't need to set
00:05:15
attack and release times it
00:05:16
automatically gain stages this so these
00:05:18
softer sections you can bring up in
00:05:20
volume and take a look at this because
00:05:22
what we'll do is we'll drive this into
00:05:23
the ceiling a bit when it's hitting the
00:05:25
ceiling none of this will be engaged
00:05:27
this upward compression so in this loud
00:05:28
section have a
00:05:30
[Music]
00:05:34
listen but in the softer section if we
00:05:37
have a look
00:05:38
[Applause]
00:05:41
here adding that 2.9 DB again to those
00:05:44
soft sections so it's a great way to add
00:05:46
loudness to softer sections
00:05:48
transparently now colorful parallel
00:05:50
compression I use this particular
00:05:52
setting in D one tonalization strong and
00:05:55
I use it in two ways either with
00:05:57
completely neutral thresholds so these
00:06:00
thresholds are all exactly the same and
00:06:01
what ends up happening is the low end
00:06:03
gets compressed the top end stays a bit
00:06:05
more open and more of the detail from
00:06:07
the top and gets Blended back in so you
00:06:09
get more of a snappier punchier Vibe so
00:06:11
let's take a listen to
00:06:12
[Music]
00:06:14
that it feels
00:06:17
like associate why do I feel like
00:06:20
there's something High let me in let me
00:06:23
see feel you want to see when it feels
00:06:26
[Music]
00:06:27
like why do I feel
00:06:31
so that's the first way you get a bit
00:06:32
more color in the top end and punch and
00:06:34
everything gets a bit more Fuller and
00:06:35
louder the second way I use this
00:06:38
particular one is to actually balance
00:06:40
out the thresholds here so if we press
00:06:42
option click the link so that the way
00:06:44
it's all evenly compressing so instead
00:06:46
of getting a shift in the tone you just
00:06:48
get a fuller sound all up so I'll
00:06:50
quickly set that up now we'll play this
00:06:53
through let me in let me sit for you
00:06:56
want to see what it feels like inside
00:06:59
yes associate why do I feel like there's
00:07:02
something high you're running you
00:07:04
running right through me you look at me
00:07:06
got
00:07:07
wish
00:07:09
yes let me in let me sit through you
00:07:13
want to see what it feels
00:07:15
[Music]
00:07:19
[Applause]
00:07:21
like all right now take a listen so
00:07:23
we'll go a b which is that version so
00:07:27
we'll swap between these two versions
00:07:29
and you'll get a more Fuller sound so
00:07:32
more Fuller and forward and more Poppy
00:07:34
and top end version so let's have a
00:07:36
listen now let me in let me sit through
00:07:39
you when I see when it feels
00:07:42
like associate why do I feel like
00:07:46
something let me in let me see you want
00:07:49
to see when it feels
00:07:51
[Music]
00:07:55
like so two different ways to play with
00:07:58
parallel actually three different ways
00:07:59
you got the transparent way and then you
00:08:01
got the two different versions of the
00:08:02
color colorful parallel compression with
00:08:04
leapwing d one now have a funny
00:08:06
relationship with saturators I never
00:08:08
understood how any of the harmonic
00:08:09
exiters could be used in mastering
00:08:11
especially when I started out in 2014
00:08:13
they were all super obnoxious with their
00:08:15
sound that's until I came across tone
00:08:18
projects Culver this was sort of my
00:08:20
first subtle flavor maker which had me
00:08:22
start to dabble with more forms of
00:08:24
saturation in mastering outside of the
00:08:26
analog domain and digitally but I really
00:08:29
got started going with this diamond
00:08:31
saturator I think it was last year or
00:08:33
the year before when it was released I
00:08:35
love the way it can get transin to pop
00:08:37
and sound loud and fool and slap on a
00:08:39
dance record without killing my Headroom
00:08:43
take a listen to this before and
00:08:45
after it's wild when when I turn this
00:08:48
off and on take a
00:08:49
[Music]
00:08:54
[Applause]
00:08:58
listen
00:09:00
[Music]
00:09:07
huge it makes all the trends in just pop
00:09:09
out of the speaker in such a unique way
00:09:12
I I really enjoy working with this and
00:09:14
also the Michaelangelo EQ which you've
00:09:15
probably seen me use on the channel as
00:09:17
well now on to our fourth way of making
00:09:19
things louder and that is working with
00:09:20
Isotopes IRC limiters and most recently
00:09:23
available they've been able to put the
00:09:25
Delta mode in it which I've been able to
00:09:27
really push and tweak the loudness with
00:09:28
prec de ision like I haven't been able
00:09:31
to before and I'm going to show you
00:09:32
exactly how we use the Delta mode to
00:09:33
really dial it in so let me give you an
00:09:35
understanding of what the IRC or
00:09:37
intelligent Release Control actually
00:09:38
does uses a set of time domain
00:09:41
parameters which all operate at the same
00:09:43
time so there's multiple of the same
00:09:45
signal different attack and release
00:09:47
times the Delta for each of those
00:09:49
parameters so parameter 1 parameter 2
00:09:51
parameter 3 as like um attack and
00:09:54
release one attack and release 2 attack
00:09:55
and release 3 attack and release four
00:09:57
those Delta signals are scored against
00:09:59
to Psycho acoustic threshold for audible
00:10:01
distortions those scores are then marked
00:10:03
against the modes markers and that time
00:10:06
date and the time demain parameter which
00:10:09
is scored closest to the mode is used to
00:10:12
limit the signal now I read that whole
00:10:14
pattern and did a whole video on it you
00:10:15
can go check that out but this is
00:10:17
happening constantly and the IRC is
00:10:20
constantly switching up parameters based
00:10:22
on which set of attacker release times
00:10:25
match that Mode's best psycho acoustic
00:10:28
distortion model it's it's a bit
00:10:30
convoluted but basically what it's doing
00:10:32
is it's picking the best attack and
00:10:34
release times based on what will produce
00:10:35
the least distortions relative to the
00:10:38
mode that that you're picking now you
00:10:41
can disregard everything I said but you
00:10:43
need to remember this from one mode to
00:10:46
the next the frequency response of what
00:10:49
information gets removed from those
00:10:51
limiters also changes so it's almost
00:10:54
like there's a limiter and a tonal EQ in
00:10:57
one so I'm going to get isotope ozone up
00:11:00
and we're going to flick through these
00:11:02
modes of limiting and monitor the Delta
00:11:04
to really understand and appreciate how
00:11:07
it's changing the sound of the limit the
00:11:09
sound of the master so we can actually
00:11:11
make really good decisions to have loud
00:11:13
Masters that are either a bit more
00:11:14
brighter a bit more beefier in the low
00:11:17
end a bit more thinner in the low end if
00:11:18
we need to tweak it a little bit have a
00:11:27
listen
00:11:54
so if you're listening to some of those
00:11:55
ones the two you probably noticed that
00:11:57
are the most
00:11:59
were the balanced mode and Transit mode
00:12:01
at least those are the two ones which I
00:12:03
think balance mode will leave your low
00:12:05
end as big as it is and transient we'll
00:12:08
focus on keeping those transients intact
00:12:10
but because it's moving faster it's
00:12:12
going to clamp down on a bit of your
00:12:13
lower Subs now I picked transm for this
00:12:15
master cuz it's suited at the best but
00:12:18
let's actually take a listen to without
00:12:20
the Delta so we'll actually listen to
00:12:21
the signal the tone of the master change
00:12:24
between transient and balanced okay take
00:12:26
a
00:12:27
listen
00:12:32
[Applause]
00:12:35
[Music]
00:12:45
so you could hear that the balance mode
00:12:47
thicker Transit mode a bit more
00:12:49
controlled in the lowend I preferred the
00:12:51
more controlled low end but that's just
00:12:53
how I roll now while we're in this
00:12:55
session that brings me to my next thing
00:12:57
which I sort of had to swallow my ego
00:12:59
one all right and that was stem verse
00:13:01
stereo mastering now early on in my
00:13:04
career I got rinsed absolutely rinsed in
00:13:07
a form of Pros because I was dismissing
00:13:10
just completely dismissing stem
00:13:12
mastering as faux Mastery I was a bit
00:13:14
young naive I felt a bit insecure
00:13:17
because I hadn't actually done a great
00:13:18
deal of stem mastering and because of
00:13:21
that I sort of put it down uh brace
00:13:23
yourself though because St St mastering
00:13:26
is probably one of the best things I've
00:13:27
pushed myself into learning take a
00:13:30
listen to what you can do before and
00:13:32
after in a stem mastering
00:13:34
[Music]
00:13:40
[Applause]
00:13:41
[Music]
00:13:57
session
00:14:06
that sound Bonkers now this doesn't mean
00:14:09
every session should be stem mastered
00:14:11
it's technically a caseby casee sort of
00:14:13
thing but there's no way I could have
00:14:16
manipulated those kick drums the high
00:14:17
perk bring the base up half a decibel or
00:14:20
the or or the vocal line down 1 dbel in
00:14:24
a stereo master and do it cleanly having
00:14:27
the stems there allowed me to the
00:14:29
Integrity of the that mix but just
00:14:30
elevate it
00:14:31
all I think that's a huge win I I never
00:14:34
used to do stem mastering when I started
00:14:37
out and it was more about my
00:14:38
insecurities not really freaking knowing
00:14:40
how to do it but it's something that we
00:14:42
should really Embrace as mastering
00:14:43
Engineers especially competing in
00:14:45
today's day and age you're in for a
00:14:47
surprise with this next one analog verse
00:14:49
digital when I started I was completely
00:14:51
hybrid 50/50 analog digital pretty much
00:14:53
every session bit of analog bit of
00:14:55
digital last year I did a mastering sh
00:14:58
shootout shootout or shout out I did a
00:15:00
mastering shootout with a my good friend
00:15:01
will Bower where I did a digital master
00:15:03
and he did an analog one of the same
00:15:05
song and I never really revealed the
00:15:07
blind files on the video people
00:15:11
commented on which one they thought was
00:15:12
analog which one they thought was
00:15:13
digital and what was funny was people
00:15:16
were completely split 50/50 in the
00:15:18
comments and half of them got them wrong
00:15:21
as well so people thought oh this one's
00:15:22
the analog this one's a digital etc etc
00:15:24
also what was crazy is half of them
00:15:26
preferred the digital half of them
00:15:28
prefer the analog again 50/50 split I'm
00:15:31
going to play you that exact same AB for
00:15:33
you guys to enjoy let me know in the
00:15:35
comments which one you think is the
00:15:37
analog and which one is the digital and
00:15:38
I'm just going to have a laugh seeing
00:15:40
all the comments being split 50/50 again
00:15:44
um let's roll with it and then I'm going
00:15:46
to tell you how I'm going about things
00:15:48
now it's going to be all
00:15:53
right be patient
00:15:56
child be patient
00:16:00
child it's going to be all
00:16:05
right be patient
00:16:09
child be patient
00:16:12
child it's going to be all
00:16:18
right be patient
00:16:21
child be patient
00:16:24
child it's going to be all right
00:16:29
[Music]
00:16:31
be patient
00:16:33
child be patient
00:16:37
child so I would you guys to leave in
00:16:39
the comments B or C which one was analog
00:16:42
which one was digital but what do I do
00:16:44
now so I still use analog in some
00:16:48
sessions and by some sessions this is my
00:16:50
mentality when it comes to it the low
00:16:52
end's really big but it is pocketed
00:16:55
nicely like there's nothing to choose
00:16:57
about it's just a big low I don't really
00:16:58
want to eat C it I want it to feel
00:17:00
natural and dynamic I'll head to the
00:17:02
analog chain cuz the varu has a really
00:17:04
nice way of managing that quite
00:17:07
musically at least to my ear I can't do
00:17:09
anything what that this varu does to
00:17:12
mixes that are big low ends that are
00:17:13
quite even um that need that that sort
00:17:16
of dynamic movement the veryu deals with
00:17:18
that really well or if I want a very
00:17:20
unique kind of limiting I'll go through
00:17:22
the analog chain so I get a very smacky
00:17:25
Pokey kind of limiting because it's a I
00:17:28
don't know why it it does something
00:17:30
really nice to the transients when I'm
00:17:32
driving it hard um this analog chain so
00:17:35
if I need a very specific kind of
00:17:37
limiting it tends to be more still on
00:17:39
rock songs I won't do this in um
00:17:41
actually know sometimes I done it on
00:17:42
dance music where I've wanted a
00:17:43
particular tone things have been a bit
00:17:45
too sterile but that's like 10 20% of
00:17:47
sessions I'm doing that the other 70 to
00:17:49
80% of sessions I'm doing I'm doing
00:17:51
completely digitally in the box and why
00:17:54
because of precision and flexibility and
00:17:57
ability to really tweak can get in there
00:17:58
and do something magical um which is
00:18:02
funny because you're going to have to
00:18:04
brace yourself to cringe at this next
00:18:06
thing um which is on this list because
00:18:09
while I'm talking about precision and
00:18:10
flexibility there's also a deal of
00:18:12
restraint that I had to learn over the
00:18:14
last 10 years because when I first
00:18:15
started out I believed everything I did
00:18:17
in mastering had to be surgical okay I'm
00:18:20
going to show you how I used to manage
00:18:22
low end in a mix and uh this is
00:18:25
embarrassing this is so embarrassing all
00:18:27
right so this is is how I used to use
00:18:29
linear phase EQ I would basically carve
00:18:33
out the Peaks and the trough so if there
00:18:35
was a trough I'd boost it up and if
00:18:37
there's a peak I'd cut it down and I'd
00:18:39
use only the low shelves so they would
00:18:42
stack on top of one another and you've
00:18:44
got like three four different impulse
00:18:46
responses now happening to keep this
00:18:48
linear phase so listen to the transing
00:18:51
the kit how you completely lose it you
00:18:55
completely lose it and I'll show you
00:18:58
this simplest solution to this okay so
00:19:00
listen to this before and
00:19:04
[Music]
00:19:16
[Music]
00:19:24
after it was just such a stupid way to
00:19:26
do things cuz I thought I was being more
00:19:28
fancy linear phase you got fancy you
00:19:31
know like it's e this Precision sort of
00:19:35
way of dealing with it you know what's
00:19:36
about this you know the best
00:19:38
way to deal with this and if you clicked
00:19:39
on this video you probably seen the
00:19:41
thumbnail you want to control the lowend
00:19:43
yeah and you don't want to the
00:19:44
transients up so to speak you know what
00:19:46
you do okay I'll get rid of these here
00:19:50
we go two DB there all right we're still
00:19:53
going to keep our punch and we're going
00:19:54
to control the low end probably put this
00:19:56
down to 50 and take a listen now before
00:19:59
and after we're still going to keep our
00:20:00
punch at
00:20:04
[Music]
00:20:11
least how simple the lesson keep it
00:20:14
simple stupid because freaking H I used
00:20:17
to do weird like that even with
00:20:18
like midside EQ trying to notch things
00:20:20
out um because mastering had to be
00:20:22
surgical no it doesn't always have to be
00:20:24
surgical you have to have a level of
00:20:26
precision and care and that is a
00:20:27
complete completely different thing to
00:20:29
doing 1,0 freaking EQ
00:20:34
moves let's move on let's move on cuz
00:20:37
that one was a hard one to put out the
00:20:39
next thing on this list I can't believe
00:20:41
I used to say this to clients and it was
00:20:43
when they weren't happy with how loud
00:20:46
their masters were I would say it was
00:20:47
the mixer's problem regardless if it's
00:20:50
true or not it wasn't professional and
00:20:52
it was a cheap attempt to absolve
00:20:54
responsibility for my duty as an
00:20:55
engineer and I've owned that since as
00:20:58
I'm grown into my role as an engineer
00:20:59
I've learned we need to take
00:21:00
responsibility and solve problems for
00:21:03
our clients that's what they pay us to
00:21:04
do not to pass blame onto an arbitrary
00:21:07
thing yeah the mix might not be good and
00:21:09
different mixes have different qualities
00:21:11
but it's our responsibility to own that
00:21:13
and in today's day and age in 2024
00:21:16
isotope ozone 11 stem Focus mode with
00:21:19
transient sustainability yeah it makes
00:21:22
it impossible to find excuses unless a
00:21:25
mix is completely so far gone you should
00:21:28
never have taken it on in the first
00:21:29
place there's a lot you can do to remedy
00:21:32
small elements of a mix to help you get
00:21:33
a better Master Okay simple as that so
00:21:37
I've got the drums here I can focus on
00:21:39
the trans in of the kick which is a
00:21:41
little bit too big and I can manipulate
00:21:43
and control that take a
00:21:57
listen
00:21:59
the next one will make people cry after
00:22:01
reading their credit card statement and
00:22:02
that is price does not equal quality I
00:22:05
have hundreds and hundreds of plugins
00:22:07
some are very expensive and one this one
00:22:11
here TDR konicom completely free and I
00:22:14
use this plugin every single day there's
00:22:17
no reason to feel like you need to spend
00:22:19
big big money to get great results when
00:22:21
there are tools like this out there so
00:22:23
here are four plugins I is in the studio
00:22:25
which you can get right now right now it
00:22:28
won't they free or won't cross you a lot
00:22:30
T konikov free go download it free clip
00:22:34
free go download it standard clip it's
00:22:37
$25 barely anything mjuk which is
00:22:40
actually an emulation of the varu which
00:22:43
actually does it pretty damn Faithfully
00:22:45
is only
00:22:47
€4
00:22:49
so go do it moving on to attack and
00:22:53
release of fake news now hold on to your
00:22:55
seat because I'm going to take you
00:22:56
through my stages of understanding the
00:22:58
parameters of attack and release first I
00:23:01
thought attack and release was how long
00:23:02
it takes a compressor to start which is
00:23:05
wrong attack and release is how long it
00:23:07
takes for the compressor to heat its
00:23:08
maximum gain
00:23:09
reduction better but it's still wrong
00:23:13
and that's because attack and release
00:23:15
times correlate as logarithmic charge
00:23:16
and discharge values so when you set an
00:23:19
attack time the charge cycle of one
00:23:20
logarithm which is one over eish is
00:23:23
actually 2/3 of the maximum gain
00:23:25
reduction and subsequently following
00:23:27
this logarithm it'll take about five
00:23:29
times longer than the attack time itself
00:23:31
to reach maximum game reduction and the
00:23:33
inverse goes for release now this isn't
00:23:36
a recent discovery I've stumbled across
00:23:38
this has been the case for over 15 years
00:23:40
of circuit design with compressors and
00:23:41
capacitors and many designers having
00:23:43
managed actually to manipulate and use
00:23:45
different topologies to interact with
00:23:47
the linear the linearity and
00:23:49
characteristics of how different
00:23:51
capacitors charge and discharge and
00:23:52
different Arrangements in order to make
00:23:54
things faster or slower but for me this
00:23:57
was literally my learning curve no pun
00:24:00
intended it was just that understanding
00:24:02
the ballistics of attack and release
00:24:03
times was something that I had to learn
00:24:05
and I think um anybody that goes down
00:24:07
that rabbit hole there's a lot of value
00:24:08
on the other end of that Discovery this
00:24:10
next one is cringe because you're not
00:24:11
going to believe what I used to send out
00:24:13
in emails when I was starting out take a
00:24:15
read of this did you read that -3 DB of
00:24:18
head room head room for mastering in the
00:24:20
digital age is a myth 32-bit float mixes
00:24:23
means I don't care what level your
00:24:25
mixers finish at and here the full
00:24:27
breakdown of 32-bit 24-bit and head room
00:24:30
and how it all works modern doors like
00:24:32
Protools logic qbase all internally
00:24:34
processed audio with 32-bit float
00:24:37
systems or sometimes even 64-bit a
00:24:39
32-bit float system uses a 32-bit word
00:24:41
made of a 24-bit fixo word bit and an
00:24:45
added 8bit exponent the 8bit exponent
00:24:47
allows the mix engine to use a 24-bit
00:24:49
word to accurately represent values
00:24:51
outside the constraint of a 24-bit fixo
00:24:54
architecture so you can represent values
00:24:57
greater than 144 DB of dynamic range or
00:25:00
the 16,777,216
00:25:02
present values that are within that
00:25:04
range now if you want to full breakdown
00:25:06
behind the geeky science behind this
00:25:08
I've left a link in a publicly available
00:25:10
exerpt from the book Protools Advanced
00:25:12
Techniques and workflows in the
00:25:13
description below in Practical terms
00:25:16
when mixing this means you can go over 0
00:25:18
DB full scale in your Daw as the 8bit
00:25:21
Xmen and provides that extra head room
00:25:23
to maintain all the relevant information
00:25:25
without clipping and I've set up a test
00:25:27
here to show show you that the maths
00:25:28
check out the first file is at 0 DB full
00:25:31
scale 32bit the second sine wave is at +
00:25:35
3 DB full scale 32bit the third S Wave
00:25:39
is also at plus 3 DB um but it is 24 bit
00:25:43
so it isn't 32bit with the 8bit float
00:25:45
it's 24bit and the final one is a -3 DB
00:25:50
32bit they're the same sign wave at
00:25:52
different levels exported so we got 3
00:25:54
32bit files 24bit these have all being
00:25:58
level matched to the -3 dbel 1 so the 0
00:26:02
dbel 1 is gained down to the -3 the
00:26:05
positive 3 DB 32bit export is gained
00:26:08
down to the -3 and the positive 3 dbel
00:26:11
24bit export is gained down to the3 now
00:26:14
I've level matched this to see they n
00:26:15
with one another everything knows except
00:26:18
the 24-bit file which hard clips above
00:26:26
zero
00:26:37
so you can hear that hard clipping
00:26:39
straight away and if you want to see
00:26:40
some waveform analytics here where's the
00:26:43
plus3 dbel 24bit so if I go on the plus
00:26:47
3 dbel 24bit I zoom all the way into the
00:26:49
waveform you can see the hard clipping
00:26:51
sample points there but if I go to the
00:26:54
32-bit
00:26:56
file
00:26:58
you can see that those sample points
00:27:00
actually reconstruct above zero because
00:27:04
it's got that extra head room and if we
00:27:05
actually look at the full scale of this
00:27:08
you can see all that extra head room you
00:27:09
have on a 32bit file now when I started
00:27:12
back in 2014 it was Universal everyone
00:27:15
would say TR Peak limits have a sound
00:27:17
even the best Engineers would avoid Tri
00:27:20
Peak limits for their sound now
00:27:23
quotation marks sound because we were
00:27:25
all M looks without any clue of to what
00:27:28
was actually going on true Peak limits
00:27:30
don't have a sound they just measure
00:27:32
Peak levels different the ballistics of
00:27:34
the attack and release are identical and
00:27:37
here's my research and the sum of these
00:27:39
syn wave values represented as a
00:27:41
waveform which we digitally see in a Daw
00:27:44
can produce over and unders shoots below
00:27:46
and above the given peak of the given
00:27:49
sample value now this presents a
00:27:52
significant issue with the accuracy of
00:27:53
digital limiters at most conventional
00:27:55
sampling rates 441 or 4 AK there can be
00:27:58
up to 3 DB of difference between the
00:28:00
discrete PCM sample values and the
00:28:02
resulting waveform which means there's
00:28:05
often audio passing through the
00:28:06
threshold of a digital limiter without a
00:28:08
corresponding PCM sample point for it to
00:28:11
react to now this is an issue and in
00:28:14
fact a major flaw in the function of
00:28:16
normal digital limiters so how does a
00:28:20
true Peak limiter seek to solve this to
00:28:23
achieve the same physical Behavior as
00:28:24
analog TR Peck digital limits emulate
00:28:26
the data a converter and measure its
00:28:28
output digitally speaking one way to
00:28:30
emulate the daa conversion process is to
00:28:32
interpolate more discrete values in
00:28:34
between the present ones so the limiter
00:28:36
can measure the level of a signal from a
00:28:38
more accurate sample Point itbs
00:28:42
17704 lays out a consideration for
00:28:45
accurate Peak metering of digital
00:28:47
signals now with the standard of four
00:28:50
times oversampling this minimizes a
00:28:52
sample Peaks underere from 3 DB down to
00:28:55
668 DB and and at eight times
00:28:58
oversampling this minimizes a sample
00:29:00
Peaks underere down to1 169 which is the
00:29:03
same standard used for Fab filter Pro
00:29:05
l2's true Peak limiter I found this
00:29:07
article published in 2015 by Isotopes
00:29:09
principal software engineer where he
00:29:11
lays out some code in Python to compile
00:29:13
this syn wave offset a fraction of a
00:29:15
sample the sample Point has been offset
00:29:18
by 38 of a sample or
00:29:21
375 now I actually failed to compile
00:29:24
this code so I reached out to a few
00:29:25
friends within the programming space and
00:29:27
one of them was able to compile the code
00:29:29
and produce me with the necessary PCM
00:29:31
encoded sync wave for my test which is
00:29:34
this one here so a big shout out to
00:29:36
Kevin thanks for that the test is simple
00:29:38
given that we've now established that
00:29:40
limit is function by how they interpret
00:29:42
the signal a digital limiter interprets
00:29:44
that signal by the PCM encoded sample
00:29:46
Peaks PCM encoded data can underere up
00:29:49
to 3 DB at 48 khz by interpolating new
00:29:53
sample values we minimize this under
00:29:55
within 668 DB And1 169 DB using four and
00:30:00
8 times oversampling respectively as
00:30:03
calculated and presented in the it
00:30:05
documentation so can I get Fab Filter
00:30:08
prel 2 true Peak limiter to null with
00:30:10
Fab Filter pral 2 using eight times
00:30:12
oversampling when setting the threshold
00:30:14
to the PCM encoded sample Peak and in a
00:30:17
second test matching the threshold to
00:30:18
the reconstructed or interpolated true
00:30:21
Pak now with this shifted syn wave we
00:30:23
have a sample point at
00:30:25
8.13 so here I have a true Peak limiter
00:30:28
set to 8.13 and here I have a normal
00:30:31
limiter set to 8.13 FB Filter paril 2
00:30:33
interpolates the data at 8 times
00:30:35
oversampling this one does not so what
00:30:37
we're going to do is we're going to play
00:30:39
this back in a loop and then what I'm
00:30:40
going to do is in the middle of that
00:30:42
Loop I'm going to switch eight times
00:30:43
over sampling on here and null it and
00:30:46
you'll see that it almost nulls
00:30:48
absolutely perfectly well below the
00:30:50
noise floor so let's do that
00:30:54
now okay so one thing we notice is the
00:30:57
true pick limit is definitely doing some
00:30:58
limiting the normal one isn't
00:31:00
recognizing that sample point but what
00:31:02
happens when we turn over sampling
00:31:04
on oh we've reconstructed sample points
00:31:08
or interpolated sample points in between
00:31:10
the existing ones and it's now
00:31:12
recognizing that it has to be limiting
00:31:13
it's actually missing that information
00:31:15
so take a look at this null test when I
00:31:18
so it's about 140 DB is the difference
00:31:21
indistinguishable now something happened
00:31:23
in 2020 I think you know what and it
00:31:25
forced me into using headphones when was
00:31:27
away from the studio or when I was at
00:31:29
home in the mornings during curfews
00:31:31
because I couldn't leave the house
00:31:33
anyway up and until this point I hated
00:31:36
headphones and it wasn't until I started
00:31:38
using Odyssey headphones that I actually
00:31:41
enjoyed listening and working on them
00:31:43
this isn't a plug but they managed to
00:31:45
remedy all the things and limitations of
00:31:48
traditional headphones which I couldn't
00:31:50
stand and here's a breakdown of how
00:31:52
their drivers work and actually achieve
00:31:54
such an awesome sound for mixing and
00:31:56
mastering sound production is physics
00:31:58
and something no product can hide away
00:32:00
from fortunately for you I've read the
00:32:02
patterns and come to understand
00:32:03
objectively how and why these sound the
00:32:06
way they do so your standard headphone
00:32:07
with a voice coil and Cone is driven
00:32:09
from the center outwards the surface
00:32:11
area of the diaphragm won't move in
00:32:13
unison with itself and when the force
00:32:15
compromises the diaphragm structural
00:32:17
Integrity many Sonic limitations ensue
00:32:20
such as breakup modes mechanical
00:32:21
filtering transmission delay and local
00:32:23
resonances a PL of magnetic headphone
00:32:25
seeks to solve much of this by susp
00:32:27
bending a thin diaphragm above a planer
00:32:29
array of bar magnets when a current
00:32:30
passes and induced magnetic field is
00:32:32
created the conductor and the attached
00:32:34
diaphragm then begin to be drawn in and
00:32:36
away from the planer magnetic bar
00:32:38
array that was a little bit of rhyming
00:32:41
there that's an explanation of
00:32:42
traditional plom magnetic headphones and
00:32:44
we don't see them often in the studio
00:32:45
space because of the net negatives this
00:32:47
brings to the headphone design typically
00:32:50
these become not this unit but PL of
00:32:53
magnetic headphones become very heavy
00:32:55
because of the magnet arrays and also
00:32:58
they tend to be a higher impendence
00:33:00
which means to drive them you need more
00:33:01
powerful dedicated headphone amps the
00:33:04
technology behind mm500 solves much of
00:33:07
these issues Force equals mass by
00:33:09
acceleration and these diaphragms are
00:33:11
ultra thin how thin 1.8 microns thin a
00:33:15
red blood cell is 5 microns which means
00:33:17
they don't have a great deal of Mass to
00:33:19
work with nor do the respective
00:33:21
conductors carry a great deal of current
00:33:23
because if the conductors are too heavy
00:33:24
it compromises the diaphragm and this is
00:33:26
an issue because headphones become less
00:33:28
efficient unless you pack this with more
00:33:30
bigger powerful magnets but then that's
00:33:33
going to make this very heavy and this
00:33:35
is where the patented flux or magnetic
00:33:37
array plays a big hand in maximizing the
00:33:40
magnetic flux or field without adding a
00:33:42
ridiculous amount of weight to the
00:33:44
resulting driver and it actually
00:33:45
increases and improves the efficiency of
00:33:47
how these headphones operate I can go
00:33:49
into a whole video on that alone if you
00:33:50
do let me know in the comments but in
00:33:52
the meantime let me summarize this in
00:33:54
music producer and mixing engineer lingo
00:33:57
accurate low frequency extension into
00:33:59
the subs fast transient response without
00:34:02
being super heavy another side effect
00:34:05
and actually it's a benefit of this
00:34:07
array is that it becomes low impendence
00:34:09
it's 18 ohms in fact compared to iPod
00:34:12
headphones which are 42 ohms which means
00:34:14
you can drive these from your MacBook
00:34:16
which is actually something Manny marer
00:34:18
mentioned in his interview with sound on
00:34:19
sound now if you stayed till now this
00:34:21
last lesson is the cherry on the cake
00:34:23
and it's the one worth waiting for in
00:34:25
the best lesson I've learned and that is
00:34:27
nothing is done until it's done
00:34:29
literally you have not finished
00:34:31
mastering a project until it is released
00:34:33
and out in the world and likewise you've
00:34:36
never entirely finished learning until
00:34:38
you're literally in the grave I know
00:34:40
that sounds a bit morbid but you get the
00:34:42
point these are King lessons and things
00:34:44
I've took on board over my first 10
00:34:46
years and what the next 10 hold who
00:34:48
knows but there's no stopping and nor
00:34:51
should there be for you so I'm going to
00:34:53
leave you there and until next time take
00:34:55
care
