Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:04):
Welcome to tech Stuff, a production from I Heart Radio.
Hey there, and welcome to tech Stuff. I'm your host,
Jonathan Strickland. I'm an executive producer with I Heart Radio
and I love all things tech and recently I received
a tweet from Twitter user Salvatore del Knock, a k
(00:26):
A non juror, asking if I would do a breakdown
on how analog to digital and digital to analog audio
converters work. And that's a great request. Um, it is
incredibly technical when you really get down to it. So
I'm going to do a very high level view of
the concept because otherwise we're gonna have to get into
(00:49):
the various methodologies that DAC and a d c's work,
and uh, it would quickly become like a technical manual.
But if people want that, then I can do a
subsequent episode and go into more detail. But one of
the things about this is that lets us talk about
the differences between analog and digital audio and why converters
(01:13):
are necessary in the first place, and to open up
the eternal argument about whether one is inherently better than
the other. This one goes out to all you audio
files out there, so get ready to send me angry messages,
because no matter what I say, some of y'all are
going to get upset. Anyway, let's start with what it
means to be analog versus digital. Now, when I was
(01:35):
a young boy, nobody loved me. I was a poor
boy from a poor family. No, hang on, that's now,
that's Queen's bohemian rhapsty Now, when I was a young boy,
analog was the standard. Digital did not even enter into
my awareness until I was a teenager, when compact discs
were starting to become popular. They had been around for
(01:56):
a while before I was a teenager, but I was
not really aware of them, because I mean, I grew
up in rural Georgia. We would get technology a few
years behind everybody else. Anyway, I grew up thinking analog
essentially meant old and digital meant new, Like that was
the sort of the abstract distinction between the two in
my head. But the differences are obviously more complicated than that,
(02:20):
and we need to understand how sound works, which I
know I've covered many many times, but it's important so
that we know how the analog and digital methods of
recording and thus reproducing and eventually playing back sound. You
know how they work with relation to the original sound
that existed. So sound is, when you really get down
(02:43):
to it, vibration or pressure waves. Now, we mostly experienced
sound by hearing these vibrations travel through the air, but
you can also experience this underwater. Sound can move through
different media, including solid material. Like if you put your
ear against a table, a really long table, and some
one on the other end is tapping very lightly on
(03:06):
that table, you'll hear it. And it's not because the
sound is traveling effectively through the air, though it is
doing a little bit of that too, but that it
travels through the table to you. Sound also travels at
different speeds through different media, and in fact, stuff like
air temperature can affect how quickly sound travels, which is
(03:26):
why when we talk about the speed of sound, we
technically actually need to be a little more specific than that.
So the standard way of describing the speed of sound
is to say that it moves at three per second
in dry air at twenty celsius, that's about sixty eight fahrenheit.
And if you start changing those parameters, you know, if
(03:47):
you introduce, say a lot of humidity into the air,
or you change the air temperature like it goes up
or it goes down. Well, sound will travel at a
slightly different speed than at that standard I was talking about.
Now I could get into how the vibrations cause air
molecules to move back and forth, creating little changes in
air pressure. And it's these pressure waves, these air fluctuation changes,
(04:11):
that our ear drums pick up and transfer to our
inner ears. That's where special nerves pick up these fluctuations
in our inner ears, and then our brains process those
those nerve signals as sound. But most of this isn't
important for the rest of this episode, so instead, let's
talk about sound waves, all right. So we can think
(04:31):
of a vibration as something in which a particle is
moved out of its usual place and then it snaps
back to its usual place, and it might do this
several times. Think of a guitar string. If you pluck
a guitar string, you're pulling the string out of where
it usually sits, and then it snaps back and forth
and oscillates around its normal position until it settles down again.
(04:56):
So we can describe the number of times that a
particle does as a frequency, you know, or the number
of times a string goes from one point all the
way across and back to that starting point over the
course of a second. So with sound, we usually use
the unit hurts to measure frequency. If a particle only
did one cycle of vibration per second, if it took
(05:19):
a full second for it to go from the you know,
the one crest to the next crest, uh, then it
would be one hurts. That would also, by the way,
be a frequency that was way too low for us
to hear. Typical human hearing has a range of around
twenty hurts at the low end, to twenty thousand hurts
(05:40):
or twenty killer hurts, in other words, on the high end.
So for stuff vibrating in a cycle that's twenty times
a second all the way up to twenty thousand times
a second, that's something we could potentially hear. Now. I
say potentially because that is typical human hearing. There are
people who can hear outside of that range a little bit,
and then there are some of us, especially as we
(06:02):
get older, who can hear a more narrow range of frequencies.
But frequency is just one part of how we describe sound.
We can also describe sound by how loud it is.
The volume of sound. So from a physics perspective, we
can think of this is how much pressure the sound
places upon our ear drums. You know how dramatic those
(06:24):
fluctuations and air pressure are. In other words, But loudness
isn't just down to physics. The way we experience loudness
depends not just on that sound pressure itself, but stuff
like psychoacoustics. That's how our brains perceive sound in the
first place. But now we've got two criteria we can
use to assign to any sound correct Like, we can
(06:46):
talk about the frequency of that sound, you know, how
frequently that those particles are vibrating, And then we can
also talk about the displacement of those particles vibrating, or
what we might think of as the loudness or volume
of that sound. We could then plot a sound wave
as a transverse wave on a graph, and we could
have the X axis, you know, the horizontal axis of
(07:09):
this graph representing the passage of time. So on the
left side we might say zero, and we say time
increases as you go to the right. The y axis
we could have being displacement, which kind of you know,
amplitude or volume in other words, and we could then
plot all the points where a particular vibrating particle would
occupy over a given span of time. If we had
(07:33):
a sound of a steady frequency, then we would end
up with a wave that would look a lot like
a sign or cosign wave. The distance between two consecutive
crests of this wave would be the wavelength for that sound,
and the sounds wavelength has an inversely proportional relationship with
the sounds frequency, So the higher the frequency of sound,
(07:56):
the shorter the wavelength will be. So deep bay Ace
notes would have sound waves that have much longer wavelengths
than very high pitched high frequency notes. Uh frequency relates
to pitch. There isn't like an easy mathematical way we
can kind of relate pitch, by the way, There are
(08:17):
easy ways we can relate frequencies, but it gets a
little tricky anyway. The reason I even talk about plotting
sound waves at all is that it makes us easier
for us to consider the differences between analog and digital
audio recording. Keep in mind, if we plotted that sound wave,
that's not that's not the physical sound wave that we've
(08:38):
just plotted. That's our description of that sound wave, its frequency,
and its loudness. Um The classic sign wave like depiction
of the sound wave shows us that there's a continuous
representation of sound across time. It is unbroken. We can
put plot, you know, even complicated sounds with changes in
amplitude and frequency, and the shape of the waves tells
(09:01):
us a little bit about the tambre or quality of sound. Now,
by quality, I don't mean, oh, this sound is very
good quality or this sound is really bad quality. Instead,
I'm talking about the elements that differentiate say piano playing
middle C from a guitar playing that same note middle C.
(09:23):
Both instruments are producing the same note at the same frequency,
assuming both instruments are you know, properly tuned, and both
of them are using the same pitch tuning, but you
would hear a difference in the type of sound between them, right,
A piano and a guitar sound different. Otherwise all instruments
would produce exactly the same kind of sound as each other.
(09:45):
But you know, you can tell the difference between a
piano and a guitar, or a clarinet or a flute
or whatever. The tambre is different, even if the instruments
are all producing you know, technically the same frequency, even
at the same volume. This leads us to the fact
that sound is this continuous thing for us. It isn't
happening in discrete units. It's kind of like the difference
(10:08):
between jumping into a pool filled filled with water, which
is you know, continuous to us because we can't you know,
experience it down on the molecular level, or jumping into
a pool that's filled with plastic balls. So to us,
sound is kind of like a fluid, and analog recording
captures that. The analog approach to recording is older than digital.
(10:33):
So way way back in the nineteenth century, folks like
Alexander Graham Bell, we're trying to figure out how to
transmit the human voice across great distances using electricity, and
the microphone was one half of what was needed to
do this, the loud speaker being the other half. And
the basic way a standard microphone works is to convert
(10:54):
sound that continuous you know phenomena of pressure wave changes
to a varying electric signal, an electric signal that has
varying voltage. This is another continuous phenomena, right, it's unbroken,
it's it's like another wave. Here's how it works. So
inside an analog microphone is a tiny little diaphragm, typically
(11:17):
made of very thin plastic, and it behaves in a
way similar to how our ear drums work in our ears.
So when sound, you know, these pressure waves hit that microphone,
it moves the diaphragm back and forth, and the diaphragm
is actually attached to an electro magnet. A simple microphone
could have a permanent magnet inside it, and wrapped around
(11:40):
this permanent magnet is a little coil of metal wire
that connects to the diaphragm. So the diaphragm moves the coil,
which then moves along the length of this permanent magnet.
That introduces a fluctuating magnetic field, or rather, you know
the effect of a fluctuating magnetic field. The permanent magnets
magnetic field is stable, but moving a coil through a
(12:01):
magnetic field, it's the same thing as if you were
to fluctuate a magnetic field around a you know, non
moving coil, you get the same effect. Now, the laws
of electromagnetism tell us that if you have a conductive
material and it encounters a fluctuating magnetic field, that field
will then induce an electric current in the conductive material.
(12:25):
So now you've got the microphone producing an electric current,
and again the voltage of this current varies depending upon
the sound hitting the microphone. That means the microphone is
a type of transducer. That's a device that converts one
form of energy, in this case acoustic pressure, into another
form electric signals. Now, you could send this electric current
(12:49):
with varying voltage somewhere to do something else interesting, like
you could have it go directly to allowed speaker for playback. Now,
of course, this electric current is really uh there are
you know, very small elements in your microphone, right, so
it cannot produce an incredibly strong electric current. So typically
(13:11):
you would first pass this electric current through an amplifier,
which increases the strength of the signal. I'm not going
to go into how amplifiers work. I've talked about in
other episodes, and it would mean that this this episode
would go like an hour and a half long if
I were to to dive into that. The important thing
to think of is that amplifiers take incoming week signals
(13:33):
and then push out a stronger version of that same signal.
Assuming the amplifiers working properly, then that signal could go
to a speaker and you would have the same process
that you had with the microphone, only in reverse. The
speaker also has a voice coil inside it, a coil
of you know, conductive of metal wire, and also a
(13:56):
magnet inside the loudspeaker. So the incoming current goes to
the wire, and we know by the laws of electro
magnetism that this means the flowing current through the wire
will also produce a magnetic field. I mean, this is
how electro magnetism works, and that this magnetic field will
then pull and push against the magnetic field generated by
(14:16):
the permanent magnet that's already inside the speaker, and this
in turn creates the force that pushes and pulls the
cone inside the speaker that connects to another diaphragm. This
is a much larger diaphragm than the one that's on
the microphone on the other side. Right, Because you've boosted
the electric signal, it can then have enough power to
(14:37):
move this larger diaphragm. So this larger diaphragm begins to
move in and out, and it's pushing and pulling air
and it's just recreating the acoustic pressure waves that we're
used to go into the microphone and generate the electric
signal in the first place, so you're kind of preserved
this experience from sound going into a microphone. The microphone
(15:01):
as a transducer, transforming that acoustic pressure into an electric
current with varying voltage, sending that to an amplifier, and
then a speaker, which then does the opposite. It's also
a transducer. It takes this electric current with varying voltage
and converts it back into acoustic pressure and we get
the playback. That's an analog chain from start to finish. Now,
(15:24):
if you've got a good quality microphone and a good
amplifier and a good speaker, you can transmit sound pretty effectively.
And because the whole process is using that continuous and
varying signal, it is analogous to the experience of hearing
the sound itself. We've transformed the energy from one kind
to another, but apart from that, it is an unbroken chain. Now,
(15:49):
analog media includes stuff like magnetic tape and vinyl records,
which are produced in a way where you are transmitting
analog signals and they are effectively carved into a surface
that then can be picked up with a stylus on
a turntable and then converted back into an electric signal
(16:11):
that then can be sent to speakers. So either way
you are preserving that analog signal with magnetic tape. You've
got a recording device set up that takes that varying
electric signal from the recording and then creates a magnetic
field with the the writer the right head. Uh, And
you've got a little electro magnet in this thing, and
(16:33):
that magnetic field rearranges particles that aren't a strip of
plastic tape. That's how cassette tapes work. That's all VHS
tapes work. So attached to this strip of plastic that
is the actual tape in a tape, are these tiny
magnetic particles that are bound to that plastic. And by
(16:53):
applying the magnetic field to the tape, using in a
tiny electro magnet, you can change the direction that these
particles are facing on the tape itself. So this process
arranges particles on magnetic tape in a specific way to
record that original electric signal you were using. The magnetic
particles represent the original signal and then in turn represents
(17:15):
the sound that was used to generate the electric signal
during the recording process. So when you play a tape back,
the tape passes underneath an electro magnet at a distance
that's close enough that the electro magnet is picking up
the magnetic fields of all those tiny particles, and the
particles have been arranged in patterns because of that, you know,
(17:36):
recording process, right. So the fluctuating magnetic field that is
created because these particles are now passing by an electro
magnet are again reversing that process. The electro magnet starts
to generate an electric signal because of that magnetic field,
and then can go to an amplifier and then go
out to speakers. So again we use a lot of
transformational processes to record this sound, right, because you're in
(18:00):
this case, we took pressure waves, vibrations, The sound went
into a microphone, creates an electric current with varying voltage.
That electric current then goes to a tape recorder essentially
that uses magnetic fields to record onto tape. We take
that tape, we put that tape into a tape player,
(18:21):
and that magnetic record then produces an electric current in
our tape player, which goes to an amplifier and then
goes to drive speakers and replicate the sound that we
record in the first place. So again we transformed things
multiple times, but the analogous sound process has remained stable. Now,
(18:41):
there's a lot in this process that I have not covered.
The equipment and methods you use in recording and playback
determine whether or not the copy you have is a
really like accurate representation of the original sound like does
it sound like you were actually there? Or is the
nuance lost? And the same is true for a back.
Playback on a really sophisticated system will likely sound better
(19:04):
than one that's played on some super cheap stereo. Though
pretty quickly you do reach a point where the returns
are harder to detect, right like where you might listen
to something on a good system, and then you might
listen to that same thing on what's considered like the
highest of high end systems, and you might not be
able to tell a whole lot of difference. But the
(19:26):
basics for analog recording and playback are all there. Now.
When we come back, we'll talk about the digital approach,
but first let's take a quick break. Okay, So now,
we've got an idea of how the analog process of
recording and playback works. We transform stuff, but we still
(19:49):
have a continuous signal that represents sound, which is, you know,
a continuous phenomena as sound changes, as the pitch and
the frequency shifts, or as the volume changes, or as
different instruments or voices produced sounds. All those subtle and
maybe not so subtle shifts are part of that recording method.
It's an unbroken wave. Digital recording uses a different approach
(20:14):
in a way. Digital recording is like taking snapshots of
what is going on during a recording session. And I
thought of a kind of goofy analogy to sort of
explain what I mean. So imagine for a moment that
you are in a soundproofed room and you cannot hear
anything that's going on outside of this room. However, you
(20:34):
do have a little panel like almost like a hatch
in this room, and it happens to be facing a
really big orchestra pit, and the orchestra is playing. And
you know this because there's a light in the room
that lights up when the orchestra is playing. But you
can't hear anything because the rooms sound proved However, next
to the panel is a button, and if you press
the button, the panel opens up, but only for a
(20:55):
split second. Next to the panel, you have a table,
you get some paper, you got a pen, and your
job is to press the button, listen for that split second,
and then write down what you think is going on
in the orchestra. You know, like you could write down
everything from the specific instruments that you're hearing, the relative
(21:15):
volume of those instruments, any sort of harmonies you're hearing.
Maybe you're even just trying to play name that tune. Now,
let's say there's some other rules in place too. If
you push the button, you are not allowed to push
it again until five seconds have passed. So every five
seconds you get another instant of sound as the panel
opens and closes. This is that little snapshot of what's happening.
(21:39):
It would be really hard to accurately describe the music
because you wouldn't have a lot of information to go by, right,
you would just have this instant of sound every five seconds.
It might as well be noise at that point. But
then let's say we start to decrease the delay, where
you get to have the panel open so that you're
getting these instants is of sound more close together. As
(22:04):
that gets closer and closer, it will start to sound
more like uninterrupted music. Maybe we even rig up the button.
We tape down the button so it's always pressed down,
and the panel still has to open and close, but
it can open immediately after it shuts, so it's effectively
a shutter. At a fast enough rate, you wouldn't necessarily
even notice the shutters effect on the music. To you.
(22:26):
It would sound unbroken if it were fast enough, And
then you could accurately describe the music you could write down,
you know, depending on how quickly you can write, you
can write down a really accurate explanation of what is
going on with the music, or maybe you're just identifying
what pieces playing. But uh, you know, in this case,
if you've got that shutter going at a high enough rate,
(22:49):
it's almost like you're not in a soundproof room at all. Well,
this kind of is how digital recording works. So rather
than preserving an unbroken sign Knoll, the digital process breaks
up a signal into discrete units. It has to because digital,
when we get down to it, we're talking about binary
(23:10):
data zeros and ones. You cannot use zeros and ones
to uh to to do anything other than talk about
discrete units. It can't be a continuous thing. Now. As
I mentioned earlier in this episode, there are a lot
of quantifiable elements we can look at when it comes
to sound. We can describe how loud it is, or
(23:30):
what frequency or pitch it is. We can describe the
timbre or quality of the sound. That that kind of
gets us into areas that are a little less concrete
at least in human language and digital equipment like computers
are pretty good at handling things that are discrete and quantifiable.
This is the realm of computers. And remember, ultimately computers
(23:52):
are relying on those zeros and ones to describe everything.
Just to be clear, to get to this point, we
would need to use an analog to digital converter, but
I'm actually gonna circle round back to that later on.
For now, we're just going to focus on the basics
of digital recording because understanding that makes the whole you know,
a D C and d a C stuff way more
(24:14):
easy to understand. So, the way digital recording systems work
is that they take snapshots of a continuous wave. They're
measuring precisely all the elements at that moment in time
of the wave and it's or signal signal is probably
a better word than wave. Really, we're talking about the
(24:35):
electric signal generated as you're using a transducer to pick
up sound from you know, wherever. So in this way,
they're like that panel in that soundproof room. If the
sample rate is too low, if you are not sampling
the signal frequently enough, then you do not get an
accurate representation of that original signal. You know, You're you're
(24:59):
having to make a lot of guesses of what's happening
between each snapshot. Just like if you had a camera
and you were taking pictures of a fast moving, you know,
scenario in front of you. If the rate of which
you're taking pictures is pretty slow, you've got to make
a lot of interpretation of what happened between picture one
(25:19):
and picture two, and picture two in picture three. Same
thing with these digital recording systems. If you were to
try and play a recording like that back, it would
not sound very good because it would not be a
good representation of the original signal. So you need a
really fast sample rate to get an accurate representation of
what was really happening. This is the major difference between
(25:43):
analog and digital. Analog is continuous and unbroken. Digital is discreet.
But if you are using a very fast sample rate,
you can create a digital record of a continuous signal
that to human ears appears to be continuous itself. Again,
if that shutter is opening and closing fast enough, it's
(26:05):
almost like it's not even there. Now, let's imagine that
we've got two graphs that are showing the same signal, right,
and on the left side we've got the analog signal represented,
and on the rights that we've got the digital signal. Now,
let's say at first glance, these two are identical. They
both look like, you know, a typical sign wave. But
(26:28):
then you zoom into the analog representation. But no matter
how how far you zoom in, you see it's just
a continuous, unbroken line that's representing this this signe wave. Now,
let's say we take the digital one and we zoom
way in. Well, as we zoom wag in, we and
we get closer, we see that rather than being continuous,
it's actually a series of discrete moments, like almost like
(26:50):
steps or stairs. That's kind of what we're talking about here.
The question is how many stairs do we use? Like,
what's the resolution that we're using here. You can kind
of think of it like megapixels in a picture. If
you don't have a lot of megapixels, then you might
see some blockiness in a photo once you get to
(27:11):
a certain density, depending on you know, the size of
the image you're looking at, Like if you're looking at
on the side of a building, you're gonna need a
lot of megapixels so it doesn't look blocky. But depending
on that, uh, it may look really smooth. Same sort
of thing with sound. Now, if you've ever played with
digital audio recorders, you've probably seen something labeled sample rate
(27:33):
or project rate. This refers to the number of samples
that the recording is taking every second, and to record
a sound, that sample rate has to be fast enough
to take two samples within one wavelength of every sound
that's appearing in that in that recording. And remember I
said that it sounds wavelength is inversely proportional or has
(27:56):
an inverse proportional relationship to the sounds frequency, So the
higher frequency sounds have shorter wavelengths, and you do need
two samples per wavelength to capture the data necessary. To
have a recording of that sound. If the wavelength is
too small, then your sample rate will not be sufficient
to get all, you know, the full information about that
(28:17):
sound wave. You won't be able to record it, at
least not accurately. So remember I said the typical range
of human hearing is between twenty hurts to twenty killer
hurts or twenty thousand hurts. That's twenty thousand cycles per second,
and you have to gather two samples per wavelength or
or cycle. So that means you need a sampling rate
of at least forty thousand times per second or forty
(28:41):
killer hurts to be able to sample everything that's within
the typical human hearing range. Well, a basic sample rate
that a lot of people will use for various recording
projects is forty four point one killer hurts, uh, and
then they go up from there. In fact, we use
forty eight killer hurts when we're recording our episodes. I'm
using forty eight killer hurts right now. I had to
(29:02):
check because I did accidentally do forty four point one
for an episode a few weeks back, and Sary needed
to gently remind me that I need to fix that.
So it's a forty eight Killer Hurts. So I also
mentioned that we're quantifying all those elements about the sound.
We want as accurate a representation of the original sound
(29:23):
as possible. That means we're not just concerned with the
number of snapshots that we're taking every second. We're also
concerned with the quality of each of those snapshots. If
we were using a literal camera to take pictures, we
would want stuff like the lighting and the lens to
be perfect so that every single photo we got was
an accurate representation of what we were seeing when we
(29:45):
were there. Well, with digital recording, you know, we're not
talking about lights and cameras. We're talking about how much
data we're using to describe the original signal. This is
called bit depth. Bit depth refers to how many potential
values we can assign to a signal in an effort
to describe it. The more potential values we can use,
(30:07):
the more accurately we can describe the signal. So let's
do another analogy. All right, Let's say that we're in
a room. It's you and your best friend. Your best
friends all the way, I across the other side of
the room, and I hand you a picture. Your job
is to subscribe that picture to your best friend who's
across the room. Your best friend cannot see the picture,
(30:29):
They can only hear your description. Their job is to
try and recreate the picture, to draw it as you
describe it. However, I give you some more restrictions. I say,
you can only use five adjectives. Uh, you can only
use five sentences, and they have to be simple sentences.
They can't be compound or complex or anything. Five simple
short sentences with a maximum of five adjectives to describe
(30:52):
that picture. Well, chances are your best friend would draw
something that's kind of similar to the picture I gave you,
but it wouldn't be an accurate copy of it. Right.
You might be like, Okay, I can see where you
got that based upon the description. But let's say we
repeat this task, and each time we repeat it, I
give you a little more freedom and how you can
describe the picture you're looking at to your friend. So
(31:15):
you get to use more adjectives, you get to use
more complex sentences, and each time you're given a larger
set of potential values that you can express to your
best friend. Well, that's kind of like bit depths. If
you're using sixteen bit bit depths. That means you're using
sixteen bits to determine the range of values that can
describe the signal. So a bit is either a zero
(31:39):
or a one. With sixteen bits, you can represent up
to sixty five thousand, five hundred thirty six values. However,
let's say you were to go to thirty two bit,
so sixteen to thirty two, you would think, oh, you
could do twice as many. That's not that's not the case.
With thirty two bit depth, you wouldn't be talking about
twice as many as sixteen bit. With thirty two bits,
(32:01):
you would be able to describe up to four billion,
two million, nine d sixty seven thousand, two hundred nineties
six values. So the greater the bit depth, the more
accurately you can describe something. Essentially, uh So, it's both
the sample rate and bit depth together that can allow
(32:22):
a digital system to create a digital recording that represents
that continuous signal. It was sampling. Again, the digital recording
is not continuous. If we zoomed way in, we would
see it's a bunch of these little steps that are
all linked together. But if the sample rate is high
enough and the bit depth is great, enough, we can
reach a point where the human ear really can't discern
(32:45):
the difference. Does this mean at lower settings we would
actually notice a difference if you go low enough. Yeah,
but really, most of the time even sixteen bit is
sufficient for just plain old recording and playback. However, if
you want to work on a project. Let's say you're
an editor and you're you're trying to edit together music
(33:06):
files or audio files, larger bit depth gives you much
more space to work in without introducing stuff like distortion.
This is called headroom. And if you remember the character
Max Headroom, that name is a pun on this very
sort of thing. Technically, at the lower rates you get
deviations from the true sound. You're essentially inserting errors into
(33:31):
the digital file. Uh As you increase semple rate and
bit depth, you can decrease those errors until you reach
a point where any errors that exist are are impossible
to detect, at least with our natural equipment. Maybe you
could detect them if you had supersensitive electronic equipment to
indicate it, but it wouldn't be something that would be
(33:52):
necessarily perceptible to human ears. One other interesting thing, or
a couple of interesting things that I should mention with
sample rates. So I said, like, your sample rate has
to be fast enough to capture two points of data
along the wavelength of every sound, and for most of us,
that hearing range caps out at twenty killer hurts. That
might lead you to the question, well, why would you
(34:13):
bother to go higher than forty killer hurts? Now, if
twenty killer hurts is the limit of human hearing, typical
human hearing, why go to forty four point one? Well,
there are some other things that we need to think
about that play a factor in this. One of those
are harmonics. Uh Now, harmonics are way too complicated for
me to really fully get into in this episode. But
(34:36):
harmonics can actually exist above the range of human hearing
and yet still shape how we experience a sound. You
can almost think of it as the harmonics are sculpting
the sounds we hear. So even harmonics that are outside
of our hearing range might be affecting the sounds we
still can here. So we're not hearing the harmonics directly,
(34:58):
We're rather experiencing how they are affecting the rest of
the stuff we can perceive. If that makes sense, Well,
if you're sampling at a rate that's too low to
capture those harmonics. Those harmonics are not going to be
in the digital recording, so they won't be in the playback.
When you listen to it, you lose that sound. So
when you do listen back, you're gonna be losing those
(35:19):
effects and you're not going to experience the sound as
you would had you been in the place when it
was being recorded. Also, one thing that we can do
with recordings is we can change the pitch when we
record stuff. You know, like if you have a digital recording,
you can digitally change the pitch. In fact, Tari, if
(35:39):
you would like to digitally alter the pitch of my voice,
maybe increase the pitch so that I get that kind
of chipmunk sound to it. That's you know, boosting the
frequency up or maybe bringing that frequency way down and
giving me that deep, bass, booming voice that I know
(36:00):
I'll never have and I'll never be able to really
play like a baritone in a musical. Feel free to
do it. The world is your plaything. So you can
record audio with a sample rate of forty four point
one killer hurts. Then on playback, maybe you decide you
want to pitch everything down well, you'll hit a ceiling
(36:24):
of the sounds that you'll have in that recording once
you get to killer hurts or so. So, if there
were sounds that were above killer hurts, you're not really
going to be able to hear them with the pitched
down recording. Remember that pitch down recording will bring stuff
that is outside human hearing into the range of human
hearing because you've pitched it down. But if your sample
(36:46):
rate is too slow, too low, in other words, you
won't have captured those higher pitches. So let's say that
you're recording something that's in a very very high frequency,
like beyond the range of human hearing. But then you
want to do a pitch adjustment so that people can
actually hear a sound, even though you know normally they
(37:08):
wouldn't be able to hear it at all because it
would be outside their range. Maybe you're doing like a
nature documentary and there's a critter that makes sounds that
typically we cannot hear, but by pitching it down, you
can say this is what it sounds like once we
reduce the pitch. Well, you have to have a sample
rate that's high enough so that you capture that range
of sound in the first place. Right, So that's one
(37:30):
reason why you might want a very high sample rate.
I just thought that was neat all. Right, we need
to take another break. When we come back, we'll talk
about the process we need to follow in order to
go from analog to digital and back again. It's gonna
be a lot of us talking about some of the
stuff we just chatted about, and we'll also talk about
audio files a little bit. But first let's take another
(37:51):
quick break. Now, before I dive into the converter's part,
I should add there are some outliers, right There are
digital microphones. For example. Now there's some digital microphones that
are analog at the front end, so in other words,
(38:13):
they still have the diaphragm, they still have the electromagnet,
they're still generating an electric current with varying voltage. But
then they'll have an analog to digital converter built into
the microphone itself. So you have an A D C
and it's right there in the device, and then you
have the signal go to other elements of your recording studio.
(38:37):
There are other digital microphones that use the pressure waves
to move elements that immediately convert into digital data, getting
into that is pretty complicated. They are not super common.
It's not like that's the type of microphone that everyone
is using. Um, they're important, but you could argue that
(38:59):
it's a microphon own and a d C all in
one because you're taking audio, which is an analog you know, signal,
and you're converting it immediately into binary or digital information.
But we're really going to talk about analog to digital
and digital to analog, which is what most equipment is
(39:21):
dealing with. When we're speaking about this kind of stuff,
We're not gonna worry about stuff that's native digital because
it's just it's not that common. Um like digital speakers
are a different thing altogether as well, and um yeah,
we're just gonna wipe those out. We're gonna look at
what most people use, which is that you know, your
(39:42):
typical stereo system or your typical audio recording setup. So again,
typically the end equipment that you use to either record
or listen to audio, the stuff at the very ends
of that chain are typically analog in nature. Again, there
are outliers, but for the vast majority of cases, we're
(40:03):
talking about an analog device that generates an analog signal
or plays back an analog signal. So we take an
analog phenomena, the pressure waves that make up sound. We
feed that through a transducer to create a different but
still analog signal, in this case, an electric current with
variable voltage. But now we get to a point where
(40:25):
we say, all right, we want to transform that into
a digital file that quantifies this signal. When we play
the digital file back, that signal ultimately needs to go
through some kind of loud speaker for us to hear it.
Maybe that loud speakers in our headphones, maybe it's a
stereo system, maybe it's you know, the speaker on your smartphone.
(40:48):
Maybe it's a sound system in a stadium. But we
need a way to transform that digital information, all those
zeros and ones into an electric signal with variable voltage,
and we probably have to amplify that nal so that
it's strong enough to drive whatever speakers were using to
create the sound, which again we experience as an analog phenomena. Now,
(41:08):
if there was some way that we can interface directly
with machines and have those digital signals interact with our brains,
maybe we wouldn't need to do this kind of transformation.
But as it stands we do have to do this,
and this is where converters come into play. The converters
could be standalone devices, or frequently they're worked into the
(41:29):
design of various pieces of equipment. So for example, a
USB microphone, if you have one of those that you
plug into your computer, like I'm using one right now
to record this, they have that a d C converter
built into them. And that I'm being repetitive because that's
analog to digital converter. And then I said a DC converter.
(41:50):
It's like saying a t M machine. Anyway, the microphone
still acts just as a traditional analog mic on that end,
but then the electric signal has to go through a
converter converts into a digital signal, and that's what transmits
through the USB cable too, you know, whatever you got
hooked up to, like in my case, it's my work laptop.
Now here's the thing. There's more than one way to
(42:13):
convert an analog signal into a digital one. All of
these ways get pretty technical talking about the way it's sampled,
the way it ends up taking these measurements of the signal. So,
for example, with analog to digital converters or a d
c s. There are several popular methodologies, but generally speaking,
(42:35):
they all do the same thing on a big picture scale.
They all sample a signal. This is the snapshots that
I was talking about earlier. They look at a signal
and a specific frequency, like a specific They look at
the signal a specific number of times every second, and
they quantified the signal. They measure the signal, which determines
(42:57):
the resolution that you get of the signal. Obviously, if
you want high quality sound, you need both a good
sample rate and a good resolution, which we can think
of as you know, the accuracy in capturing the nature
of that signal. You can think of it as an
A d C is measuring the electric current many many
times per second and quantifies that measurement as digital data.
(43:21):
And it's not just like how important is this signal
at this specific moment in time, but also how important
are the changes in that signal over greater lengths of time. Now,
the bit depth we can think of is how detailed
these measurements can be. So the number of measurements and
the detail we get together determine the quality of the
(43:43):
digital signal compared to the original analog signal. And again
we're talking about digitally describing an electric current. At this point,
we're not talking about describing the sound necessarily. We're describing
the electric current that the transducer created after the sound
went through the transducer. Now, if the sample rate of
(44:04):
an A d C is too low, you get what's
called alias sing. Now, this means that the digital signal
will differ greatly from the original signal. Uh. And that
means that you're not going to have a good representation
of what was originally creating that signal in the first place,
in this case, whatever the sound was. UH. So that
that's what alias sing means in this context. Now, a
(44:28):
A DOCK or d a C is a digital to
audio converter, and it's basically the same thing we just
talked about, but in reverse. The d a C takes
digital information, which essentially is describing an analog signal an
electric current of variable voltage. Then it produces that analog signal.
(44:48):
The way it does again depends upon the type of
d A C. Just as A d C s have
different methodologies, so do d a C s. H. I
might do an episode that goes into more detail, like
I mentioned at the top of this episode, But honestly,
once you really start diving in there, it gets incredibly
technical very quickly. Generally speaking, we're talking about sophisticated circuit
(45:13):
boards that are designed to convert digital to analog or
vice versa, to switch between the data made up of
zeros and ones and a continuous electric signal. And again,
if there's interest, I'll go into more about how that works,
but believe me, it gets really complicated, and without visual
aids it's really hard to kind of get it across. Anyway,
(45:37):
Now let's talk about audio files. Also, I should mention
there's a ton of stuff I did not talk about, right,
I didn't talk about multiplexing or anything like that, So
there is a lot more to it than just the
general information I'm giving anyway. Audio files. So, back in
the day when c d s were fairly new, there
were audio files who just despised digital media. The process
(46:00):
us of converting an analog signal into a digital file
and then back again to analog. Well, that represented a
potential loss in quality, right, the playback experience might not
be as vibrant. Audio files typically use words like warm
or full to describe sound. These are words that are
(46:20):
hard to quantify they are experiential, I guess, and they
would lament that digitization removed some of those elements from recordings.
The thing is, depending upon how you're digitally sampling a signal,
some of that could be actually happening. You could be
losing harmonics. And this isn't even touching on the issue
(46:40):
that you start getting if you're if you're doing stuff
like compression file compression in this sense. I'll talk about
audio compression in a bit, but file compression can involve
using what are called lossy formats. A lossy format discards
part of a digital file that describe a signal, and
(47:01):
typically the way it does this is that the encoding
process is getting rid of information that it deems as
being irrelevant. So let me explain that last bit. I
did a full series of episodes about MP three's that
goes into this into far more detail, but i'll give
it down in dirty version for this episode. So, the
MP three method of compressing a file takes a psycho
(47:25):
acoustic approach in part when figuring out how to make
an audio file size smaller, because raw audio files can
be huge if you're really using a very high sample
rate and a big bit depth. During your recording process,
you're generating enormous files, right because the system is taking
(47:48):
data many many many times, many thousands of times every
second and using an enormous amount of information to try
and describe that signal each time, every single snapshot. That's
a lot of information and that isn't really convenient if
you want to store that file on like an old
MP three player. You know, if you remember those where
(48:11):
you had to like in the old old days, you
had to connect them physically to your computer. You would
download or rip music and you would then send that
music file to your device. These devices had very limited
storage space on them, so you couldn't really hold a
lot of raw audio. Like a single file might end
up taking up the entire storage on your m P
(48:33):
three player. And maybe you really like journeys Don't stop Believing,
but you might want some other songs on there too.
This is also more complicated if you want to do
something like stream music. You don't want to have enormous
files that would require like a gigabit Internet connection in
order to be able to stream it, So you have
to have a way to compress files down to sizes
(48:55):
that are easier to handle well. The way the MP
three algorithm does this is that once you set some
general parameters, like you decide how compressed you want to
make this file, essentially you're telling the MP three algorithm
how hard it needs to go. When it's starting to
cut stuff, well, then the algorithm begins to toss out
(49:16):
data that, at least in theory, should not affect your
experience when you listen back to the audio playback. So,
for example, let's say you've got a sound file and
in that sound file you have a very soft sound
that immediately follows a very loud sound. So loud sound happens,
soft sound happens immediately after that. Well, you wouldn't actually
(49:38):
hear that really soft sound, and that's just because of
how our ears work. The loud sound effectively masks the
softer one, so it's almost like the soft one didn't
exist at all. Well, if it's like the soft one
didn't exist, then there's no reason to keep it right.
If you couldn't hear it anyway, there's no reason that
(49:58):
that should be in uh the file, right, So the
algorithm effectively, through analyzing this data says, ah, that doesn't
need to be in there. No one would hear it,
so it tosses the data out. That's why it's a
lossy file format. The same goes for frequencies that would
be outside the range of human hearing. The logic is, well,
(50:20):
you can't hear something that's at killer hurts, so we're
just gonna get rid of anything that's occurring at that
frequency because there's no reason to keep it. However, depending
on how much you want to compress that file, those
cuts can really start to affect the quality of the
playback audio when you put it back through you know,
a decode er and you get the audio on the
(50:42):
other end. By the way, as I mentioned, file compression
is not the same thing as audio compression. I'll explain
what I mean by that, but first let's take one
last break. Okay, before the break, I said that audio
(51:04):
compression and file compression are two different things. It does
get confusing, and I myself have been guilty of kind
of interchanging the words or not clarifying enough while talking
about compression and uh, thus I have been guilty of
confusing it even more so. My apologies for that, but
let's get to it. Audio compression refers to reducing the
(51:25):
dynamic range of volume in a recording. Uh So, in
other words, it's about reducing the volume distance between the
softest sounds and the loudest sounds. Now, this can be
really important for certain types of recording. I'll give you
an example that I frequently run into that drives me nuts.
And this happens a lot with like streaming media for me,
(51:49):
so movies and television. Have you ever watched like an
action movie where you can barely hear some of the dialogue,
especially if people are speaking in like low voices and
you know they're trying to be secretive or whatever. And
then so you turn the volume up so that you
can hear what people are saying. But then the next
time something explodes, you're worried that you've just destroyed all
(52:09):
your speakers, or maybe you've caused yourself permanent hearing damage.
This happens to me all the time, where the softest
sounds and the loudest sounds are so far apart that
there's no comfortable volume. I can select where I can
hear everything and not feel like one I'm missing out
on some dialogue, or two my neighbors are going to
(52:31):
come over and complain that I've got my volume turned
up too loud. So compression in a case like that
can narrow the gap between the softest parts and the
loudest parts so that you can find that kind of
comfortable volume where you can hear everything. However, going overboard
with audio compression will reduce the dynamic range in a
(52:53):
recorded piece of audio, and if you do that too much,
it can make the audio sound flat and uninteresting, where
everything is just coming out at exactly the same volume.
If there's no real volume range, then your ears just
kind of get tired of hearing everything played back at
that same level. Some digital recordings really suffered from this
(53:16):
kind of processing, Like there was an era of music
where audio files in particular were really complaining that everything
that was being laid down had so much compression in
it that there was no real dynamic range and audio,
and it just meant that the music wasn't as interesting,
like there wasn't there wasn't enough variation, and it makes
(53:39):
music kind of boring. Uh, it wouldn't matter if you
had an analog pressing of a digital recording session, because
analog does not magically fix the problems of the recording process.
So if you're recording digitally, and then you make a
vinyl record pressing of that digital recording. I mean, all
that processing you did on the digital side, that's still
(54:02):
going to be part of what ends up being recorded
on the vinyl. It's it's not like vinyl suddenly cures
all sins of digital. So even if you were to
to go with analog audio media, you would still have
the same problems that were introduced in the digital processing. Now,
(54:24):
this does not mean that all digital to audio is
inherently flawed. Even if we just look at the analog chain,
we have to acknowledge that the process of recording in
playback means, you know, you're taking pressure waves of the
original sound, you pass them through a system that converts
those pressure waves into an analog electric signal, and then
(54:45):
you've got to reverse that process during playback, and stuff
can happen along that pathway that could affect either the
recording process or the playback process or both. So, in
other words, analog does not necessarily mean better, because flaws
can exist in the analog approach just as they can
with the digital approach. And there are other elements as well,
(55:08):
such as low level noise. Analog systems can introduce a
low level noise into a signal. Digital avoids that. Now,
that does not mean that digital is better, mind you,
because there are other ways to reduce and eliminate noise
and analog systems and digital can introduce other artifacts that
didn't exist in the original signal, and then that comes
(55:30):
across as like errors in your playback, Like you might
hear some weird blip noise and think, what the heck
was that, and it wasn't necessarily present in the original
recording session, but was introduced as a digital artifact. This
is just another example of how one format is not
necessarily superior to the other. It depends on way too
(55:52):
many other factors. They're just different. And honestly, I'm fairly
confident that if you were to do a double blind test,
and just in case you're unfamiliar with that term, double
blind is a type of scientific test where neither the
subject that's going through the test nor the person who
is in charge of administering the test knows which version
(56:18):
anyone is getting. So if you have a control group,
the person administering the test doesn't know if that's a
control group or if it's an actual test group. That
they're testing at any given time. That way, the person
administering the test does not give bias to the person
who's experiencing the test. The thought is, if I know
as an administrator, then I might give a tell to
(56:42):
the test subject. Right. So let's say it's a double
blind test and the audio files are going into a
room and they're going to experience the same piece of
audio recording, but it's over different systems. So like some
of the pieces of the same stretch of audio are
analog sources, some of them are digital sources. They might
(57:04):
even include different like systems like premium systems, like super
super high end systems that cost maybe upwards of hundreds
of thousands of dollars, and maybe just on some that
are really good systems, like you know, they're still expensive.
Maybe it's a few thousand dollars, but they're not, you know,
monumentally expensive. My bet is that most audio files would
(57:25):
have trouble picking out which ones are analog systems versus
digital systems unless there's some giveaway, like if you hear
the scratch of a needle hitting record, then that's kind
of a dead giveaway. But let's say you know you're
you're talking about, like the highest of high ends. I
don't think they would be able to tell the difference
very easily. And that's because our approach to digital processing
(57:48):
has become sophisticated enough that to our human ears, it's
pretty close to an analog signal. And that you know. Also,
I want to mention the returns on those high end
audio equipment, like the differences that you start to see
when you're really hitting that upper echelon of audio equipment.
(58:10):
Some of those returns are so minor that after you
reach a certain point, they are largely meaningless. Like like,
as far as perception goes, you wouldn't be able to
tell the difference. Uh, And for people like me, people
who have had some hearing loss, it matters even less
than that, right because like for me, like you could
(58:30):
almost say, it's like I have an unsophisticated palette, Like
you could serve me an amazing meal, but I'm not
likely to notice it being any better than you know,
a cheeseburger. But again, this is my hypothesis. I I
believe this is probably true. It is entirely possible, and
I admit this that if I actually were to conduct
(58:54):
this kind of study, I might find that I'm totally
wrong that the audio files are like no, that is
clearly the premium, and maybe the differences are subtle, but
maybe they're detectable. Right, It could be that I'm wrong
about that. Uh. I just think that there gets to
be a point where people start to buy into a philosophy,
(59:15):
especially with audio files that isn't necessarily supportable by you know,
quantifiable evidence. It becomes so subjective that once you remove
the subjective element, like you remove the ability for them
to know whether or not they're listening to their preferred
set up, that it starts to disappear. Maybe I'm wrong
about that. I don't think I am, But that's the
(59:38):
overview of analog and digital and why you have to
have the converters. As I said, to get into specifics
would take more time, you know, talking about delta Sigma
processing and that kind of stuff. But if you want it,
let me know and I will try and put that
episode together. It will just be far more niche oriented
than even this one was. B It is a fascinating subject,
(01:00:02):
like there's some really cool technology that goes into making
this all work, and the fact that that technology does
work and that it has become so sophisticated is why
I feel pretty confident in saying that with a sufficiently
good system, you wouldn't be able to tell the difference.
But that's it for this episode. If you would like
(01:00:23):
me to cover any kind of topic, whatever it might
be in the tech world, let me know. The best
way to get in touch is on Twitter. The handle
for the show is tech Stuff h SW and I
greatly appreciate it. I'm getting some wonderful suggestions. Really makes
my job easier because I know exactly what people want
to hear um. So yeah, reach out and let me
(01:00:43):
know what you think and I'll talk to you again
really soon. Tech Stuff is an I Heart Radio production.
For more podcasts from my Heart Radio, visit the i
Heart Radio app, Apple pine Casts, or wherever you listen
to your favorite shows.