[Music] Is there an audio quality difference
between 24-bit linear and 32-bit float audio recordings when you're doing spoken word audio? Well, let's listen to some samples and see
what we can learn here. [Sample Audio] We'll talk more about those
samples in just a minute, but first of all, what is 32-bit float recording? Well, first of all, I think it's kind of an
unfortunate name that a lot of product companies that make audio recorders have started using
to describe their technology. 32-bit float is part of the technology. It is basically the format in which they save
the audio or encode the audio into a WAV file, but it isn't the full description. So really, what 32-bit float is, it's just
a way to record a wider dynamic range. And dynamic range is, of course, the range
between the quietest sounds and the loudest sounds. That's all that 32-bit float enables, but
it requires more than that to really get the most out of it, to really record wide dynamic
range. As an example, think about soft rain falling
on asphalt and then a lightning bolt striking and you hear the thunder. That would be a very, very wide dynamic range
of sound. And the beauty is that 32-bit float containers
or files or encoding schemes can record really, really wide dynamic range like that. But to do that, there are a couple of things. Number one is the microphone. The microphone has to be able to handle as
wide a dynamic range as well. So it's really great that these recorders
can handle a wider dynamic range and encode them in 32-bit float. But if you don't have a microphone that can
handle the sound pressure levels or the wide dynamic range of what you're trying to capture,
then you're still going to get some distortion. But it's going to happen at the microphone
before it ever gets to the recorder. So that's the first thing. Secondly, there are some other things that
are necessary to get wide dynamic range recordings. And typically what these wide dynamic range
recorders do, something like a Zoom F6 or a Zoom F8n Pro or a SoundDevices MixPre 2
series recorder, they have multiple analog to digital converters. So just very quickly, the microphone captures
audio and it creates an analog electrical signal that goes into your recorder. And then the recorder converts that electrical
signal. Well, first of all, before it does that, it
amplifies the signal using a preamplifier, makes it louder, and then it converts it into
digital from analog. So to capture this very wide dynamic range,
what recorders, these wide dynamic range recorders do is they have multiple analog to digital
converters per input. And so what those converters do is one of
them, for example, and they can have two, they can have three, they can have even more
than that. But one of them, as an example, will handle
the very, very loud sounds. And the other one will handle the quieter
sounds, and then it will bring those back together and encode them into a 32 bit float
space. So that's how these work. And those are all the different ingredients
that have to be there. So it's really important you need to understand
number one, which microphone you use makes a big, big difference. Number two, the placement of the microphone
matters a lot. And so I remember when 32 bit float recorders
first came out, people were saying, oh, a bunch of people are going to lose their jobs. These production sound mixers aren't going
to be needed anymore. I don't think that's the case at all. I think it's just a tool that makes their
jobs potentially easier. And honestly, most production sound mixers
aren't using 32 bit float recorders normally for capturing dialogue, but they can. But in any case, getting the right microphone
that can handle the sound pressure levels and the dynamic range you're trying to capture
is one important factor, and having this wide dynamic range recording capability on your
recorder is another factor. So it's not quite so simple. And honestly, I wish they'd call them wide
dynamic range audio recorders as opposed to 32 bit float recorders, because you see it's
not just 32 bit float, it also requires multiple analog to digital converters. So that's another factor. Now, the interesting thing about 32 bit float
as well is that 32 bit float, while it can manage a much wider dynamic range, it can-
it can provide a much wider range of values to store the audio in, it's very much like
scientific notation or using exponents. So you can represent very large or very small
numbers. However, when you do that, at some point you
lose precision. And there are potential precision errors. And so the big question for me is, using 32
bit float to encode your audio, does that reduce the audio quality of what you're hearing? And from the samples that we heard here for
spoken word audio, I honestly think that we're not really pushing the limits of the microphone
or the recorder that much that it makes a difference. I think for sound effects, the answer may
be entirely different. But for spoken word audio, and what we did
specifically for those 32 bit float recordings is we intentionally set the gain or the input
level- or in the case of the Zoom F6, we set the fader- so that it would intentionally
get the audio above zero dB full scale, which is kind of like the big magic of 32 bit float
audio, when you first see a demonstration of this, when there's audio that hits up against
zero dB, and that you can actually reduce the level of the audio, and you can actually
restore the quality of the audio, that seems like magic to people. And indeed, it's something we're not used
to. And it's- it's really, it is, I guess, kind
of magical. But if you understand it, it makes sense how
it's possible. In my experience, listening to those samples
and with other samples that I've listened to, I think what matters more... I don't- I don't really feel like the 32 bit
float recordings were substantially worse than the 24 bit, I don't think the 24 bit
recordings were substantially better than the 32 bit float recordings. But I think what is a bigger difference is
the choice of microphone that you use, where you place the microphone, and that you're
using a quality recorder that has good preamplifiers, quality preamplifiers, and quality analog
to digital converters. Of all those samples, the only one that sounded
really kind of substantially different to me was the- was especially when we recorded
with the Rode NT1 5th generation, which is an XLR and USB microphone. And incidentally, we used the same microphone
through all of those samples intentionally, so the analog front end of that microphone
is the same through all those samples. What I noticed is that when we recorded with
the USB connection on the NT1 5th generation, that means we're actually using a preamplifier
and analog to digital converters inside the microphone. And in that case, it sounded not quite as
smooth as the others to me, but I think that's- that's not so much an issue of 32 bit float
versus 24 bit linear. That's more an issue of the preamplifier and
the analog to digital converters that they're using. So all of this to say, I think for spoken
word audio 32 bit float is fine. I don't think it's a... some people call it a game changer. It hasn't changed my game much. I still record mostly in 24 bit because I'm
mostly recording things that are scripted and/or in controlled environments where I
know what's coming. But if you're recording a, some sort of situation
where you've given the actor full freedom to do whatever they want and it's unscripted
in some way, and maybe, or maybe you're working on a horror film where there are going to
be whispers to screams all in the same take. Those are cases where it may be useful, but
I think it's important to keep in mind you will also have to do work in post to bring
that into a range that actually can be mastered and sent out in your final video. So those are some things to keep in mind as
well. But in short, I think it's fine to use 32
bit float for spoken word audio for your own projects. If you're working on a project with other
people involved, make sure that they can handle 32 bit float in the overall workflow in the
pipeline so that you're not giving them something they can't work with. And if you can't afford a 32 bit float recorder,
a quality 32 bit float recorder, you're going to be fine. Your audio is not going to be worse for it. Just learn how to set your input level, your
gain, and you'll be fine. I think one thing to consider as well, there
are a bunch of these body pack recorders on the market that are 32 bit float capable as
well. So they're wide dynamic range recorders, but
they come bundled with a lavalier microphone. A few things about lavalier microphones. First of all, the microphones that are bundled
with those kits usually are not very good. They're, they're okay, but they're not amazing. And secondly, lavalier microphones by nature
and by design are not generally capable of handling as wide a dynamic range as a more
capable microphone. So for example, like the Rode NT1 5th generation
here, for example. So that's an important thing to keep in mind. If you're expecting that you can go out and
record jet engines to softly patterning rain on a piece of concrete somewhere with that
lavalier microphone, maybe, maybe not. I also noticed that they- most of those microphones
sound a little bit brittle. So just some things to keep in mind there
as well. I think beyond the consideration of whether
32 bit float or 24 bit linear sounds better as a recording, there are a lot of other factors
that matter. So as a final word here, get out there, make
some great sound. I don't care whether you're doing it in 32
bit float or 24 bit. Just get out there, get your message out and
have a good time doing it. We'll talk to you soon.
