The difference in audio file quality, between a 24 bit fixed point WAV and a 32 bit floating point file is negligible. 32-Bit Floating does however, have a lot of headroom, that lets you make mistakes during recording, without experiencing any unwanted clipping.
Which is Better 16 bit, 24 bit, or 32 bit?
To understand how different types of audio files compare to each other, it's necessary to know how each type of audio file differs from one another.
The main difference between 16, 24 and 32 bit float wav files is dynamic range, or, the difference between the loudest and quietest possible parts of the signal.
Let's go through these one by one! First of all, 16-bit audio.
16-bit WAV files store uncompressed audio samples, each of which is represented by a binary number, with a total of 16 digits.
Unlike our 32-bit float file in question, 16-bit and 24-bit files will have a “fixed point” numbering.
Simply put, it means that within your 24-bits, we are working with whole integers, rather than numbers with decimal points.
These 16-bit numeric values essentially represent a voltage level, that corresponds to signal amplitude. The Maximum signal amplitude or loudness within a 16-bit framework is 65,536, while the lowest or quietest value is represented by a 0.
Basically, this means our entire signal amplitude has a resolution of 65,536 levels.
Running this through some calculations to convert it into dB shows us, that the max level is at 0 dBFS, while the noise floor, or lowest level is at -96.3 dBFS.
This shows us the total maximum dynamic range of a 16-bit WAV file, which is 96.3dB.
24-bit files are easily the most widely-utilized files in professional audio, today.
Just like a 16-bit wav file, 24-bit files are also represented with fixed point values.
24-Bit however, improves on 16-bits, by adding 50% (or 8 more bits) to your 24 binary digit word. With more bits, the range of individual signal voltage levels grows exponentially.
With 24-Bit Audio Files, you have a total of 16,777,215 Levels of signal amplitude, with 0 being the lowest value, and 16,777,215 being the loudest.
Running this through the same maths, essentially leave us with 0 dBFS as the maximum volume, while the noise floor drops down to -144.5 dBFS.
From this we can see, that the maximum dynamic range of our 24-Bit WAV file is 144.5 dB.
In programming, a floating point value is a decimal number, that lies between two integers.
Imagine if you have 24-Bits, you have 24 possible values of 1 or 0. With 32-bit Float, you also have the decimal values between 1 and 0, which end up creating some massive numbers
Unlike 16 and 24-bit recorded audio, where the bits represent actual values, 32-bit float uses it's bits in a different manner. 32-bit float is more complex, since it's meant more for computers, rather than human eyes.
The first bit in your word indicates whether the value is positive or negative, the next 8 bits are to indicate the exponent of your value, while the last 23 bits are your mantissa (the specific floating-point digits, in scientific notation).
These end up so unfathomably big, that if we were to write out the number of total possible amplitude levels, this number would have over 70 zeroes, so it has to be written as an exponent.
That being said, 32-bit float files don't have their maximum volume at 0dB. Running the numbers through the same formula, we're left with a maximum dB value of 770 dBFS, while the noise floor is at -758 dBFS
This gives us a total dynamic range of 1528 dB, over 10 times the range of a 24-Bit file.
While this won't make much difference in the quality of your audio, but what you get as a result of this massive dynamic range, is a sh*t load of headroom.
Since 32-Bit Values have a maximum value that is above 0 dB, and most DAW's set their max loudness at 0 dB, out 32-bit recordings are lined up with the 0 point of a 24-Bit or 16-Bit file.
This means that the loudness of the three are identical, but the 32-Bit file has an additional 770 dB of headroom on top. We'll talk about how this is helpful, shortly!
What Difference Does 32 bit Make To My Audio Quality?
In terms of the actual quality of sound, the difference between a fixed point 24-bit bitrate and 32-bit floating point is negligible, if it exists at all.
The main improvements come with the increased dynamic range when recording instruments, or using 32-bit files for sound design.
Nobody needs 770 dB of loudness. The largest volume disparity of anything on earth is around 210 dB, so the massive 1528 dB of dynamic range, is comically over-the-top, and as a result extremely useful when you don't want clipping.
In terms of pure audio quality, choosing 32-Bit float over 24-bit, is pretty much redundant.
Why 32 bit Float Is Great For Recording
While your masters will have no benefit from being in 32bit, over 24-bit – when recording, 32-bit float is the way to go.
Imagine you're recording an extremely loud sound source. If you record in 24-Bit, everything above 0dB gets cut off or clipped, which leads to hard clipping artefacts, that can be very unpleasant.
If you record in 32-Bit, none of the information you record will get clipped at any point, unless you're recording at dB levels, that are literally loud enough to create black-holes.
With the theoretical maximum sustained loudness being at 194dB, you can start to see how much headroom you have, when recording in 32-Bit.
Since your DAW will automatically max out at 0dB, if you just import a very loud track into your DAW, it will still be clipping. Where this added headroom and massive dynamic range come in handy however, is, when scaling your audio.
Say, your loud recording is so loud, that you're getting a lot of unwanted distortion.
If you recorded in 24-bits, anything above 0 dBFS in your signal, gets clipped during recording and you can't bring any of the clipped audio back.
With 32-bit float files however, nothing above 0 dBFS gets clipped, so when you go to reduce the gain, you bring back all of the signal that would have clipped otherwise.
This goes both ways and you can increase gain just as much as decrease it with 32-Bit audio, without worrying about any clipping or other distortion.
Essentially, if you're recording in 32-bit Floating point files, you don't have to worry about any digital clipping. This means that you don't lose anything if you make mistakes with recording gain.
The key word to what 32 bit floating point wav files are all about is: flexibility.
You're not getting any increased audio fidelity, or a magically better sounding audio track. However, what you do get with 32 bit floating files (in contrast to 12 or 24 bit fixed point values), is the option to make mistakes.
No matter how good a producer you are, everyone makes mistakes. And, using 32-bit float when recording, lets you make these mistakes, without any lasting impact.
That being said, as long as you check your levels before recording, recording with 24-bit audio is just as good when talking about audio quality.
When you record in 32 bit floating digital audio file format WAVS, the only positive you're getting is that dynamic range and headroom. It has little to no impact on your final product, which will be exported in 24-bit tracks, so don't worry about it too much.
The difference in audio file quality, between a 24 bit fixed point WAV and a 32 bit floating point file is negligble, if it exists at all. What 32-Bit Floating has going for it however, is a ludicrous amount of headroom, that let's you make mistakes during recording, without experiencing any unwanted clipping.
Hopefully we've cleared up some of these, obviously complex concepts for you. While the difference between 16, 24 and 32-bit float audio files can be negligible most of the times, it's necessary to know what the differences are, so you can make a more educated choice.
Whether you're interested in the technology, or just want to record better digital audio files, we hope this article has been helpful to you.