On a larger scale, a human whisper can be around 20 to 30 decibels, while a typical conversation is around 60 decibels. A motorcycle goes through about 90 decibels, a really loud concert can be in the 110 decibel range. Much higher than that, and you get into a range where the voice becomes physically painful. But if that’s the case, why would any recording device need more than the 144.5 dB dynamic range of 24-bit audio?
set level (or not)
The crazy dynamic range of 32-bit floating audio is the source of the claim that you never need to set levels, although it’s a bit more complicated than it seems. The highest level the device can record is called 0 dBFS (FS here stands for “full scale”).Anything above this will be clipped, which is why it sounds distorted when YouTubers scream like that.
Now, usually you would set the audio level when setting up the device to avoid hitting that limit. Setting these levels requires applying gain to the signal from the microphone, an irreversible step that destroys the dynamic range of even 24-bit recordings.
“When you’re recording sound live, you usually apply gain. Some recorders have gain between 30 and 90 decibels,” explains Pereira. “It’s great when you have a quiet scene of two people whispering. So you can turn up the dial on the recorder, say, plus 60 [decibels]. So now when someone decides to yell between 60 and 145, the dynamic range isn’t that great. “
On the other hand, for 32-bit floating point recording, no gain needs to be applied before recording. “When you record in 32-bit, there is no volume knob, it’s essentially just creating a mathematical graph of the data that can then be interpolated in post-production,” Pereira said.
Setting levels can be tricky on 24-bit systems due to the noise floor. Simply put, no matter how quiet you set your recording space, there will always be some noise from background objects, or even the electronics you’re recording from. Adding gain to your signal while recording amplifies noise as well as your audio source, and once it’s baked into the recording, it’s there forever.
32-bit floating point recordings allow for greater flexibility to adjust after the fact (in some cases it may even Help with low-level noise issues). That said, it’s important not to let this give filmmakers and sound producers a false sense of security. “It’s not going to solve the inherent problems of where you’re shooting. For example, recording in 32-bit isn’t going to eliminate noise if there’s AC or a big fan nearby,” Pereira said.
It’s always important to manage live noise and make sure the microphone is capturing the signal correctly, but once your equipment is set up correctly, being able to capture audio even beyond the peak point is a useful tool. But don’t expect to be left with 24-bit audio forever.
link in chain
So, if 32-bit is so good, why isn’t it the default? For starters, many steps of production—including editing, mixing, and especially distribution—will use a 24-bit workflow, which means extra data will be lost at some point. And the audio engineer will need to make adjustments at some stage to make sure the audio signal doesn’t get clipped when downsampling to 24-bit, like when the levels weren’t set correctly during the initial recording.
Essentially, this means that work originally done on set is moved into post-production. So you have a choice: either set the levels correctly on set and record directly in 24-bit, or record in 32-bit and add an extra step later. Either way, it’s a step you have to do, and some would argue that you’re better off doing it on set.