Sound Science Corner Pub

[gregorio]

The topic was extra dynamic range for editing-in the case of audio, trying to raise the sound level in a quiet passage and not get noise. Or not to get clipping in the loudest passage.

The issue is the dynamic range of the recording, not the available ADC dynamic range. In difficult conditions the dynamic range of the recording might only be 20dB or so (filming in a busy city centre for example), 40dB is typical, the most I’ve ever seen is about 70dB. So even in this most extreme case, with the addition of a relatively huge 30dB of headroom, that would put the noise floor of the recording at -100dB which is still roughly 20dB higher than the noise floor of a decent (24bit) ADC. When raising the level in post, it’s the noise floor of the recording that’s always going to be the issue, not the noise floor of the ADC.

With your profession, and that you are actively monitoring microphone levels, I can understand why you're not the intended consumer. Rather "content creators" as it's known as YouTube creators.

I’m not a production sound mixer but a competent one won’t really need to monitor mic levels, they will have a rough idea of the likely peak levels and set their gain to be 20dB or so lower than that, which will cover any eventuality, so it’s pretty much fire and forget. They will still likely monitor the input levels just to be sure but they will be adjusting the output (mix) levels of the mics rather than the recording levels. The issue of clipping is virtually always a case of incompetence, an inexperienced/misinformed production sound mixer who tries to record the hottest signal possible, with only 6dB or so of headroom, a mantra from the days of analogue production sound recording back in the 1990’s. There might be some valid use cases for 32bit recording (beyond extreme incompetence), where there may be some truly huge peak levels and dynamic ranges, weapons/explosions recordings for action films for example, but even then you’ve still got to know what you’re doing because the issue of most concern would be overloading and physically damaging the mic capsules.

There are use cases for 32bit float audio files during editing and I’ve used them myself but they’re rare. For example, we had a scene with an actor in a large gymnasium, the lav was faulty (almost constant breakup), it was quite a wide shot, so the boom mic was well beyond the optimal distance, the levels were low, the noise floor was high and there was a ridiculous amount of reverb. It was a no-brainer to schedule the whole scene for ADR (dialogue replacement). Unfortunately, Covid hit and killed the actor, so ADR was obviously off the table and the original production sound had to be rescued. This involved a great deal of off-line processing, de-clicking, de-hum, several rounds of de-reverb and numerous rounds of various noise reduction tools with gain and EQ boosts between steps to get the dialogue up to useable levels. Many hours of work for a borderline acceptable result. Recording the result down as 32bit audio files (to re-import into the session) instead of the original 24bit format was prudent because if clipping had been inadvertently introduced in any of the steps (and not noticed) the whole process would have to be started again from scratch with 24bit, as opposed to just lowering the level with 32bit files. There are related potential use cases with music production, bouncing down the final mix for example (for transfer to the mastering engineer) but again, 24bit is more than enough for any competent mix engineer.

G

 

[Ryokan]

That's not what my exchange with gregorio has been about. He and I have both accepted that current standards with audio engineering are fine with music production. That you have a professional like him that knows how to capture audio and process for 16bit/44.1kHz, without needing advertised 32bit float audio recording, We seem to have argued what consumer base 32bit float audio is advertised for.

'However, there are some scenarios where having room for error can be a benefit. “Where I do really see 32-bit being a huge advantage is for journalists who are recording out in the field, or for documentary filmmakers that are capturing things that you only get once and can be unpredictable,” Pereira says.

https://www.wired.com/story/32-bit-float-audio-explained/

 

[colonelkernel8]

Interesting take. I grew up with modern SS stuff heck actually started with SS and always have this so called listening fatigue and even with treble eq all the way down to where it sounded much lower in volume relative to bass, I still feel that fatigue. With my first exposure to tubes a couple of years back, I’m giggling so much in awe of how pleasing the sound was that I can finally crank up to party time listening levels (90+ dB SPLs) with the addition of pleasing harmonics and natural/subjective treble roll off (because it also is affected by harmonic distortion from tubes and output transformers too not just literally rolling off treble). I don’t have my eardrums or brain feeling tired at all even after listening to 2 hours straight from very loud volumes

That "fatigue" is 100% in your head. Sorry to burst your bubble there.

 

[theveterans]

That "fatigue" is 100% in your head. Sorry to burst your bubble there.

That's okay, not everyone experiences the same thing. Some have hyperacusis while others don't. It's simple as that

 

[colonelkernel8]

That's okay, not everyone experiences the same thing. Some have hyperacusis while others don't. It's simple as that

That's not what hyperacusis is.

 

[theveterans]

That's not what hyperacusis is.

No intended pedantic, just an analogy that not everyone perceives the same way as I do

 

[bigshot]

Not everyone’s brain processes it like your does. You talk about things that exist in your mind as if they exist in the sound. But they don’t.

The main cause of listening fatigue is headphones that aren’t comfortable. That has nothing to do with sound. Neither does your expectation bias.

But I will say this, I’m glad I’m me and don’t have all these mental biases interfering with my enjoyment of music.

 

[theveterans]

That’s your experience which is definitely different from mine. Without my brain, I could never interpret sound waves do I. This is where discrepancies between what is measured and how it’s perceived lies. I don’t interpret the same sound waves as you and if the specific accurate measured gear sounds like headache inducing fatigue to me, it’s crap to me no matter what. The distortion laden tube amplifier that I use OTOH sounds like real music to my own personal perception. Zero listening fatigue with tube distortion to my perception

 

[bigshot]

Sound is measurable. It’s what gets recorded and played back. Digital audio is essentially measuring sound in order to record it and play it back. If what you perceive is different, it isn’t because of a wire or a DAC, you’re right. It’s because of your mind and all of those preferences and biases flying around between your ears. What happens up there in your noggin is unique to you. No one else experiences what goes on up in your skull. It means absolutely nothing to me, and it is irrelevant when I go shopping for gear because I want fidelity, not something that suits your placebo and bias. Your experience is purely solipsist.

Now that we’re agreed that you’re talking about what happens in your head, not what happens in your ears, I can cheerfully dismiss your experience as inapplicable to anyone else.

 

[theveterans]

Sound is measurable. It’s what gets recorded and played back. Digital audio is essentially measuring sound in order to record it and play it back. If what you perceive is different, it isn’t because of a wire or a DAC, you’re right. It’s because of your mind and all of those preferences and biases flying around between your ears. What happens up there in your noggin is unique to you. No one else experiences what goes on up in your skull. It means absolutely nothing to me, and it is irrelevant when I go shopping for gear because I want fidelity, not something that suits your placebo and bias. Your experience is purely solipsist.

There's always choices for everyone. You choose fidelity, I choose what sounds best to me

 

[bigshot]

I have absolutely no objection to you choosing anything for whatever reason you want. I only have a problem when someone tries to suggest that their personal subjective impression might be applicable to other people, or might be more applicable to others than measures of objective fidelity or controlled listening tests.

Subjective impressions apply to you alone. Objective fidelity applies to everyone.

 

[gregorio]

'However, there are some scenarios where having room for error can be a benefit. …

While not entirely wrong, it’s not exactly correct either. In response to this quoted assertion for example; Having room for error when recording live sound is always a benefit, 16bit recording allows a lot of room for error, far more than the analogue recorders in the 1990s and 24bit allows way more room for error than 16bit. In other words, 32bit recording would be useful in those circumstances where a massive amount of room for error still isn’t enough, and such circumstances are very rare. So maybe for a journalist in the middle of a battlefield but even then, they’d still need to be mindful of levels (and of not being blown to smitheries)!

No intended pedantic, just an analogy that not everyone perceives the same way as I do

And how many people do you think would perceive cranking up the levels and adding additional harmonics to be less fatiguing than significantly lowering the levels and rolling-off the mid/high freqs? Presumably even higher levels and adding even higher amounts of harmonics, that would have normal people covering their ears in discomfort, would be even less fatiguing to you, so relaxing in fact that you’d drift off into a nice deep sleep?

Without my brain, I could never interpret sound waves do I. This is where discrepancies between what is measured and how it’s perceived lies. I don’t interpret the same sound waves as you …

What sound waves do you interpret then? That should be easy to test, set up a microphone next to one person listening to a track then replace that person with you, repeat the same track and measure the difference in the sound waves recorded by the microphone.

The distortion laden tube amplifier that I use OTOH sounds like real music to my own personal perception.

Which proves there’s something seriously wrong with your “own personal perception”, because real music (EG. Acoustic music) is not laden with any tube distortion at all! And non-acoustic music is already laden with exactly the amount of distortion desired by the musicians/engineers, so ladening it with even more can only make it sound less real.

As your perception is, by your own admission, so screwed up that it’s pretty much the opposite of reality, why would it be of any interest to anyone else or be a valid basis for assertions here, and why haven’t you visited an audiologist?

G

 

[theveterans]

What sound waves do you interpret then? That should be easy to test, set up a microphone next to one person listening to a track then replace that person with you, repeat the same track and measure the difference in the sound waves recorded by the microphone.

My voice sounds weird with mic. I recognized it's me, but it's distorted relative to my natural voice.

And how many people do you think would perceive cranking up the levels and adding additional harmonics to be less fatiguing than significantly lowering the levels and rolling-off the mid/high freqs? Presumably even higher levels and adding even higher amounts of harmonics, that would have normal people covering their ears in discomfort, would be even less fatiguing to you, so relaxing in fact that you’d drift off into a nice deep sleep?

People that love tube distortion share the same sentiment that I do.

Which proves there’s something seriously wrong with your “own personal perception”, because real music (EG. Acoustic music) is not laden with any tube distortion at all! And non-acoustic music is already laden with exactly the amount of distortion desired by the musicians/engineers, so ladening it with even more can only make it sound less real.

As your perception is, by your own admission, so screwed up that it’s pretty much the opposite of reality, why would it be of any interest to anyone else or be a valid basis for assertions here, and why haven’t you visited an audiologist?

Real music without going through ADC or any amplification sounds natural to me. Recoded music with tube distortion is closer to how my brain perceives unamplified live sound. This is also true for those that exclusively listen to tubes

Audiologists put test tones that are electronically generated. FWIW, I have no issues with hearing 10K test tones down to whisper levels (30-40 dB SPL)

 

[castleofargh]

My voice sounds weird with mic. I recognized it's me, but it's distorted relative to my natural voice.

That has nothing to do with mics.

 

[Ryokan]

While not entirely wrong, it’s not exactly correct either. In response to this quoted assertion for example; Having room for error when recording live sound is always a benefit, 16bit recording allows a lot of room for error, far more than the analogue recorders in the 1990s and 24bit allows way more room for error than 16bit. In other words, 32bit recording would be useful in those circumstances where a massive amount of room for error still isn’t enough, and such circumstances are very rare. So maybe for a journalist in the middle of a battlefield but even then, they’d still need to be mindful of levels (and of not being blown to smitheries)!

There must be more to it than that for a well regarded publication to state this as a good reason in favor of a 32 bit float?



Some pros and cons;

32-bit float audio allows for more flexibility in post, as it prevents digital clipping by enabling the adjustment of gain without introducing distortion.


Files are 33% larger compared to 24-bit, which could be a consideration in storage and processing.

High Dynamic Range

Captures a dynamic range up to over 1500 dB, making it ideal for environments with unexpected loud sounds.

Requires clear communication in workflows, especially if others are expecting standard 24-bit files.

Eliminates the Need for Perfect Gain Staging

Compatibility Issues

Gain staging becomes less critical during recording, as levels can be adjusted later in the DAW.

Some older DAWs or hardware may not fully support 32-bit float files, potentially complicating the production process.

Ideal for Uncontrolled Environments

Possible Clipping at the Pre-Amp Stage

Perfect for live music or location recording where unexpected changes can lead to clipping.​

https://www.production-expert.com/p...t-float-audio-when-is-the-best-time-to-use-it



This guy gives an example of what a 32 bit float in a music file means,