Cam Audio Recording Request: 32-Bit Float

Requesting that future cameras get 32-Bit Float Recording for their internal recording. Not just from the on camera microphones, but also through the XLR inputs.

One thing I hate when shooting direct into camera at times is that adjusting audio levels suddenly on camera is not as perfect as when using a mixer/recorder. I had levels set on the last shoot, and when the speaker suddenly went up in volume I reached "RED" levels, and was super close to peaking. If the camera was 32-Bit Float then I wouldn't have to worry about that. But I had to adjust and ask that we get that section of the teleprompter script again. If it was an interview with a subject that was going through an emotional story and suddenly they cried loudly a heartfelt part of their story I wouldn't want to miss that going into camera with audio.

Ideally we'd be recording audio externally. But sometimes that doesn't happen. It's become a common occurrence for us to shoot with audio going direct into camera. So in that instance 32-Bit Float would be a lifesaver. Thank you!

+1!

If the camera was 32-Bit Float then I wouldn't have to worry about that.

You would still have to worry.

While having float 32 for audio is a great advantage for both accuracy and mitigating wrong initial levels it doesn't increase the dynamic range of the recording. The saturation of a signal is dependent on the ADC and pre-amp capability.

You would still have to worry.

32 bit float recording still isn't broadly understood.

It is a completely different recording pipeline than regular 16/24bit recording, including hardware.

It's similar to dual gain output sensor recording, two signals are being recorded separately at vastly different gains and then processed and combined internally. This requires different hardware than regular recorders.

The biggest advantage is that you don't need to ride gain while recording, at all. It doesn't matter if it's set to zero or 100, you will always get the perfect gain, you just have to dial it in in post.

Those always perfect levels also make the quality of converters / preamps less important. What it does not compensate for is proper placement or clipping of the microphones themselves.


With that out of the way, 32bit float recording may be more expensive and difficult to implement in a camera.

You would still have to worry.

32 bit float recording still isn't broadly understood.

It is a completely different recording pipeline than regular 16/24bit recording, including hardware.

It's similar to dual gain output sensor recording, two signals are being recorded separately at vastly different gains and then processed and combined internally. This requires different hardware than regular recorders.

The biggest advantage is that you don't need to ride gain while recording, at all. It doesn't matter if it's set to zero or 100, you will always get the perfect gain, you just have to dial it in in post.

Those always perfect levels also make the quality of converters / preamps less important. What it does not compensate for is proper placement or clipping of the microphones themselves.


With that out of the way, 32bit float recording may be more expensive and difficult to implement in a camera.

1. Recording in float 32 instead of fixed-bit recording has nothing to do with dual gain recording.
2. The dynamic range depends on the quality of the ADC and pre-amp it has nothing to do with using float 32.
3. The implementation of float 32 is mostly a CODEC issue it does not have any grave technical hurdles.

I think this Sound Devices fact sheet explains it well:

https://www.sounddevices.com/32-bit-flo ... explained/

And this fact sheet explains how it's recorded (in their devices):

https://www.sounddevices.com/how-is-a-3 ... -recorded/

Crucially:

"Four distinct subsystems are required for 32-bit float operation. These include:

A very high-dynamic-range analog microphone preamplifier
Multi-stage A/D converters
32-bit floating point processing and signal paths, including USB interconnection
32-bit float WAV recording file format"

To a degree I'm more camera oriented than sound. But I know when I'm using the MixPre series and have it set to 32-Bit Float that if the person suddenly gets really load that I'm not losing anything. I can recover it in post. I don't understand all the technical aspects, but I do love that. As a camera person that makes it far easier for me to deal with sound.

Thus why I'm requesting it in camera for the audio recording. My preference would always to have a professional sound person on set recording all audio externally. Let them handle the sound while I work on lighting and shooting the images I want to capture. And, I'll let them pick up the additional sounds I'll need for me to edit in post the sound design I'll want.

But when it comes to the technical aspects of audio I'll defer to the audiophiles. For me I simply want 32-Bit Float Audio in Camera because it will make life easier when I do have to do sound into camera. I understand as long as the microphone doesn't clip that means the audio file won't, even if the subject gets loader than the gain levels set in camera. That's a wonderful feature of 32-Bit Float. And, if they get really quiet I can still bring up the volume and it will sound just as good. That to me is a major benefit.

For me I simply want 32-Bit Float Audio in Camera because it will make life easier when I do have to do sound into camera.

It could probably be done, but would definitely increase the cost of the camera and can't just be done with a firmware update: you need new preamps and ADCs in the camera. A new generation of cameras could incorporate it, albeit at a higher purchase price. I doubt the existing preamps and ADCs in these cameras are up to the job, but I could of course be wrong.

1. Recording in float 32 instead of fixed-bit recording has nothing to do with dual gain recording.
2. The dynamic range depends on the quality of the ADC and pre-amp it has nothing to do with using float 32.
3. The implementation of float 32 is mostly a CODEC issue it does not have any grave technical hurdles.

You clearly don't know how 32bit float recording works, please inform yourself before making these claims.

For me I simply want 32-Bit Float Audio in Camera because it will make life easier when I do have to do sound into camera.

It could probably be done, but would definitely increase the cost of the camera and can't just be done with a firmware update: you need new preamps and ADCs in the camera. A new generation of cameras could incorporate it, albeit at a higher purchase price. I doubt the existing preamps and ADCs in these cameras are up to the job, but I could of course be wrong.

The Deity PR-2 I got that does 32-Bit Float costs only $229. So say you increase the cost by doing two XLR Ports that both go into Pre-Amps, were maybe talking about a $500 price increase to the consumer I would guess. But I’m completely going based off the cost of one device that does 32-Bit Float.


Sent from my iPhone using Tapatalk

The Zoom H1 essential can record 32bit float and is just $100.
But I would think getting this inside a camera requires knowledge and experience with the tech and probably a bit more space.

I wonder if anyone has compared audio quality of these less expensive recorders with the Sound Devices MixPre-3 which is their budget option and intro-level quality.

I'd love to see a Pocket 4K Gen 2 with good mics and 4 channel 32 bit Float myself.

float 32 for audio...doesn't increase the dynamic range of the recording.

Say what!?

"24-bit audio recordings can capture a dynamic range of up to 144.5 dB. Meanwhile, 32-bit float audio can capture the ludicrous range of up to 1,528 dB."

float 32 for audio...doesn't increase the dynamic range of the recording.

Say what!?

"24-bit audio recordings can capture a dynamic range of up to 144.5 dB. Meanwhile, 32-bit float audio can capture the ludicrous range of up to 1,528 dB."

That's like saying that adding a 10-bit or 12-bit mode in a low-quality video camera will suddenly increase the dynamic range.

The ADCs and the pre-amp determine the dynamic range not the bit-depth of the CODEC. Recording 32-bit float by itself will not increase the dynamic range. It will however avoid amateur mistakes of totally setting the wrong recording levels.

1. Recording in float 32 instead of fixed-bit recording has nothing to do with dual gain recording.
2. The dynamic range depends on the quality of the ADC and pre-amp it has nothing to do with using float 32.
3. The implementation of float 32 is mostly a CODEC issue it does not have any grave technical hurdles.

You clearly don't know how 32bit float recording works, please inform yourself before making these claims.

I suppose we have to agree to disagree.

The Deity PR-2 I got that does 32-Bit Float costs only $229.

How do you like that recorder? I was about to pull the trigger on the Tentacle System when I saw Curtis Judd's review of the PR-2. I like that you can use the little screen as well as the app and that their software runs on Windows and not Mac only like Tentacle. When I make my decision this week, then I could:

I wonder if anyone has compared audio quality of these less expensive recorders with the Sound Devices MixPre-3 which is their budget option and intro-level quality.

Try this out for Rick's interest since I own the Mixpre3ii. Then I could test further and tell you that 32 float only sounds best with silver interconnects and when the recorder is sitting on my Black Diamond Racing Pyramid cones atop my Townsend Seismic Sink.

As someone with a career in production sound lemme clear the air a bit:

First up, getting a signal from multiple ADCs to increase dynamic range has been around forever. Hell, my Zaxcom Nomad12 has had that since I got it in 2012 (called NeverClip in their feature list) and I can confirm that even on its "meagre" 24-bit recordings I don't know if I've ever heard it clip. I'd take 32-bit float recordings if it was an option, sure, but it's not required by any stretch. 24-bit still has more dynamic range that pretty much any mic we use in production and setting levels really isn't that hard folks.

2nd, 32-bit does have its place but it's not a complete panacea. It can only record what you feed it, so if your mic placement is dogshit, your mics can't handle the SPL of the source, the pre-amps in your wireless system clip before the audio is even transmitted, etc., then 32-bit ain't gonna be much help. 32-bit is only really helpful if the only problem you have is that you can't set your input trim to save your life. Honestly, to deal with this problem more effectively I would rather have AES support and digital sources so I could skip analog gain staging completely.

3rd, and kind of carrying over from the previous point and what Cary Knoop is saying, most of the options out there for 32-bit float recording are built on pretty shoddy hardware platforms. I've heard a bunch of 32-bit float recordings where you could clearly hear when the recorder was switching between ADCs and it was extremely distracting. Was it as distracting as clipped audio? Arguably not, no, but it still highlights the fact that—even with infinite dynamic range—you still need to know what you're doing if you want to get good sound.

All of this to say that I would not turn down 32-bit float recording in any of my production sound recording hardware. I'm not the kind of person interested in trying to argue that more isn't more or that progress is a bad thing, but what I will say is that a lot of the benefits to 32-bit recording simply don't present themselves in a lot of contexts. I know it's not an apples to apples comparison, but it's worth mentioning that we've been able to record 192kHz audio for a while now an the standard in production recordings remains 24-bit 48kHz for a variety of reasons. I genuinely don't see that changing until solutions are found for other bottlenecks in the signal chain.

The Deity PR-2 I got that does 32-Bit Float costs only $229.

How do you like that recorder? I was about to pull the trigger on the Tentacle System when I saw Curtis Judd's review of the PR-2. I like that you can use the little screen as well as the app and that their software runs on Windows and not Mac only like Tentacle. When I make my decision this week, then I could:

I'm loving the Deity PR-2. Used it on a professional shoot last week. Only annoyance was that I hid the Lav with some moleskin in the subject's shirt and because I couldn't monitor it at all times got some clothes rustling. But that's a common lav problem when hiding them in clothing. I even get that with my Wireless Lav system, so it's not like it's the PR-2's fault. Still, some 20% voice isolation in Resolve got rid of that quick.

The battery life on it is great. The audio sounds great. And, I really do love having the LCD on the device. I can hit the record button on it or the app. I can lock it. I forgot to lock it so the subject sat in between some takes and accidentally hit the record button to stop recording. Luckily I had the boom mic going. So just a learning moment that if I'm hitting record on the device and leaving it I should probably lock it.

The screen is definitely one of the best parts. But it has far better battery life than the Tentacle. Honestly, since I got the Deity TC1 kit I've mostly been skipping using Tentacle. Even when the sound person brings Tentacles I opt for using my Deity Kit because I prefer it that much. And, on the feature film we shot last year we used Deity for the whole shoot for timecode. I really do love having the LCD on each device. Quickly checking it without having to log in to the app. But the app is still great too.

Overall, I'm very happy. I'd say pull the trigger on it. Now we only need Blackmagic to support the Deity Timecode System with the iPhone and Android BMD Camera Apps.

Thanks for the detailed information! I'll pick it up tomorrow.

I have an interview situation coming up in Mexico (two people) and I have never done that before. I figure I'd get two of these, the TC-1 and some sort of travel boom so I can put my MKE600 overhead so I'm capturing both. Probably the boom mic into the BMCC6K. I'll keep it in mind that I need to lock the device. I'll leave the lavs out since it's not a big deal in this context. This is helpful.

I wonder if anyone has compared audio quality of these less expensive recorders with the Sound Devices MixPre-3 which is their budget option and intro-level quality.

The MixPres have exceptionally good mic preamps, with multiple ADCs (analogous to the dual gain principle used used in sensors), to achieve a high DR regardless of bit depth. All Sound Devices mixers/recorders do as well, though they are not all of the same construction. I have a Mixpre 6 II and a 302 mixer and they are both very good, in this regard. On the Mixpres they are quoted to be 142 dB dynamic range min (A-weighted). The full specs of 10 Hz to 80 kHz +/- 0.5 dB re 1 kHz @ 192 kHz sample rate, greatly exceeds that of any camera too. I would venture they are also usefully superior to the Zoom products.

As has been mentioned 32 bit float should be unnecessary - the real problem is the BMD cameras have no limiters nor autolevels, which are common on most every other camera particularly ENG ones; and would be more useful often, for where audio needs to be recorded in camera.

32 bit float may be useful and popular for set and forget non professional audio recording but it is also contingent on very good preamps and exceptionally good mics, with low noise floors and wide DR, particularly max SPL. Most mics will not cover even 24bit full DR. Professionally 32 bit float is often used for special FX; an example might be rain in a thunderstorm with lightning; where the huge DR range from light patter to crack could both be accommodated, usefully.

Sound Devices limiters tend to be very high quality also; better than those found on most cameras and this is normal too.

1. Recording in float 32 instead of fixed-bit recording has nothing to do with dual gain recording.
2. The dynamic range depends on the quality of the ADC and pre-amp it has nothing to do with using float 32.
3. The implementation of float 32 is mostly a CODEC issue it does not have any grave technical hurdles.

You clearly don't know how 32bit float recording works, please inform yourself before making these claims.

I suppose we have to agree to disagree.

No we don't.
You are spreading misinformation because you have no idea how 32bit float recording works, and then you double down on it. Get informed.

32 bit float may be useful and popular for set and forget non professional audio recording but it is also contingent on very good preamps and exceptionally good mics, with low noise floors and wide DR, particularly max SPL. Most mics will not cover even 24bit full DR.

I disagree on this a bit.

- DR/SPL of mic is irrelevant as mic placement is the same regardless of recording option.
- noise floor with 32bit float recording will always be at it's optimal which can compensate for noisier mics.

Pros for professionals:
- peace of mind, no gain errors possible in diifficult environments or scenes.
- purely focus on mic placement / boom operating for single person sound department.

I disagree on this a bit.

- DR/SPL of mic is irrelevant as mic placement is the same regardless of recording option.
- noise floor with 32bit float recording will always be at it's optimal which can compensate for noisier mics.

I can't see where you can rightly take issue here with what I said; 32 bit float is irrelevant if the quality of preamps and mics is not there; and it's all pretty self-evident, to my mind. Max SPL is very necessary sometimes, especially for loud sudden transients. In a previous life I was a location sound recordist; not that that means my word is gospel but I've built up a lot of good gear, in that direction; especially mics. YMV of course, Michel.

I have no need for 32bit float in camera audio because I always use external recorder for audio meant to be used and camera audio is just extra if it happens to be usable, the recorder has 32bit float but I hardly use it and use 24bit 99.99%.
But still I can see it possible as handy to have rather than not, it could save some possible usable audio when shooting in a situations like when you could expect unexpected loudness in run and gun and needing to use the audio from the camera.

So in my opinion it is not a must have but still nice option to have if really needed for some reason.

Max SPL is very necessary sometimes, especially for loud sudden transients.

I didn't disagree with that. It's just that it's equally true for 16, 24 or 32bit float, so it's not really a factor when comparing 'standard' to 32bit float recordings.

32 bit float is irrelevant if the quality of preamps and mics is not there.

I still disagree with that a little because the greatest feature of 32bit float is that no gain control is necessary and yet you'll get perfect levels, always. The two combined recordings guarantee minimum noise floor at all times and no clipping at all times (given proper mic placement of course). With lesser converters you may actually get better recordings compared to suboptimal gain riding with great preamps and ADCs, especially in a difficult recording environment.

A good sound recordist doesn't need it - but the big differentiator is not a better sound quality but the different work flow that can free up resources on set (but add work in post).

32 bit float may be useful and popular for set and forget non professional audio recording but it is also contingent on very good preamps and exceptionally good mics, with low noise floors and wide DR

In practice, the cheap $99 hand-held products offering 32 bit float can be abused with very loud noises and they won't clip. Where they do fall down is the noise floor, which would likely be an issue with implementation on lower cost cameras as well. But the higher end Zoom recorders and the Mixpres won't have either issue and they're still pretty cheap.

A professional sound recordist doesn't need 32 bit float, but ain't that sort of beside the point? How many people on this forum employ them? Hit record, forget and get great levels has obvious value.

In practice, the cheap $99 hand-held products offering 32 bit float can be abused with very loud noises and they won't clip. Where they do fall down is the noise floor, which would likely be an issue with implementation on lower cost cameras as well.

Indeed, a device encoding data using 32 bit float does not guarantee any better dynamic range.

32 bit float encoding is the future not just for audio but also for video. It would finally allow recording in linear light without fixed bit-depth restrictions (as you would need 16+ bit-depth to capture a good dynamic range) and do away with log encodings.

But don't fall into the trap that if a device records 32-bit float that somehow guarantees a better dynamic range, that depends on the ADCs and pre-amps.

I still disagree with that a little because the greatest feature of 32bit float is that no gain control is necessary and yet you'll get perfect levels, always.

I see what you're saying now Michel. I still reckon it would be all unnecessary, for most users, with good limiters and sometimes ALC too, in camera. I agree broadly with the 2 previous comments, John and Cary that is. The high end Sound Devices recorders have no 32 bit float it mighty be noted; and they are pretty de rigeur amongst pro location recordists, these days.

… Hit record, forget and get great levels has obvious value.

For a sole operator in theatrical or music video use cases where I cannot predict when I’ll have very low sound levels and very high sound levels, the MixPre-6 II has never failed to manage my sound levels well (within the limitations of my noise floor) as a starting point for my editing in Resolve.

Again, I never peaked on the last shoot I did. But it was close on the camera gain level. That's why I think it would be beneficial to have 32-Bit Float based on my experience using a MixPre, setting, and leaving it. Last year we did a short film where a girl's horrifying screams at the death of her father were so load they peaked on the MixPre. But because it was 32-Bit Float and the microphones didn't clip the information was recoverable. That was a nice benefit. I wasn't the audio recorder on set for that short, but it was someone who wasn't a full experienced sound professional so the 32-Bit Float saved them.

That's all I want in camera for some instances. Mainly documentary/corporate shoot scenarios where we do end up just recording audio into camera. Is that my preference? No. But, it is a reality of some productions. So there is a benefit to improving our audio recording capabilities in camera with better preamps and as well taking it up to 32-Bit Float. I wouldn't mind having a PolyWav file either that I can label the tracks so that I could label one as Boom and the other as Lav.

I'm still of the belief that we should always default to recording audio externally. Have a proper sound professional. Leave the camera to the visual capture. But that isn't the reality of the modern digital era like it was during the film era.

Nonetheless, I'm pleased to have sparked an informative debate on this subject. I'm learning some things I didn't actually know about audio recording in 32-Bit Float. Maybe this could inspire Blackmagic with their future camera products.

It's frustrating that some people confidently take part in this discussion when they very obviously don't know how 32bit float audio recording/recorders works.

So I'll leave this link here for them to hopefully read.

https://tascam.jp/int/feature/32-bit_fl ... 20clipping.

It's frustrating that some people confidently take part in this discussion when they very obviously don't know how 32bit float audio recording/recorders works.

Don’t take this personally, but in fairness it is equally frustrating to see folks parade around marketing materials like they’re articles from a peer reviewed scientific journal. I can see at least one glaring issue with that Tascam piece.

it is equally frustrating to see folks parade around marketing materials like they’re articles from a peer reviewed scientific journal. I can see at least one glaring issue with that Tascam piece.

If Tascam's science is wrong, let's hear it, but the fundamental claim, which you can find in Sound Devices documentation as well (and backed up by them with recorded samples), is that if the mic doesn't clip and the preamps are of good quality (like theirs), level settings are irrelevant. The material will never clip or be drowned in the noise floor, with any levels setting or no levels setting. This includes 32 bit float material intentionally recorded with levels "wrong" by 30 or 40db too much or too little. The recordings will sound every bit as good as if they had been set to "correct" levels. SD has files to prove it.

If you dispute these claims, minds are open....

Tim, I agree with having the ability to label the audio channels as I can with the MixPre. I either use the microphone name or which instrument that mic is recording on a music video such as piano, bass, vocals, guitar etc. That metadata carries through to Resolve so it’s clear how I should proceed to edit and mix the audio.

Tim, I agree with having the ability to label the audio channels as I can with the MixPre. I either use the microphone name or which instrument that mic is recording on a music video such as piano, bass, vocals, guitar etc. That metadata carries through to Resolve so it’s clear how I should proceed to edit and mix the audio.

Absolutely! And, when shooting narrative labeling a track that is a Lav on an actor with the Character name... brilliant!

There's several ways Blackmagic can improve the in camera audio recording. Yet, no matter what using an external mixer remains the best. Especially when it does what Sound Devices has done.

it is equally frustrating to see folks parade around marketing materials like they’re articles from a peer reviewed scientific journal. I can see at least one glaring issue with that Tascam piece.

...

Tim, I agree with having the ability to label the audio channels as I can with the MixPre. I either use the microphone name or which instrument that mic is recording on a music video such as piano, bass, vocals, guitar etc. That metadata carries through to Resolve so it’s clear how I should proceed to edit and mix the audio.

Another cool yet little-known feature is that if you get the $100 Musician Plugin (which records individual WAV files for each channel, no polywav file that you need to split up in post) and use the "wav all ISOs" command in the render menu, any markers you added while recording will be embedded in the metadata and will show up in some DAWs, such as Reaper, and editors such as Audacity and Acon Digital's excellent Acoustica. I believe the "wav all ISOs command" involves resampling, which might be a dealbreaker for some people but I've never experienced any drawbacks. You can also name markers using the recorder and those names will show up in the DAW. Never tested Fairlight to see if it uses the marker metadata, as I'm giving Fairlight some more years to mature before I start using it for anything more than the basics.

For longer concert recordings I usually use markers between songs instead of stopping/starting the recorder as sometimes you lose something you want in the brief instant between stopping and restarting recording. And I have an efficient workflow set up in Reaper to split long recordings into individual files using markers.

Thanks for the tip on the use of markers in recordings. I just record everything in a concert and then add markers in Resolve post as appropriate as well as deleting the empty space between songs if present.

The high end Sound Devices recorders have no 32 bit float it mighty be noted; and they are pretty de rigeur amongst pro location recordists, these days.

See the 20s mark of this May 22nd Sound Devices video:

8-Series v10.01 Available Now!

Toronto sound recordist Nicolas Field on setting it up:

How To Enable 32-Bit Float On Sound Devices 833, 888 & Scorpio

Thanks for the detailed information! I'll pick it up [Diety PR-2] tomorrow.

Note that you can get the international version of the PR-2 from Canadian suppliers.

Thanks for the detailed information! I'll pick it up [Diety PR-2] tomorrow.

Note that you can get the international version of the PR-2 from Canadian suppliers.

Thanks, but I looked to see if I could get it up in Vancouver and didn't see them at any of the stores there. I was thinking I'd drive up since a friend of mine works up there pretty regularly. But no one had it on their web site.

I'm no professional by any means, but I know a few pro audio people, and every one of them agree that 32-bit recording is overrated and ultimately unnecessary. I've heard some audio post houses refuse to even accept 32-bit source files, requiring you to re-render them before sending.

On a professional recorder? Maybe. On a camera, though, seems an unnecessary expense...

Thanks for the detailed information! I'll pick it up [Diety PR-2] tomorrow.

Note that you can get the international version of the PR-2 from Canadian suppliers.

Thanks, but I looked to see if I could get it up in Vancouver and didn't see them at any of the stores there. I was thinking I'd drive up since a friend of mine works up there pretty regularly. But no one had it on their web site.

Vancouver is a centre for film production. Trew Audio is just one of the vendors of sound gear there. Maybe ask Rick Lang who his preferred Vancouver vendor is.

Deity’s website lists five Vancouver area camera shops that carry its gear.

I'm no professional by any means, but I know a few pro audio people, and every one of them agree that 32-bit recording is overrated and ultimately unnecessary.

It's debated on professional sound forums as well, with some insisting it's a marketing fraud and the others that the naysayers don't actually understand it and refuse to consider anything outside their own experience of 16 or 24 bit integer. Or regard it [inaccurately] as nothing more than 24 bit integer with 8 additional bits.

There's an earlier thread here which follows the same path. Marc Weilage and Mark Hemsley (an actual sound guy) were insisting it's all nonsense and the rest of us were arguing on the other side. Mark Hemsley eventually conceded the issue. Or at least appeared to. If he's listening, he can speak for himself.

viewtopic.php?f=33&t=189729&hilit=32+bit+float

As far as being "necessary", nobody ever said that. Typical consumer equipment has automatic gain control for adjusting to audio levels, which creates an unpleasant "pumping" effect. Similarly, an amateur sound person constantly "riding the gain" creates all kinds of other problems in post. 32 bit float avoids both.

On a professional recorder? Maybe. On a camera, though, seems an unnecessary expense...

I'd argue the contrary actually.
Especially with small crews where there is no dedicated sound department on set, 32 bit float makes a lot of sense. It frees up (already limited) capacities as creators don't have to control or even worry about audio gain anymore.

Again, 32bit float is not about a better sound quality - it's about a different workflow (which happens to result in better recordings for non-sound recordists).

Cinematography and location recording both suffer from traditionalist hall monitors who view innovation and new approaches with deep suspicion. I don't have any 32-bit float recorders myself and have survived fine without them but I think it's a wonderful feature for one-person crews or for people who are recording in unpredictable environments where sudden loud sounds are a potential. Limiters can handle that stuff but if you can avoid them kicking in so much the better.

Agree. But it's not just about not clipping, the more beneficial part in my opinion is not raising the noise floor when normalising a quieter signal.

This gets overlooked (or isn't even understood) often.

I’ve had a recorder that records 32-bit float for five years. I’ve used the function very occasionally. I used it for this stereo recording of a storm. This is 52 seconds of the recording. The spike on the left is a thunder crack, one of the best that I’ve recorded. I’ve included data about this recording in the top right of the image.

I recorded this in anticipation of the possibility that I might get a thunder crack like that. I say possibility because, where I am, this happens once or twice a year. I could have set my recording level low enough to record without 32-bit float. However, the sound level of the rest of the recording would have been too low, rendering it useless. I could also have used the recorder’s limiter, but I chose not to because I wanted to manipulate the sound, which is why, as the screen capture shows, I recorded at 96kHz.

A lot of people are now using 32-bit float as a matter of course. That’s cool, but it doesn’t interest me. I know how to set levels, which I’ve found is not rocket science, and I’m not keen on make-work.

I also don’t care whether Blackmagic offers 32-bit float. However, I have no intention of paying for it. For me, that would be a reason to buy another camera make.

iZotope RX screen capture:

rx.jpg (300.78 KiB) Viewed 7139 times

Cinematography and location recording both suffer from traditionalist hall monitors who view innovation and new approaches with deep suspicion. I don't have any 32-bit float recorders myself and have survived fine without them but I think it's a wonderful feature for one-person crews or for people who are recording in unpredictable environments where sudden loud sounds are a potential. Limiters can handle that stuff but if you can avoid them kicking in so much the better.

I think that this should be discussed in conjunction with what’s actually happening in the market. Courtesy of saturation marketing campaigns by Røde, DJI, Hollyland and others, we’re seeing the widespread adoption of sound recording via 2.4GHz wireless transmitters combined with features such as 32-bit float, noise reduction, etc. 32-bit float is not itself the product. It’s just a facet of the product.

These companies, and the influencers who work for them, have figured out that most people want to spend as little as possible in cash and effort on sound recording; don’t care about reliability, sound quality or what their subjects look like when being recorded; and don’t want to know anything about how to record sound properly.

Røde created this entire market. It’s genuinely interesting to research how it did it. Have a look at the early videos and written marketing material. Røde knew from Day 1 that its biggest challenge was to get filmmakers to accept, as normal, aesthetics that were considered completely unacceptable. Røde succeeded in this campaign, so much so that DJI and others started copying it.

Having created the market, Røde and its copyists have now moved to stage two - they're increasing the cost of entry.