24-bit vs 32-bit Float Audio Recording Quality Is There a Difference?

[@Podcastage]

Fantastic Video Curtis. Thank you for sharing your samples and expertise with us. You're the best.

[@curtisjudd]

Thanks, Bandrew 🙏

[@michalwalks]

I am just a entry-level hobbyist, but I got into 32-bit because I went to an airshow earlier this year and just wanted audio that would work even when the jets were ear-piercingly loud. It was nice to put on my binaural headset, have my GoPro on my chest and enjoy the day. Afterwards I was thrilled with the audio.

[@curtisjudd]

👍

[@marczakian3511]

Thanks for the content. When we moved from the Tascam DR40 to the Portacapture it was a game changer for our history podcast, as the actors and narrators could shift volume levels without us worrying about clipping. I have recovered takes that would have been unacceptable with the DR40 at 24bit. This will be a big advantage for lower budget drama, where you can’t control the environment and are not able to do lots of post with audio.

[@curtisjudd]

👍 this seems to be the most common use case. But of course, you have to address any cases where the audio exceeds 0 dB in post so it is more work in some cases.

[@SoundSpeeds]

Spot on Curtis. And about 32 bit floating point... Most 32 bit float recorders have a lower EIN than extremely low self noise microphones so boosting low audio will result in the same noise floor when levels are raised. When there's a loud sound that clips a microphone, it doesn't matter how you record it because the sound is still clipped. It's like having clear water that gets dirt in it. The water doesn't magically clean itself because you drink out of a sterile cup vs one that was washed in tap water. Just wanted to add that...

[@curtisjudd]

Thanks, Allen!

[@LensVid]

First - thanks for another great video Curtis. My take on this and we discussed this recently on LensVid Talk is that if there is no actual drawback to 32bit float (and I am not familiar with any technical drawback compared to 24bit) then you should always use 32bit float. It basically gives you the peace of mind that if you did forget to set the levels correctly or something unexpected happens you can still salvage the audio no matter the levels. It is not a magic bullet against all problems but it does solve a very frustrating issue that many single operators faced for many years, especially in the field. In other words - when all other things are equal - 32bit is always better than 24bit. Really the only reason why this is even still a question is that we are in a time where not everything is yet 32bit float - I suspect that in a couple of years, every new recorder (maybe even many cameras) will have 32bit float built in and there will not be any other option. I am very curious as to why none of the small portable (Rode/DJI/Hollyland etc.) wireless systems are not 32bit already - I am guessing the next generation will be (Rode actually has some mention of 32bit in their software but I am not aware that the hardware supports it - maybe you know something about this).

[@curtisjudd]

With the RODE NT1 5th gen, I'd rather record to a quality recorder with 24-bit linear than with the on-board 32-bit float USB. That actually did sound better to me. There are a TON of nuances. The quality of the preamp, quality of the converters, dynamic range and max SPL handling capability of the microphone, mic placement, etc. For spoken word, yes, generally it isn't going to hurt, but in the case I cited above, the preamp and converters on a MixPre sound better to my ear than those built into the microphone.

[@tom_k_d]

A drawback is that 32bit float recorders need to combine multiple, usually mediocre, preamps to achieve their very high dynamic range. Using a single high-quality preamp, at the cost of spending 10 sec more on proper gainstaging, should usually provide better results.

[@LensVid]

@@curtisjuddHi Curtis (always fun to talk). I understand where you are coming from, but I was talking about all other things being equal so exactly the same setup with 24bit and 32bit (I think most recorders will give you the option to choose). If you are in a completely controlled environment and you have the time to fine-tune the audio completely and you still think that 24bit sounds a little better (not sure why - maybe the 32bit implementation is not perfect on the mixpre II for some reason), sure. But for most other users/use cases - especially for less experienced users (let's face it most video shooters are not audio engineers) - 32bit is really the way to go. P.S. any idea what is the 32 bit field on the rode wireless go II software?

[@LearnLightAndSoundSessions]

@@LensVid It sounded BETTER on the MixPre, likely because of much better preamps and converters than those built-in to the RODE microphone. That's the thing - all things are usually not equal.

[@LearnLightAndSoundSessions]

@@tom_k_d *multiple converters. Usually a single preamp with 2 or more converters per preamp.

[@SunriseWaterMedia]

I often hear people saying that they are recording 32bit float and they don't have to worry about distortion, but as you mentioned, 2:40 distortion can occur before the conversion.

[@curtisjudd]

Yes, and one thing I didn’t get into is that the preamp also has to support a wide range.

[@aristotle_4532]

If you know what mics you will be working with, you can design the maximum level to be above the distortion point of the mic. Zoom F2 does this with its bundled lavalier. The input clips at a voltage level which the microphone cannot actually reach. Soon, 32bit will be able to record the full dynamic range of any microphone.

[@johnmellor932]

True. I recorded a feature film and despite recording in 32bit still got mic distortion. I was using the Countryman B6 lav the high spl ones used recommended for theatre and still got mic distortion. You need good speakers to hear it though. The actor had a theatre background and went hell for leather with his performance.

[@aristotle_4532]

​@@johnmellor932 The maximum SPL also depends on the power the unit supplies. You can calculate that the distortion point output of the mic is 450mV or whatever at x dBA SPL using its spec sensitivity, but if the voltage is low the microphone may clip lower than that and very asymmetrically. In order to know what the recorder can do you need all the details. Sometimes, lavaliers are on gain stages through the transmitter, which makes it even more complex. You need to measure the distortion points or get precise info from the manufacturer, and then pick the mics according to the expected SPL.

[@SomewhatAbnormal]

Great advice! I think that one of the best cases for using 32bit is when you have no control of your sound, like documentary style filmmaking. In New Orleans you never know when a police car, ambulance or even a brass band my suddenly appear (seriously we have rogue brass bands) and 32bit helps with this. Also I’ve found it quite helpful for capturing foley sound for these types of vehicles. I’ve stood in front of an ambulance (with permission of the drivers) while they blasted me with their sirens - growlers and all - to capture sound in 32bit.

[@curtisjudd]

👍

[@JaredHoyman]

Completely agree! Although I have 32bit float 99.9% of the time I shoot in 24bit because I am in a controlled environment. For event videographers I can see where 32bit float is a great safety net.

[@curtisjudd]

👍

[@AllThingsFilm1]

Speaking for myself, I've never expected 32-bit audio to sound "better" than 24-bit. I only sought it for the flexibility 32-bit audio has in post. I own the Zoom F3 32-bit audio recorder. I got it for the reason you mentioned in the text in the beginning of the video. That being someone who often has multiple roles in filmmaking scenarios where a 32-bit float audio recorder like the F3 would seem to help me out. That being said, I did not ignore the fact that I need a good quality mic to record with. Currently, I use a Sennheiser MKE 600 and the Rode NTG5. Thanks for this video. You always do a great job at making audio recording easier to understand so that I have a better chance of capturing quality audio.

[@curtisjudd]

🙏 Thanks!

[@kneepayne]

Exactly.

[@drmatthewhorkey]

I like the last message! A lot of your stuff is WAAAYYY over my head in terms of need but I still love learning about it. I have a big project coming up and WILL get out there and get my message across. Thanks again

[@curtisjudd]

👍

[@MichaelBabich]

One BIG advantage of 32-bit float that not so many reviewers mention is that the first bit of this 32-bit in float format represents a sign: plus (+) or minus ( - ). In other words, above 0 dBFS or below it. So, 32-bit float technically may go up to +770 dBFS above 0 dBFS and go down to -758 dBFS below 0 dBFS. This gives effectively 1528 dB dynamic range. Not that it's even technically possible to produce such levels. But mathematically 32-bit float can save it. It allows the whole dynamic range of a signal―lets say the whole 144 dB (like in 24-bit range)―to be allocated anywhere within 1528 dB scale of the 32-bit float range. Even above 0 dBFS. E.g. like from +72 dBFS to -72 dBFS (72 + 72 = 144). The problem of FIXED depths―16-bit, 24-bit, and 32-bit fixed―is they all have 0 dBFS as an upper limit despite their dynamic range. I.e. their range only goes down below 0 dBFS. They only have minus ( - ) range but don't have (+) like 32-bit float. It means all fixed depths may clip easily despite their range and require careful gain control to make sure the signal is below 0 dBFS. Unlike fixed depths, 32-bit float both 1) has a huge range and 2) has 0 dBFS in the middle of this range, not as an upper limit. That's why 32-bit float doesn't clip and the gain level to keep signal below 0 dBFS doesn't matter―the whole signal will always be within the 1528 dB scale (+770 dBFS to -758 dBFS), will be saved in 32-bit float, and may be recovered in post.

[@curtisjudd]

Yes. *Assuming your analogue front-end and converters can deliver this. For spoken word audio, they generally can.

[@zbyszekolko3998]

No. A sign bit is not relevant. You also have sign bit in 24 bit signed PCM. 8 bit exponent value is what determine below and above sample level of "former linear 0dBFS" . In case of recording sound with multiple linear ADs what determine maximum analog recording level is a clipping value of the least sensitive AD converter. 1500dB of sound dynamic range is abstract and has no physical meaning.

[@curtisjudd]

@@zbyszekolko3998 Ok. Thanks. Yes, clipping is a factor of the analogue stage and the converters.

[@zbyszekolko3998]

@@curtisjudd You probably do not remember when I asked you a question what additional headroom 32F recording really provides? I figured it out myself. You have as much of additional headroom as you would set gain in mic preamp using 24I because the least sensitive A/D should operate at unity gain. You can clearly see that in case of Rode WGII vs WGPro. Both have fixed 20dB preamp gain. Max SPLs are accordingly 100 and 120dB SPL.

[@DigiDriftZone]

32bit is like a seatbelt, it doesn't improve your car driving experience until you or someone near you makes a mistake.

[@curtisjudd]

👍

[@hypnosecoachinghombach7517]

I remember one aspect that many were very excited about: With 32bit it was no longer possible to clip. That's why I bought a MixPre II. I regularly record sessions for my clients in an untreated room. Depending on the topic, I really shout into the mic and whisper in the same recording. I use a Neumann KMS 104+ and I still record in 24bit. So far I haven't been able to clip it yet.

[@curtisjudd]

👍

[@JoeBaermann]

If you know how to set the pre-gain and work with compressors for set bit rates it’s not an issue, 32bit float is more for ease of use and cutting costs, the latter does still require that preamps, converters and mic’s are up for it, I can’t unsee how many corners Røde have cut to get that into one single mic. It probably still needs some manual set gates when multitracking with multiple mic’s though?

[@Andreas-ov2fv]

I think this video should be like the go-to video every audio tech uses to send out to coworkers. A clear and succinct explanation showing that A) 32-bit float isn't just about the digital format B) requires hardware (ADCs) in the recording chain to benefit and C) doesn't magically improve the experential quality of audio. However, it does understate the wonderfullnes of 32-bit float audio for _non_ proficient techs. In your video you say "[...] Just learn to set your input level, your gain, and you'll be fine." This IS a big deal. To non-audio techs - like me - the audio side is already something like 50% science, 50% art and 50% witchcraft. On small productions, documentary productions, student productions, not having someone *dedicated* to pay constant attention to pushing faders or turning knobs is a godsend. For our "audio tech" - an amateur like the rest of us, making it up as she goes along - not having the responsibility to not _absolutely fuck our entire day's work_ on her back is HUGE. And she can pay attention to the stuff that she should be paying attention to, like sudden noise sources or bad mic positioning and not getting the boom into the shot.

[@curtisjudd]

I was hoping the main theme would come through as, "32-bit float is a great tool, but don't neglect other really important factors like using a quality microphone and a recorder with quality preamplifiers and converters, and placing the microphone optimally, managing noise in your location, etc."

[@jasonnakai4700]

Thank you for clarifying the front end of the audio chain...the microphone. My 416 distorted on a shoot but I was booming a car engine (Caffeine and Octane), and I am sure both my mic and limiters were maxed out. I did a quick playback and it was in the red, but some if it sounded usable. I should have had 32 bit float on that shoot 😢

[@curtisjudd]

👍

[@huyked]

Always good to hear your thoughts on tech, Curtis. Thank you for the rundown.

[@curtisjudd]

👍

[@MichaelGAubrey]

I've been using a Zoom F3 for going into my Sigma FP and recording 32bit-float really only as backup, but I've also been pleasantly surprised by how clear my line output is to the camera since I'm basically turning the camera's preamps all the way down.

[@curtisjudd]

👍

[@sparkestudio]

32bit float has only one real use to myself, lifting audio recorded at far too low of a volume that needs to be lifted without adding audio hiss or when mistakenly clipping from having too much volume and pulling the volume down without distortions. I use the Zoom F3. The Vocaster Two if I shout really loud during testing it is very difficult not to mistakenly clip the audio. The microphone can be an issue SPL if I am correct is the labelled spec. The Rode NTG I found to be very poor in terms of Sound Pressure Level, way too easy to distort if not using -20Db on the pad

[@curtisjudd]

Yes, good practical benefits.

[@petermcginty3636]

Thanks Curtis for keeping me up to date. Once again, you have cured my GAS addiction 😂. From your video, I have learner that this stuff I beyond my skill level and, therefore, I don't need to buy any additional great😂🎉🎉. Many thanks.

[@curtisjudd]

👍

[@hectorhivers8180]

As for Audio Samples, to my impression and to some extent, I feel that the 32-bit sounds louder, more prominent/present and clearer. But obviously the sound changes depending on what recorder or device is being recorded. Well, in favor of 32-bit, there is the obvious sound cushion to get the sound back after saturation, but you can still get back, as far as I remember, very low sounds. Also, I once saw a video comparing 24-bit silences with 32-bit. And the latter sounded quieter and cleaner. I look at it more in the following way. If you have features that contribute and help to have a better sound to some extent why not use 32 bit floating? Still, it doesn't take away from the fact that I am willing to listen to the possible negative points of 32 bit floating. If I am told that voices and/or sounds that capture them in a worse way or sound worse, I think I would question more whether to use them or not. In fact, I think that's mentioned in the video to some extent and they ask that question but haven't had the chance (or the "tools") to evaluate that. But I am always on the lookout for what new thing people more experienced than me have to say and what new discoveries are made about it. That was a good example of the horror movie where you have to record a whisper accompanied by a scream in one take. In my experience, it happens to me several times on different shoots that the actors' voices are very volatile. They naturally speak softly but from one moment to the next they speak very loudly, in a very pronounced way, without shouting (I guess that's the way many are in Latin America, or at least in my country). Nothing I couldn't handle when recording at 24 bits, but 32 bits is a great help and insurance for those situations. On a side note, that phrase "go out and make a great sound" is appreciated. I don't know, it was inspirational and make me want to go out and record sound with a smile on my face. Best regards and I wish you a good day from Chile.

[@curtisjudd]

Thanks, Hector.

[@bhgemini]

Love that initial summary. Still worth a watch.

[@curtisjudd]

🙏

[@jayforcinema]

Curtis, you're a legend. Thank you for your perspective on it, especially about the bodypack transmitters with 32 bit float

[@curtisjudd]

Thanks Jay 👍

[@ahart228]

Thanks Curtis. Consider a video in which you discuss what microphones and recorders may be best with 32bit recorders? I’m thinking of documentary, field work, where you have to contend with ambient wind, traffic, etc. Perhaps, as in this vid, a short sidebar about the recorders as well. Thanks.

[@curtisjudd]

👍

[@estebangiordano_]

Greetings from Argentina. excellent video about a reality that is little talked about on KZbin about the need to use 32 bit floating. Thanks for the explanation

[@curtisjudd]

👍

[@starsi2612]

One late comment to thank once again @curtisjudd for another good video, especially for trying to clarify the 32-bit float topic, which is very interesting and not fully explored/resolved. It seems to me that there is not a clear answer to what is (for me) the most fundamental question. Is there any disadvantage in using 32-bit float format? Or better, can the use of multiple preamps and converters + following processing cause any problems to the resulting 32-bit sound compared to the same recording performed with same setup (in terms of microphone/position) at 24-bit? We would need some more precise answers on this before deciding if we want to invest in the "insurance" of 32-bit float and pay the price of the extra storage space + longer workflow. I have no doubt that I would use 32-bit float anytime I make a recording that is important to me, but would like to be reassured that I am not compromising the quality of my audio recording! You (Curtis) considered this when saying (e.g.) "I was hinting more at the possibility of creating a crossover network to recombine the audio from multiple converters as being a potential degrader of audio quality." But then your conclusion was simply that you did not seem to hear anything wrong in your 32-bit float tests. I found it interesting to read @chakerian "There might be another test you could do. Poorly combining the outputs could result in gains/losses at the crossover point, too. Maybe test it with a known flat mic, and see what the white noise looks like across the spectrum to look for unexpected peaks or dips? And at different volumes it should move the peaks/dips (since it'll move which A/D converter it's using)." I would love if someone could try this and other methods to really stress test the 32-bit float recording system and find out if there is any potential problem and sound degradation. It does not help that there is not full clarity in the specs provided by Sound Devices, Zoom, Tascam, Rode, etc. They do not reveal the full details of the new 32-bit float recording technology and this is a real shame and makes me worry about this topic and hope for some more tests and discussions! Many thanks to all for the efforts!

[@curtisjudd]

One potential disadvantage - lower end devices use converters or crossover schemes to recombine the audio and that doesn’t always sound great.

[@randomgeocacher]

Excellent as always!! :) Use-cases are mostly about uncontrolled scenarios like fireworks, demolitions, drums, orchestra, crowds… and especially removing the SNAFU factor if you are a “one man band” feeling overworked, removing audio level worries and reducing your own stress. You set the levels for situation A, suddenly things changes, and too fast you did not have time or forgot to correct. At least I have messed up audio levels over gaining and getting garbage results. Audio quality due to bit are should rarely be an issue as final delivery quality doesn’t need more than CD quality (16bit linear), and honestly maybe that is slightly overkill since many of my consumers don’t listen in perfect headphones … mostly I deliver podcast / radio edit l, compressed audio with limiters and compressors, good for listening while commuting, Off topic: more concerned about enough bit rate for voices being h correctly transferred … back in the day some mad men claimed 96kHz/stereo was good enough… had they ever A-B tested or even listened to what they delivered? Even my poor hearing is good enough to pick up small differences in between wav and 128kbit/mono, and 96kHz/stereo is garbage. So eh, doing video a limiter is good enough for my personal projects but I would like to have a “future GH5” with 32bit float if ever produced. For spoken word / podcast where recording is being monitored while recording, 24bit has never failed me when able to monitor well. Video work back in the Zoom H1, H4N days, oh I messed up 24bit several times. That I was recording in the middle of samba bands, drums and what not was a factor for sure. 32 bit float (with a good mic) would have been a godsend gift to past me in those situation! :) Also yes: great choice of wording regarding high dynamic range recording and likening float to scientific notation. You could also drive home the point even more explicitly that it spectates amplitude magnitude from signal precision; so the point is easily understood if you don’t know the point of scientific notation. But maybe that’s overkill. Sorry if this got long and rambling :)

[@curtisjudd]

Yes, the situation with sound effects is a different matter. Definitely need to choose wisely with microphones in that case, and especially max SPL handling for loud sounds.

[@aristotle_4532]

Technically speaking, there is no reason to change gain setting on a good preamplifier and specific microphone combination. If you think you are getting a quality difference by the gain change, it is easy to prove there is no quality difference.

[@Kiwiphotog01]

Hi Curtis. Thank you so much for this. I have the Zoom F6, and have it constantly set to 32 bit float. I am a documentary maker and run the F6, Rode NTG4+, Sony UWP-D series lapel and plug in Tx, and a couple of Wireless Go's for run and gun shoots. I need your advice, I've been filming some Opera singers for a new doco, and need to get a better result. Mic placement etc are key but I'd like to get an indoor SDC for interviews... looking at your indoor mic episode I'm interested in the flexibility of the Oktava MK-012 kit, (using the hyer for interviews and cardioid for the opera recording). I've also seen the Avantone CK-1. What are your thoughts on these options?

[@curtisjudd]

I don't have experience with the Avantone, but the Oktava has a max SPL handling level of 130 dB SPL which is decent. Be careful when recording opera singers - mic placement will be critical for those who are very skilled at projecting. May need to back off a good bit. Also be sure to get a genuine Oktava - there are a lot of fakes out there.

[@kneepayne]

It was never about better sound, rather a more forgiving range of data. I don't recall manufacturers claiming "better sound quality" but if any did, shame on them. I can see people misinterpreting the information. Not much ROI in a controlled environment with a thoroughly sound treated room and a good mic (and of course knowledge). But, I do admit that not having to think about gain settings has been great!

[@curtisjudd]

I agree, I wanted to dispel that misconception - a lot of people ASSUME that 32-bit float = better sound quality.

[@johannesfruhstorfer7914]

My impression is that in all the examples 32bit produces more sibilant-sound (least in the Mac). The impression is stronger when listening on my small Genelec speakers than with headphones. Anyway, Curtis, thanks for your videos, I have learned so much from you !

[@curtisjudd]

Thanks. That’s a combination of different converters. I hear it too. Not so much a 32-bit float thing, necessarily.

[@jmalmsten]

The main reason I got into 32 bit float recorders, is that, generally, they eliminate manual gain setting. I just put the microphone in a sensible location and I hit record. No faffing about with getting the levels right and all that. I just record and I have enough resolution per sample so I can adjust things in post and still have full resolution samples at 16 bit or 24 bit export. Sound recordists in the world do not need to worry. Productions still need people who can put the right microphones at the right spots. It's just that now they can concentrate on that bit without constantly having to glance down at audio meters. I overgeneralize, of course, but for the vast majority of recording situations. It makes lifes easier.

[@curtisjudd]

👍

[@farreach84]

8:13 yup 32bit is great for someone, like me, doing video(stationary) and photo, who can't take care of the audio well in advance or can't take care of extreme gain fluctuation on the fly. I go to show/performance that goes from dead silent level whisper, to talking, to singing, and to screaming, on kids' small recital where the kids (usually K to 3rd graders, not so much from some 3rd graders to 6th graders) can sometimes go whisper monologuing cause they loose confident or forgets line and next minute sings and next minute screaming "yay" to response for leading adult teacher/actor to join in the "adventure" (in front of their parents, grandparents, siblings, aunts, uncles, etc) And I need to keep them all separate channels to prevent other kids fidgeting, talking to each other, coughing, etc going into the performing kid's audio (up to 6 kids). I bought one Zoom F2 unit last year and, despite a lot more work, this was way better in the final video. So I'll be switching to 5 more Zoom F2 from current small wireless setup of Rode Wireless Go2 + Comica Boom X-U.

[@curtisjudd]

👍

[@johnmellor932]

32bit is great but beware many video editors don't understand it. I still get back AAFs from editors where they have rendered the files in 24bit and clipped them. So if you're going to record in 32bit it would be wise to invest in Ediload and ask the editor for the edl files and reconfirm all the dialogue yourself from the original soundcards. Especially if its a feature film.

[@curtisjudd]

Agreed! 👍

[@V1aviation]

Curtis (or readers), I don't know if this comment is placed correctly here, but I can't seem to find any solution to my problem i'm having on my current audio recording. I'm using a Nikon Z8 for aviation videography with loads of variable dB's (from almost silent to clipping) and i'm recording with a Rode NTG4+ directly into the camera. What i've noticed is a lot of hiss and flat audio files (no bass or high pitch from the airplanes engines) and shooting roughly 40-50 clips for a video makes it too labour intensive to edit every single file. Is there a way to capture audio on an external audio device that plugs in to the camera (Z8 in this case) and maintain the audio quality of the microphone, or does the camera's audio card still processes it as a flat audio profile? NOTE: On my Sony RX-10, the audio coming out of the NTG-4+ is a lot richer and more vibrant. I hope I explained my situation in an understandeable way.

[@curtisjudd]

Not sure. I’d suspect the cable. Going from XLR to 3.5mm often gives mixed results from device to device in my experience. Impedance matching may be a factor, but not sure.

[@brett1934]

Hi curtis How can I know if my mic can or can't handle the 32 bit ?! Thank you, have a great day

[@tom_k_d]

Your mic can handle a max. sound pressure level (SPL), which is normally specified by the manufacturer. At this max SPL your mic will produce a specific voltage level, measured in the logarithmic scale dBu, depending on it's sensitivity, provided in mV (or dBu) per, usually 94, dB SPL (94dB SPL equals 1Pa). Max input voltage of the recorder is specified by the recorder manufacturers, at Sound Devices MixPre II it's +14dBu. A small example, based on U87ai specs: max SPL is 117dB SPL, sensitivity is 28 mV/94dB SPL, equals -28.8 dBu at 94dB SPL, This results in -5,8 dBu at max SPL, which the MixPre II could handle safely. Or take the SM7B: -57 dBu at 94 dB SPL resulting in +29dBu at its max SPL of 180dB, which would clip even the MixPre II. Now, 180dB SPL is a starting Space Shuttle in close proximity - and with the SM7B hooked to the MixPre II, you could still record insanely loud signals up to 165dB SPL.

[@curtisjudd]

@tom_k_d is correct. I'd look at the dynamic range spec for your microphone - or the self noise and max SPL specifications. Most microphones over $200 USD will be able to handle most spoken word recording situations.

[@EvanFotis]

@curtisjudd This would be a great topic for a separate video! if you can explain the above and share some recommendations for various types of microphones suited for 32bit float recorders

[@mrheng562]

I’m gonna reference this and tag team teach my students with the recorded you. :) I need to find mics that can handle my “wide dynamic range” voice. Even when I set my levels and placement correctly, I still tent to clip because I get way too excited. I think my daughter is turning out to sound like I do. 😅 Thanks for continually putting out useful content.

[@curtisjudd]

LOL! You might need 64-bit float. 😉

[@tucko11]

It's crazy man , thanks for the vid . I cannot tell 24 bit or 32 bit float. I just simply can't tell the difference to the naked ear. But definitely with 16 bit mono and stereo, I could tell the difference, however it's still hard to decipher even MP3 from wave someitmes , thanks so much appreciate the videos. I play guitar and do home recordings myself very helpful.

[@curtisjudd]

👍

[@hectorbenitez2163]

Another great video Curtis. Thanks for your contribution.

[@curtisjudd]

👍

[@LeoMahoney70]

Hi Curtis, I hope to listen to you today on your Sunday show (Sept. 15th). In case I miss you I'll ask my question now... Should limiters be off when using 32 float? I leave them off because I thought the float & limiter will fight each other during a recording. Should the limiter be on? Many Thanks, Leo.

[@curtisjudd]

Thanks for coming by the stream today, Leo. Yes, turn those limiters off, unless you're recording AND doing live sound at the same time, in which case: Turn the output limiters on. 👍

[@Legal-Canadian]

It would be awesome if you can share your thoughts on 32-bit capable microphones for different scenarios and what to look for in a microphone that is specifically going to be used with 32-bit float recorders. Also maybe elaborating more on “good” 32-bit float recorders and what separates the good from bad. Might be basic info but would be nice to have all of that in one concise video.

[@curtisjudd]

All good suggestions for another video. Thanks.

[@clashofideas50]

always teaching! thanks again Curtis!!!

[@curtisjudd]

👍

[@0LoneTech]

Float audio is like saving the exposure information while using automatic exposure. It won't save you from a broken light meter, but it may let you concentrate on other aspects.

[@curtisjudd]

Yes

[@PB-kf4qt]

Thanks Curtis! I'm a filmmaker working on a Direct Cinema documentary and not using a boom operator, as it's a very intimate situation with several characters, and they don't feel comfortable with one. Quick question: Can recording in 32-bit help me bring distant sounds forward in a room with multiple people talking, while also preventing clipping from sounds too close to the mic? I know it's not a professional setup, but since I'll have to handle the sound myself, do you think a 32-bit recorder can manage this?

[@curtisjudd]

No, 32-bit float does not give you more "reach" with your microphones.

[@tom_k_d]

Thanks for the clear words: I do not know any microphone / preamp combination that exeeds 24bit dynamic range (144dB). The best of the best preamps may deliver up to 130dB dynamic range. For example Avalon M5: noise floor -95dB, headroom +30dB, resulting in 125dB dynamic range - so an SM7B paired with the M5 would deliver 125dB dynamic range at maximum.

[@curtisjudd]

Yes, agreed. The main use case for most people actually has less to do with dynamic range and more to do with, “Oops, I was busy operating camera and didn’t set my audio gain correctly so now I have clipped audio…”

[@MrBassonist]

Or: oops, the only thing I know about levels is to set them to -6dB as someone used to say that. (aim for -30dB instead and it will not clip). @@curtisjudd

[@tom_k_d]

@curtisjudd Did further research, as I was curious now: SD claims to have 142dB DR with their MixPre II, using a single special design preamp (and several ADCs): -128dBu noise floor, +14dBu max mic input. This is were 32bit encoding really starts to make sense, as the max. 144dB range of 24bit encoding would require very precise matching, and hence starts getting impractical - chapeau to SD to reach this dynamic range with a single mic preamp.

[@YoungBlaze]

is there a lav mic that has a huge dynamic range for 32bit float?

[@curtisjudd]

I think of it differently. The wide dynamic range recorders let you get ALL of the dynamic range of your mic. Some mics have more range than others, just need to look at their specs and optimize which you choose to fit your recording situations.

[@petet66]

Thanks for another great video Curtis. What sort of specs should we look for on a mic to make sure it's capable of capturing a wide enough dynamic range for 32-bit float recording?

[@curtisjudd]

Dynamic range and max SPL are the two most important.

[@petet66]

@@curtisjudd thank you 🙏

[@PuthethuKollam]

Your videos are very informative. I would like to see a video on your comparison between Tascam X8/X6 vs Zoom R4. I hope the Zoom R4 will be on the sales only on December 2023.

[@curtisjudd]

F4 or R4?

[@PuthethuKollam]

@@curtisjudd sorry R4. It was a typo .

[@MaxoticsTV]

Hi Curtis. I couldn't hear any difference! Agree with you, it's all about the microphone and ADC. I'd like to make that point stronger. What words, or concepts we use, have significant implications. Not sure if the right words didn't come to you, or we have different understandings. You said, at 3:17, "it amplifies the signal using a preamplifier." A signal to me, is derived "information" from an electrical current. Signals are abstractions. Sure an audio engineer might say "amplify the signal", as shorthand, in a work setting, but not when it comes to specific electronics, like ADCs. I would have said, "it amplifies the current (or voltage) using a preamplifier. Next you say "makes it louder". There's no loudness is currents or signals. We certainly perceive some signals as being "louder" than others, but that has nothing to do with amplification (unless you're talking about brute sound pressure). Later you say "multiple ADCs per input." There is one and only one input (current) from a microphone. Every time you amplify/fork that current you add noise or lose accuracy. If those problems weren't difficult to overcome your high end audio gear would be a waste of money ;) As others have commented below, all the bit-depth in the world won't fix a current (high or low) where a good "signal" can't be extracted. The signal is what we're after! And we can't get a good signal if the microphone can't deliver a current that accurately mirrors the acoustic sound pressure changes it was subjected to. The old GIGO. Most audio clipping happens on the analog (current) end of the chain, where the sound pressures are overloading the microphones ability to differentiate changes in sound pressure though its output of electrons. This is the reason why, as you pointed out, NO SOUND ENGINEER is going to lose their job to 32-bit. Indeed, quite the OPPOSITE. Every producer who is burned by a sound engineer who relies on "32-bit float" to reduce clipping will find those engineers who understand what REALLY does what. Like you said, 24-bit fixed, or 32-bit float, no practical difference once you have your mics and pre-amp/ADC set up properly, optimized so you can derive the "signal" you want. If anyone believes anything I said is wrong, please correct me!!! Thanks!

[@curtisjudd]

Thanks for sharing your perspective. 🙏

[@etchd_]

Awesome video Curtis. Thanks so much. Sorry if this is already asked but say I’m using the Wireless go ii or even the Sennheiser G4s because they have a gain control on them before it’s sent to my Zoom f3. Would you suggest as a baseline to be safe in solo run and gun situations to just set their own gain levels at half way? Assuming the sound isn’t drastically too loud or too soft? Sorry just couldn’t find much information out there about using with these types of devices with their own gain controls

[@curtisjudd]

I'd suggest you do tests. The trick is that if you lower the gain too much, the noise floor will be more prominent when you boost the dialogue levels in post.

[@creemB]

The curtain on the sound is lifted and it feels clearer in 32bit float I don't know why the sound quality of 32-bit float continues to feel better to me. Do I just feel that the sound quality is better when the dynamic range is wide?

[@curtisjudd]

Interesting. I didn’t hear that.

[@JoeBaermann]

It’s the differences in loudness that tricks you, common perception: louder=better. The examples are ofc. normalized, but that doesn’t change dybamic volume in between. Any other difference in audio quality comes down to the mic’s and rest of the signalpath, not 32bit float itself.

[@ChrisFlores22]

I have the f3 and love it but it doesn’t have a dedicated TC input. Looking for an option that has a TC in. Is the f6 my only option? It seems big and way more channels than I’ll ever need. Thanks

[@curtisjudd]

The F6 has a sort of odd TC in/out port, but yes, it works. You’ll just need to figure out the right set of cables/adapters.

[@europeanambience]

Zoom F3 32Bit Recorder all the Way 😍I don't want to miss 32Bit anymore. So Fast Startup Time. Turn it on and press record and don´t worry about Gain Settings

[@curtisjudd]

neat.

[@UlliTorquato]

Thanks for the explanation! I'm an opera singer and have always recorded myself with Zoom H4n with the attached microphone. I would like to upgrade but am not sure if a Tascam 32 bit is a better choice. What do you think?

[@curtisjudd]

If the H4n is working well, I don't see a need to upgrade. but if you're finding that the audio distorts in the particularly loud passages, then the Tascam may help.

[@UlliTorquato]

Thank you so much!!😊

[@shareyourfare]

Hi there and thank you for comparing 24/32f. Let me just tell you,.that.there is a unit alled Boss Gigcaster 8,.tbat seems to do the same as the Rodecaster II pro/Mackie, but records in 32 Bit float. Just so you are aware, maybe this would be a great third unit to compare to the others. Thanksnand all the best from Vienna!

[@curtisjudd]

Thanks.

[@GadesChannel]

In my opinion (I use 32-bit float recording devices, F3 and X8, for almost two years, and now also Rode NT1 5th) there is no noticeable quality difference between 24 and 32 bit recordings. I feel rather the difference in ease of use, especially in the field recording. That's why 32-bit float is so handy :) And to be honest and as a programmer, I don't understand why 32-bit float might be less precise than 24-bit? It's much more precise!

[@curtisjudd]

Glad you find it useful!

[@ChrisFlores22]

Awesome video as always!!

[@curtisjudd]

👍

[@Glen.Danielsen]

Curtis, off topic: What think of the Hollyland Lark Max Wireless Lavalier Microphone System? I just saw a month-old review of it from Dunna. Thanks brother! :)

[@curtisjudd]

TBD. Hoping to review it in the next couple of months.

[@WobbleMatic]

Brilliant video especially now that Rode have released the Pro touting 32-bit float.

[@curtisjudd]

👍

[@paulg3388]

Your Videos are so excellent I can’t thank you enough. I have a question after seeing this. I have an Octava for my KZbin videos and currently it goes into my Zoom F3. I use an SM7B also plugged into the Zoom F3 for my Podcasts. If I am doing controlled KZbin talking heads and Podcasts, could I actually just go through my Apollo Twin with both mics and record directly into Audition for the Podcasts and Davinci for the KZbin videos without actually losing any quality? I got the F3 mainly for the Podcasts so that I could record them away from any sounds in my room (untreated for now), but I’m wondering now whether I’ve added something extra I didn’t need ?! Thanks for your professional opinion.

[@curtisjudd]

You could do it either way - I'd do some tests and see if you notice any difference in your situation. The Apollo has good preamps so that certainly won't be a step down as long as you manage your gain setting.

[@paulg3388]

@@curtisjudd ok great and thanks Curtis !

[@ThodorisKordolaimis]

one thing about 32 bit recording i think is that: lets suppose that you are recording a comedy act or a drama act and you are the only sound guy. thats mean that you are booming and you record the lav mics at the same time. Especially in comedy you come across with low spoken word volume and then some sudden bursts of really loud shouts. So if you are an one-man-show sound recordist you cannot manage easily these sudden volume changes. What was happening with a 24bit recording? you put the gain as low as possible so you wont clip your signal when the loud volume happens. The bad thing to this is that the low volume signal will need enough gain on post to match the loudness standards, introducing hiss noise. Now, with the 3bit recording you can have a higher general gain so the low spoken words are captured with a healthier signal to noise ratio without worrying about the cliping in the louder moments of the acting. To me that means less pre amp noise and healthier signal. This means also that you are prepared for the loud moments especially when setting the gain structure on your transmitters. What do you think about that?

[@curtisjudd]

Yes, that’s a use case that benefits from 32-but float.

[@MagDag_]

Thank you for the video! What is it about youtube compression of audio to 256kb for this comparison? I can hear clearly a difference. Do you have a link to the original files? BTW I saw the Kyama channel with very good quality sound, how do they do that? What is your opinion on audio codec vs processing (hardware) quality (for example Apple AAC vs LDAC Sony )?

[@curtisjudd]

Which sounds better to you?

[@MagDag_]

@@curtisjudd 32 bit, more richer.

[@honoluluhappenings5561]

Excellent analysis. Thank you.

[@curtisjudd]

👍

[@omnidivergence9846]

Thank you, that was very clear and helpful.

[@curtisjudd]

👍

[@alirezazimi]

I know the gain is important, but I think most of the mics reach their maximum SPL before 32bit extra dynamic range. so what is the maximum SPL that 24 bit can record? or what is the maximum volume that 24 bit can record?

[@curtisjudd]

It depends. The question that can be answered mathematically is how much dynamic range can be encoded with 24-bit. But even that differs at a practical level from recorder to recorder and mic to mic depending on implementation and max SPL and dynamic range specs.

[@vividmotion66]

I was talking to a friend today and he was assuming that once something is clips in a recording and other parts are not, you will always have problems mastering that audio in the end even it was in 32 bit float? Specially when it comes to compression. Is that though correct?

[@curtisjudd]

In 32-bit float, you should be able to pull it back down below 0dB FS and it should be fine, IF and only if it was not distorted earlier in the signal chain before it was encoded into 32-bit float digital.

[@vividmotion66]

Good points there, thanks!

[@curtisjudd]

👍

[@rhalfik]

Can we have 24 bit float? I think I've heard of it being used somewhere. There is surely enough DR in 24 bits for it.

[@curtisjudd]

There’s enough dynamic range in 24-bit linear for spoken work and singing vocals.

[@timothyawe5103]

Thanks for the awesome advise

[@curtisjudd]

👍

[@OrigEntertainmentOfficial]

Any chance you can do a video on upgrading lavaliere mics? We just use the mics that came with the kits. But I would be interested in getting some better mics.

[@curtisjudd]

I wish there were more mid-range priced microphones…but good suggestion for a topic. 👍

[@OrigEntertainmentOfficial]

@@curtisjudd honestly, I'd like to even learn about any upgraded mics. Hear the difference between kit mics and the upgrades. Especially under clothes applications. Maybe even add in some of the super low budget mics that are around $20 or so to hear the difference. Thank you, you rock!

[@curtisjudd]

@@OrigEntertainmentOfficial Here's a look at several pro-level lavaliers: kzbin.info/www/bejne/qonFaIKsYrhpmsU

[@curtisjudd]

@@OrigEntertainmentOfficial And a few more: SHURE Duraplex: kzbin.info/www/bejne/jmO6l5useduCq7M DEITY W.Lav Pro and Micro: kzbin.info/www/bejne/al7OfpKug5qmhqM

[@OrigEntertainmentOfficial]

@@curtisjudd Just saw these. These videos are perfect and thank you so much for all of these!

[@Beco-x2v]

Excellent video. What’s the post workflow? I’ve been recording 32 bit float for interviews but when I import the files into Adobe Audition I get 3 tracks that seem identical but I’m not sure that they aren’t capturing slight differences? Do I need all 3 tracks? Trying to find information about how to handle 32 bit float in your DAW

[@curtisjudd]

Most likely you have a stereo mix on channels 1 and 2 and the isolated microphone on channel 3. In that case you can just keep the iso channel - 3. You might also look into disabling the stereo mix if you don’t need it.

[@Beco-x2v]

@@curtisjudd oh my gosh thank you! I’ve been using a Sennheiser MKH 416 and a Zoom f6. Didn’t realize this could be a setting under record >> file format in zoom. So you can record 32 but float in mono?

[@curtisjudd]

@@Beco-x2v yes, indeeed!

[@Beco-x2v]

@@curtisjudd been struggling with this for months. You solved it in one comment. Thank you!!

[@curtisjudd]

@@Beco-x2v So glad it helped! Happy recording! 🎙

[@johnsilfen70]

What is important for making an adequate evaluation of these high resoulution digital formats is were is the limitations seated, and what kind of limitation do we have. First of all, the artefacts of these digital codings are large for very weak signals. So if we are going to be able to do listening tests that make these limitations easy to hear, we must use audio signals that arevrecorded way much weaker than what we normally do. Maby the max level shoud be at minus 100 dB or lower. athe we have to turn the volume knob way up towards maximun because the signal we have recorde is so weak. To have recording at or near 0 dB or near full level makes the listening test impossible to use fir giving a judgement on the acctual difference between the two formats. We need such avweak recording that only a few of the lowest bit arebused becausebitbis down there we find the limitations of theses systems. Second, it is of good use if we know what the limitations ate more precisely. For PCM which we have here the limitation is a noise floor at some level. For 24 bit PCM the limitation is a noise floor that is as low as -144 dB. This noise must be there for the system to work properly, and when recording and mixingmwe must make sute the noise is there and for this we add it, and its name is dithering noise. If we don't have a noise floor as such the PCM will instead give distortion. Quantisation distortion. athis is because digital code is using discrete values. And thos jagged signals are the same as distortion. With dither noise this quantisation distortion vanish and instead we have a noise floor. Noise is much better sinse we can hear very weak signals much lower than a noice. Here it becomes a noise floor were the signal can be lowered and it will drown in the noise when low enough but before this it is sounding as it should but with this noise that is fixed. So do listening tests with ridiciously low input when recording and crank up the volume and perhaps more gains adjustment when listening to the weak recording. And remember you are going to hear the noise I just described. You are very corrects about mics but we can add analig circuits through the whole path the signal takes before it is digitisized. The digital dither noice at -144 dB are way lower than any other part in the signals way to the ADC. But mics are often the worst of them.

[@curtisjudd]

Thanks for the insights

[@thesilencer6736]

Can the gain be set on zoom f3? I mean loud enough for direct monitoring purposes? I saw there is a volume button as well, but will it be loud enough for tracking oneself while recording?

[@curtisjudd]

Gain cannot be set on the F3 - one can only set the post fader digital levels.

[@thesilencer6736]

@@curtisjudd so basically it will output amplification?

[@LearnLightAndSoundSessions]

@@thesilencer6736it will boost the levels in the digital domain, yes

[@robhephner]

What stand is that MixPre II on at the beginning??

[@curtisjudd]

That’s a Sound Devices Pix-E stand

[@robhephner]

@@curtisjudd thank you sir.

[@robhephner]

@@curtisjudd I found it and bought one!

[@curtisjudd]

@@robhephner 👍

[@Chalisque]

32-bit float is basically the native language of the CPU when doing DSP. On the one hand, recording format is no substitute for properly managing the signal that goes into your DAC. On the other hand, once the digital signal comes out of your DAC, there is little point these days using anything other than 32bit float. When recording, the trick that is appearing is to use two DACs at different gain levels, and clever DSP to reconstruct a signal that is more accurate than either DAC can achieve on its own. I imagine in a few years a dual DAC with 32bit float output will be a standard commodity component. As a data format, 32bit float can represent any 24bit integer exactly, but the reverse is not true. Moreover, whereas a quiet signal in 24bit loses resolution, a quiet signal in 32bit float does not.

[@curtisjudd]

👍

[@WondefullComment]

If I want a pair of studio monitor does this affect to the quality sound of monitors ? I dont want to use microphone just know if bits are aplied to the monitors speaker while listening music for example

[@curtisjudd]

No, 99.99999% of monitors have analogue inputs so bit depth is not relevant to choice of monitors.

[@malz]

For me, it's not about the sound quality itself, it's all about the convenience. Whether you're an one man army or you use a help of less experienced crew member to execute the recording of sound on small-budget set, the thing of ignoring the sound input level is reaaaally handy. Thanks for the vid and clarifying some things!

[@curtisjudd]

👍

[@cas-cos]

I got a Zoom F6, mostly for the ambisonic capability, but also the 32-bit floating point. I decided to play with the 32-bit float setting on the 4th of July. Many booms at different distances, you know? Anyway, while the float wasn't clipping (naturally) I did notice that the mic was. So there is still an element of gain setting and mic choice to go along with extreme volume fluctuations. It doesn't necessarily mean "set and forget". At least in that particular situation.

[@curtisjudd]

Agreed! But on the F6, there is no gain control when recording 32-bit float. The knob becomes a post conversion level adjustment so it is important to choose your mic carefully for very loud sound sources like fireworks. Need a high max SPL spec.

[@PeterMossUkulele]

I think one of the big misconceptions is "32 bit float will save my bacon in all scenarios". Of course this doesn't happen for example when live-streaming, but a perfect tool for say editing a pre-recorded podcast if someone were to cough or spontaneously laugh out loud. Not preaching to you Curtis, just adding to the discussion.

[@curtisjudd]

Thanks, Peter. 👍

[@johnalmighty]

Yes, massive difference.

[@curtisjudd]

Really?

[@Joker69s]

I just started to make clips for tiktok to promote my business, but i can't see any difference between 32 bit and 24 bit.

[@curtisjudd]

If you set gain correctly when recording dialogue, there is no quality difference. The mic and recorder you use will impart a bigger quality difference.

[@chakerian]

TLDR: 32-bit float is simply a file format that records sound just like 24-bit PCM, but doesn't clip at 0dbFS. You touched on this Curtis, but the biggest thing to note is that 32-bit float is merely a format that can *represent* a wide dynamic range, but uses about the same *resolution* as 24-bit. It uses 23 bits. Normal 24-bit PCM cannot represent anything above (the arbitrary scale of) 0 dBFS... this is by design, and its dynamic range is ~146dB. At 0dbfs, all the bits are 1, so it can't represent numbers bigger even though the voltage from the mic can be higher than that. 32-bit float is less limited and can go far above what would be 0dBFS, and far above what would clip a microphone and far below where a mic signal would be lost in the noise floor. Using the 32-bit float file format, you shouldn't hear any difference in theory. 32-bit float has 23 bits of resolution "data", compared to 24 bits in PCM format, but can represent ~1528dB of range instead of ~146dB. Nobody has ears sensitive enough to hear any resolution difference between 23 and 24 bits, so it's no surprise that no difference was heard in this video.

[@robertulrich3964]

I was going to post technical stuff, but you beat me to it. the mantessa bit and file format are all that matter. 32 bit preamps are no better than high end 24 bit preamps. The dynamic range of a 24-bit digital audio signal is about 144 dB, while 32 dB weighs in at 192 dB.. you can still clip 32 bit. it's not magical by any means. it's more dummy proof for going over 0db, that's all. you can move the scale around and say in your head. I won't go over -20 db in 24 bit. the headroom is still around the same, relatively speaking. Your mic's built in SPL rating is going to clip way sooner than your 24 bit recorder...

[@chakerian]

@@robertulrich3964 You're talking 32-bit *integer*, not 32-bit *float*. It's easy to confuse the two, but when someone says 32-bit regarding recorders, they're talking about 32-bit float. I don't know of any recorder that records in 32-bit PCM, and the preamps are more expensive.. I think sound devices uses 32-bit A/D in their 800-series, but converts down to 24-bit. 32-bit float is very different than integer ... it's basically 24-bit with an 8-bit exponent that gives it the tremendous range. It can represent sound louder than anything that can even exist in the physical universe ... so it can represent sound louder/softer than any microphone can, and the file format is future-proofed! :)

[@robertulrich3964]

@@chakerian correct, while the file format is future proof, no such hardware exists that can record 1528db dynamic range!

[@curtisjudd]

I was hinting more at the possibility of creating a crossover network to recombine the audio from multiple converters as being a potential degrader of audio quality. But if it is degrading the audio quality, I can’t hear it with these particular devices.

[@chakerian]

@@curtisjudd That makes sense. There might be another test you could do. Poorly combining the outputs could result in gains/losses at the crossover point, too. Maybe test it with a known flat mic, and see what the white noise looks like across the spectrum to look for unexpected peaks or dips? And at different volumes it should move the peaks/dips (since it'll move which A/D converter it's using).

[@Pikmeir]

"It can be helpful for those situations where you forget to set your gain correctly" LOL yes, this is the main reason I require it as a solo video maker. It's also useful when recording someone else who doesn't understand how to speak into a microphone, such as during a street interview. I might have the gain set, but then the person goes too close to the mic and starts talking really loud into it, and the clip is completely ruined for a few seconds until I can move the microphone back away. With 32-bit I can still save it.

[@curtisjudd]

That'll still SOUND different when they're close to the mic - 32-bit float won't correct "distant" issues. But it will generally allow you to correct for going over 0dB.

[@Pikmeir]

@@curtisjudd Right, and it'll sound quite horrible either way when they're swallowing the microphone. Thanks for the video!

[@jasondavis4421]

For me i say it makes a differance but it cost me a fortune in equipment for it to make a difference.. allot of people use bluetooth headsets and say theres no difference.. the problem i have is with older music thats been upmixed ive heard everythind from papers being turned over.. breathing and background voices

[@curtisjudd]

Interesting, thanks!

[@platterjockey]

For music processing, it's a totally different matter.

[@MagDag_]

Finally! YEEEES!

[@curtisjudd]

👍

[@southgrove]

Is there any situation where it is a disadvantage with 32-bit float?

[@curtisjudd]

If you use a video editing app or digital audio workstation app that doesn't support 32-bit float, then that's a bit of a problem. But otherwise, for spoken word audio, it doesn't appear to be a problem from my perspective.

[@johnmellor932]

The only disadvantage I've found is that video editors just don't understand it. I get back AAFs from video editors and they've ingested the audio into a 24bit project and inadvertently clipped all the loud audio. Even after I've explained it to them. They still dont get it. So then you've got to manually find all the original recordings to get the peaks back. I cant use TC because despite jamming with the camera, they dont understand TC either. They use had-hoc methods they pick up on youtube videos to sync sound such ss using the multicam method which strips out all of the meta data.

[@hbp_]

The best thing about 32-bit float is that ebay, FB, etc are now full of used 24-bit integer audio recorders that are rendered useless and obsolete by this new technology. I got a MixPre-3 for cheaper than what I paid for a used MixPre-D a couple of years ago.

[@curtisjudd]

Hahaha! Good point!

[@hxhdfjifzirstc894]

Does anybody know of ANY movies that attempt to have a somewhat natural dynamic range (within reason)? That is to say, gunshots for example, would be very very loud, but not loud enough to damage the audience's hearing... or the camera could be moved farther back, in order to lower that sound in a realistic way.

[@curtisjudd]

My sense is that films don’t generally try to be realistic, they try to suspend the audiences disbelief. Foley, in particular is seriously interpretive and often not particularly realistic. Gunshots are often made to sound MORE impressive and BIGGER than in real life. So in short, perhaps documentary films sound realistic, but mast narrative films are intentionally contrived.

[@hxhdfjifzirstc894]

@@curtisjudd Oh, yes... I was asking from the perspective of someone who has a trained ear, and is acutely aware of the general lack of realism within film audio. I was just hoping there's some undiscovered indie genre that specializes in more realistic audio, level-wise... but I'll settle for no ADR, and no Foley 12 gauges being racked, when someone picks up a pistol. Apparently, theaters are all calibrated to 105dB peaks, so that would be the place to start (not using any perspectives that would have a realistic level of higher than 105dB due to proximity of a particular sound). Even a small .22 caliber pistol will instantly make your ears ring, if fired indoors (so the camera would have to be looking through a window, or something... and characters might have to suffer temporary hearing loss).

[@redcat.]

Well I hear a difference in your examples but it's not like a difference in quality - meaning I can't really say that 32float sounds worse, but defititely it sounds different. It's more like you compare two different recording devices and that's it. So I don't think the difference is so important, expecially if you make some post production anyway. However I noticed one common thing in this comparison for all 32bit examples - the bass sound not so reach compared with 24 bit and also it seems like there's some little tiny things in the midrange (all 32 bit sound more like they have some kind of barely noticable presence boost) but I could be wrong - it's just my first impressions for what I'm listening here but being a pro post-production sound guy for about 12 years I notice that. Anyway thanks for the video and examples. It's definitely a great tool but for the dialogs I think it's better to stick to 24 bit for most of cases (esp. talking head)

[@curtisjudd]

That's what I do - 24-bit for dialogue, especially for scripted dialogue where I generally know what's coming.

[@redcat.]

@@curtisjudd I would prefer 24 bit too for controlled scenarios

[@LawnD4rt]

For me, it's not about the quality it's about having more data. Like shooting images in RAW vs shooting Jpeg.

[@curtisjudd]

But ultimately, shooting raw is so that you'll have more to work with and achieve what you want to achieve with the image, which, in the end, is about quality?

[@ciinelog]

Sennheiser MKE 600 and Zoom H5 can handle 32bit flaw?

[@curtisjudd]

The ZOOM H5 does not. You CAN use the MKE600 with recorders which support wide-dynamic range 32-bit float recording.

[@ciinelog]

@@curtisjudd Thank you Curtis!

[@georgeogrady449]

Demanion off sound how loud is that

[@curtisjudd]

I'm sorry, I don't understand your question.

[@winnegehetoch]

Re the argument: “A skilled person like you can set the gain on a 24-bit recorder [as long as no surprise levels turn up] and make recordings that sound every bit as good as, and often better than, a person using a 32-bit float recorder who doesn’t know how to optimally position a microphone.” Surprise, surprise… comparison of microphone positioning skills? Why not compare apples to apples. Assume both cases has the same knowledge on microphone positioning. Or same lack of knowledge. Otherwise, what are you comparing? Intelligence/experience and lack of knowledge/stupidity? What does that have to do with use of 24-bit versus 32-bit float? If the person doing 24-bit does not know… etc. would you accept an argument comparing the “dabbler” to a knowledgable person using 32-bit float (and not overloading either the microphone or input circuitry)? In controlled environments equally skilled person with similar quality equipment should be able to reach similar results, whether using 24-bit PCM or 32-bit float, given decent settings and gear ability. What a surprise. In uncontrolled environments - especially run and gun in public places, where anything may happen at any moment, without any warning, and completely surprise you in the midst of videographing an interview, event or… whatever. As long as the microphone is not overloaded (can happen), and your recorder input is not overloaded (32-bit input circuits are ALSO limited on that front, but in most cases far more forgiving), however carefull you have adjusted the input levels, the chances are, that 32-bit float recordings will still be salvable, but the 24-bit recording heavily distorted, if levels suddenly skyrocket. Surprise, surprise. The 32-bit float system used correctly by a decently skilled person will have clear advantages compared to the same person using a 24-bit system correctly, in virtually all cases, where the unexpected may and can happen. Period. It’s like a travel insurance policy. Is it bad to take out e.g. travel insurance? If you don’t travel, certainly. In most cases it was unnecessary even if you travel. The only problem is, that you do not know beforehand, if decent and comprehensive travel insurance cover will be needed. Now, my argument is: “Look at 32-bit float as a kind of travel insurance”. If you never travel outside your studio, outside well planned and controlled environments, the 32-bit float “insurance” is not needed. But if you venture out into “real and unplanned, uncontrolled life settings”, you’ll from time to time need all the “insurance” you can get. Period. Maybe now is a time to start debating the value of backups (either data or on the current theme extra microphone, recorder, camera etc)😇 Regards and a biiig smile

[@curtisjudd]

The main question in my mind had more to do with the quality of combining the output from two or more converters. In short, of those samples here, it didn’t seem to be a big factor. I agree that 32-but float, when combined with a quality microphone and preamp, is a useful tool or insurance policy, I don’t believe anyone should put all their projects - free or paid - on hold until they can upgrade to a 32-but float recorder.

[@georgeogrady449]

If mixer 40bits why not build 32 to 40 bits system

[@curtisjudd]

Why build a 40 bit system when 32-bit float is more than enough to capture well beyond anything humans can hear without totally disintegrating?

[@rodrigoroaduterte9415]

I am not an expert in any dimension, but 32-bit sounds “deeper” to me where low frequencies are being presented better.

[@curtisjudd]

You have very good ears, better than mine 😉

[@gunvlog]

On my experiance, I have the feeling, that 32 Bit Float is quite a bit quieter then 24bit. At least when using simple 32 BIT recorders like the Zoom F2. In order for YT videos to get in the acceptable loudness area of what YT needs to get decent volume

[@curtisjudd]

You boost the audio in post - loudness normalize. In fact, all well produced videos are loudness normalized.

[@johnmellor932]

I find that some location sound recordists have adopted strange working practices. They pride themselves on never using limiters because they claim it's 'unprofessional' to do so and that if you use limiters you don't know how to gain stage properly. What I find funny is they don't seem to have a problem with us post guys clip gaining sometimes up to 20dB just to get dialogue to a good level. It's the opposite end of the spectrum of bad. Recording a healthy level and using a limiter is essentially the same thing, just the other way around. That to me is the difference between a trained sound engineer who went to sound school as opposed to someone who learned on the job as an apprentice. There's nothing clever about recording a piss-poor signal just so you don't have to use a limiter. It's misguided and plain stupid. Use a limiter or better yet 32bit.

[@curtisjudd]

A limiter and staying on top of the script is really the job of a production mixer.

[@DanLiveOnUTube]

I thought DAWs are processing in 32bit float. If that's good enough for 99.99999999999% of those who mix in the box I would say it is good enough :) Also per your earlier advice I now max my gain on my mix pres to take advantage of the preamp warmth :D

[@curtisjudd]

Most DAWs process in 32-bit float. But that's sort of a different discussion than what we're covering here - this is about gaining a bit more dynamic range, even if you set the gain incorrectly.

[@JGnLAU8OAWF6]

Many DAWs use 64 bit floats internally.

[@DanLiveOnUTube]

@@curtisjudd I am talking about the internal sample rate (read:bit depth) of most daws but I was wrong, Reaper (which I use) process audio at 64 bit fp which made my point moot, my argument is that if daws are mixing audio in 32bit fp then it's it should be a acceptable precision to record in.

[@DanLiveOnUTube]

@@JGnLAU8OAWF6 HA! you got me before I could post my reply, I had to look it up, don't know why I thought 32 bit float.

[@curtisjudd]

@@DanLiveOnUTube Also, sample rate and bit depth are two separate things. 😉 Nevertheless, 32-bit float depth is just fine in most cases, even for recording, to your original point.

[@乾淨核能]

Thank you for explaining the 32-bit float audio recording in such detailed and easy- to-understand way!!!

[@curtisjudd]

👍

[@helmanfrow]

I thought we put this issue to bed already.

[@curtisjudd]

For myself, I have.

[@helmanfrow]

@@curtisjudd Ha for sure, but I recall a video about 32-bit not too long ago. Maybe it was on another channel. I'm starting to forget details in my old age!

[@TonyBabarino]

Haven’t seen Dr. Judd’s video for a while…sporting sexy-quality thumbnails now I see! 👍😊

[@curtisjudd]

👍