32 Bit Floating Point Explained - Sound Speeds

Holy ** how long did you take to edit and animate the video? That's a crazy amount of info in that video. Big 👍

They're going to need a new warning sticker on boom poles.... "Warning: do not use boom pole outside during thunderstorm."

thanks for simplifying, clear as mud.

THIS VIDEO WAS SO GOOD!!! :D please make more.... i love how much detail you add to your content <3

9^9?

Edit: Great video man! Back to the original comment: 32 bit float. It's great for any high action and high dynamic sound mixer jobs where you need to be recording but also boom operating at the same time. You can record hands off from the input gain and then convert the 32-bit floats into 24 bit before you deliver the sound because if not the film editor people will be nagging cause they don't know how to convert a 32 bit float to a 24 bit. I wish I had a 32-bit float recorder that way I could use it in these types of situations.

I do have question..me and my band..three of us.. are in our studio using 24 bit recording..for vocals with music is that perfect? or should we be at 32 bit float? the reason Im asking is we do alot of cover tunes..for practice... some mid and highs.. vocal wise..
John- 5manshockwaveband..

excellent man!!

32-bit is mainly useful in the box because it can contain any level with the same accuracy without clipping. If you applied an envelope, the noise floor would "float" with it could be inveted with minimal loss. 64-bit is used in plugins such as equalizer or reverb for adding extremely small numbers to a bigger number many times and not losing them and causing filters to become unstable. It only exists in Reaper because it can and to satisfy users who peek at the signal with an analyzer. When mixing multiple chanels of sound, they are all approximately in the same range, and 32-bit is plenty. Now we have even more crazy bits in Kirchhoff EQ, where you can't even see a noise floor.

I doubt that a box like Zoom can exceed 140 db of dynamic range. Must be some space technology.

Can you repeat please

Great education, thank you! Raises a question about the Zoom F series.

The F6 lets you adjust gain to set levels in 32-bit mode. It also offers 24-bit recording.

F3 is 32 bit only, and documentation suggests there is no support for setting levels. You only control monitoring output or fade or something? So the gain is fixed?

With 32 bit, you don't need to set levels, woo! 😝 You dispelled that notion. Without proper levels, you don't get 24-bit quality in the standard DB range. That's my takeaway.

F3 is a nicely portable unit, but 32 bit only, and if you can't adjust gain to set levels ...

Hope I haven't misunderstood the zoom F series specs. Would you happen to know? Thanks so much.

Great explanation! I usually relate it to the screen resolution of video.

One Volt of current.

I've always wondered what happened to Albert Einstein's grandchildren.

which is better?
16 or 24bit?

9!^(9!)

yooooo this is so creative and cool!!!!!!

Excellent video! Great thanks!

With most all audio interfaces I can hear the transients are not being recorded clearly, even at 24 bit 96khz. But when I use my RME Fireface II I can hear that it does a lot better job with the transients of my voice. Now, no one I have ever talked to can hear bad reproduction of transients other than myself. But I hear the bad transient reproduction with all of the major audio interface brands I have tried, other than RME. The transients are not peaking and distorting, it is like the interfaces simply cannot focus on the beginning of sounds and reproduce them right. Why am I the only one that can hear this? It is like a variable recording rate is needed, like as in film with large explosions and fast motion to throw more bits at the beginning of words. Would 32bit floating point throw more bits at the beginning of sounds and make them more focused and clear? All I ever hear about 32 bit is that it can record louder sounds, does it not also record finer steps, more densely packed bits, and would this not improve sound quality as well?

Hi. Very informative video! Is this your most indepth video on ADCs and what the layered 24-bit aspect means? Have any higher bit depth ADCs come about since this video was posted?
I'm still trying to learn if people need to learn the details when saying floating-point field recording is possible.

Amazing.
The best about your content is that we get to talk real data and facts on our breaks tehehehe
Greetings and thank you for this, truly!

The present 32bit float follow some standards representation of float numbers but unfortunately also blocks any attempts to improve the quality of recording with out rewriting the present "standard" 32bit float for audio.
Had a 32bit float format been thought out carefully it would not had been sign, 7 bit exp., and 24 mantisa - it would had been a signed 3 to 5 bit exp. (4 to 6 bits used) and a 28 to 26 bits mantisa.
Remember that the exp. is a number telling you how many places you need to move the decimal point - 3 bit plus sign would be +/- 7 times !!, or like multiplying or dividing the 28 bit value in the mantisa by one million, or +/-60dB ....... 120dB increased dynamic range.
Add to that the 20dB the extra 4 bit (24 to 28 bit resolution) would give and you have 260dB of real world dynamic range with 150dB real world resolution !!!!!!!!

Keeping strickt to the binary roules of float numbers it may not be more than 104 dB extra dynamic range - but 225dB real world dynamic range with 150dB real world resolution .

Using present day 24bit's ADC's it would "only" give you 190-200 dB dyn. range with 125dB resolution - only very specialyzed microphones and sensores will challenge this.

Nice work. Thank you.

the more i look into this, the more it confuses the shit out of me! i think i still don't get the relationship between recording bit rate, and the "bits" regarding floating point
i thought floating point was in relationship to internal volume processing (volume hitting plugins, volume above 0.0dbFS inside the daw, etc.)
i am completely lost now that you said one can record at 32bit floating point? is this regardless of an audio interface being only being 24bit? are there soundcards that record at 32bit?
or are we using the terms "24, 32 & 64 bit" for multiple things?

Great video!!!

Hey Allen,

I've been watching your videos for quite some time now. I absolutely love your channel and the diligent work you put in to deliver valuable and accurate information to us.

So, here's my question...

Which sources did you use to create this video? Do you have any books or websites to recommend to budding music/sound professionals that also have a passion for the scientific aspect of sound?

Thanks, keep 'em coming! :)


- Stefan

I studied Computing in college. Wasn't very good but there are some important things I still remember. One of those things is information about how floating points work because I also studied Computer Systems Engineering, which is something I was actually good at. I came to the same conclusion about 32 bit float and the accuracy of the data. You basically confirmed my breakdown, thanks lol. Great and Informative video as usual.

whew Chile ! that was a lot .

My head hurts 🙁

What makes you claim 1V analog audio signal should represent 0dBFS? In professional world we use +4dBu level which equals to 1.2Vrms. That old agreement does not say anything what headroom to 0dBFS should be.

But why not use 32bit integers for comparison. Then it would be a fair comparison since both will use 4bytes per sample.

32bit integer has range of -2.15 billion to +2.15 billion

32 bit float has definitely proven handy when doing a one man band shoot.

So is 24bit 48kHz lossless the top of the line sound quality for movies, TV, and Video Games or will they come out with something better in a few years?

That was fantastic. I’m still using 16b recordings and ultra happy with them. But now I know what floating means. Yay!

Excellent information as always!! Thank`s for your channel!!!
By the way, can You make a video how to record a surround sound on a stage. For example room or street tone, or so. There`re obvious ways, but may be You`ll tell some tricks as U always do on yhis channel.

Thanks!

That was a great class.

I’d the Zoom F2 worth it? It records in 32 bit? I need a small portable lab recorder

Ffffuuuuuuuuuuuccccckkkkk!!!!!

9^9

WOW... all that video and math was a LOT to absorb bro! QUESTION.... I have a recording studio and I'm creating a Podcast/Vlogcast network. I do not want to use the typical Shure SM7B mics. i want to use all Lav mics. And afterwards I can great outboard gear and software to EQ the podcast vocals to sound damn near as good, if not better than using a condenser mic or shure sm7b. Because I want to do it this way, do you think having the Lav mic with 32bit float ability is the best way to go?

very, very informative. now I know to keep my levels similar to what I would for a 24bit recording if I'm recording with 32float, since that will reduce unnecessary noise(even if it's not normally noticeable). that's probably a really good takeaway for field recording/sound design.

I really love that your videos are always full of knowledge and straight to the paint!!!!!

9!^9!

It was great.

Ok now am sure am really dumb 🤣🤣🤣🤣. But I get the main point of this nice video ! « Slightly technical side » lol ..... this statement is slightly underrated

Thanks for the video and to @podcastage for sending me over. This was awesome.

Wonderful video. Lots of information. Some may need to listen few times to grab all the information. Thank you for your time and effort put into this video. You deserve better ratings.

Funny how there are now interfaces that are 32 bit integer, but we still haven't achived true 24 bit integer. So the 32 bit integer interface is still under 144db dynamic range. I think my interface records at a true 20 bit integer despite the 24bit integer marketing.