The Zoom F6 and 32 bit recording

[TBman]

Can someone explain the importance of 32 bit recording?

lkingston in another thread here touched on the subject while recommending his Zoom F6.

Thanks.

[KevWind]

Originally Posted by TBman

Can someone explain the importance of 32 bit recording?

lkingston in another thread here touched on the subject while recommending his Zoom F6.

Thanks.

If you are referring to bit rate (like 32 float bit 48khz sample as opposed to 24 bit 48khz)
The actual "importance" has been debated a fair amount . Theoretically the higher the bit rate, the more headroom before digital clipping and it adds up exponentially.... SO 32 bit float does have huge head room (as lkingston mentioned you don't have be as diligent about gain levels ( but that also means you can drive the mic into analog distortion with more gain ) .
So you can get to higher level signals and more dynamic range without clipping.... The question is can you really use all the added headroom and possible db level ?? And considering most converters have 125 db (max) having 700 db of headroom doesn't do you much good

So many claim in practical terms 24 is more than sufficient for most audio productions..

Then also in practical terms for things like uploading to streaming platforms 32 bit over 24 bit will not do anything in terms of being able to upload louder files Because for example for Youtube any thing above -14 LUFS will be compressed by Youtube anyway

[Rudy4]

This fairly recent Zoom F6 topic had a lot of information:

https://www.acousticguitarforum.com/...d.php?t=638552

[rick-slo]

No reason to record music in 32 bit (unless perhaps you like spending money on SSDs). You don't need anywhere near that large of a dynamic range. Your 32 bit words will be mostly zeros. You will end up with large files, more computer latency due to larger buffer sizes needed and more likely clips and pops if you computer can't keep up with it (especially with effects added such as reverb). More likely dither will be needed (which has its own issues) as you down size the files to lower bit rates.

[TBman]

I see Sweetwater has the F3 with 32 bit also.

[ChuckS]

Originally Posted by Rudy4

This fairly recent Zoom F6 topic had a lot of information:

https://www.acousticguitarforum.com/...d.php?t=638552

It isn't about just using 32 bit float, it's using 32 bit float in combination with a different type of ADC design. Some patents have expired, so we are now seeing this technology being used in a number of device designs.

As I posted in the thread referenced above:

Interesting features for a field recorder; using dual AD converters with auto-ranging /auto-scaling, and 32 bit float. Don't have to deal with gain staging, and don't have to worry about clipping. Still, I wonder how optimized the analog will be (i.e. noise) prior to the ADC and how much of the full-scale range of the ADC actually gets used under various scenarios.

Some interesting comments, in particular post 15:

https://gearspace.com/board/mobile-a...l#post15635623

[Rudy4]

Here's a video presentation done by Curtis Judd that focuses on the specifics of 32 bit float audio. What it is, and pros and cons of using the format:

[lkingston]

To me, the advantage of 32 bit float is not having to set my levels closely while recording. In music, the advantage is mostly for recording live music where you don’t know exactly how loud it’s going to be. It’s not like you are going to be able to tell the difference between well recorded 24 bit and 32 bit audio if the levels are set properly on the 24 bit. They are going to sound the same. It’s just in being prepared for recording somebody who does very timid level checks, then belts out the song when you go to actually record. It gives you oodles of headroom so that you don’t lose a great performance to not having set the levels right.

As far as taking up extra space, you can easily get around that by just normalizing the 32 bit audio, then rendering it at 24 bits.

Much of the time it doesn’t matter. Other times it can save the day.

Let’s imagine you are recording a live show and you have a variety of performers. You just want to start the audio recording, and mix and work the video camera. Some performers are really soft. Others are belters. 32 bit float will give you a recording that is as good as if you had meticulously set the levels perfectly for each performer.

32 bit audio isn’t that big a deal in recorded music where there is some predictability to the audio levels. It is absolutely huge in the world of movie audio for recording dialog. That is because actors might do a line three or four times as loud in a real take as they might in rehearsal. 32 bit audio has no problem with that.

My perspective is that live performance isn’t that different from live dialog. Sometimes singers really get into it and get way louder than you expected when you set the levels. Sometimes, they get all shy and are way softer. 32 bit audio protects you against unexpected levels.

It can handle everything from a hummingbird wings fluttering to a gunshot next to the mic. Yeah, it’s overkill, but sometimes overkill is exactly what you need!

[lkingston]

As far as the F6 or SoundDevices recorders go, they both have fantastic preamps. I don’t bother with a CloudLifter if I am recording to the F6, even with distance miking from a SM7B. There simply is no noticeable added preamp noise, even with large amounts of added gain. This is true at both 24 and 32 bit.

[DupleMeter]

Watch this video - BJ Buchalter (Metric Halo designer/engineer) gives you the real details about 32-bit float vs 24-bit fixed & why 32-bit float isn't as good as you might think.



https://www.youtube.com/watch?v=M_14...geMODR&index=2

[lkingston]

There was nothing in that informative talk that scares me away from recording a live performance in 32bit float.

Here’s the thing: 24 bit sounds wonderful, but if you record it with too much headroom, you aren’t really getting 24 bits. You are getting 20 bits or some lower number. With 32 bits, you are getting the full 24 bits of resolution regardless of where you set your levels.

In the sort of live performance capture I am using this for, there isn’t the heavy (and problematic) calculations that he says can be a problem. I am recording some live tracks, then adding maybe a touch of reverb, compression and EQ. That’s it. It’s not like I am using it for a DAW, where there might be realtime audio stretching and pitch correction.

If you want to avoid this sort of problem and do heavier audio processing, there is a pretty simple solution: just normalize the recording to 24 bit depth and render 24 bit audio files.

The double precision 64 bit float audio he also talks about is also interesting, but you can’t go out and by a double precision 64 bit float Zoom recorder that does two tracks for $350 (Zoom F3) or six tracks for $700 (Zoom F6).

My point is that an F6 will give you 24bit precision at unpredictable volume levels, and that can be super convenient when you are recording live music.

A 24bit recorder will give you precision at some number less than 24 bits, depending upon how much headroom you are using.

Another thing that is really cool about the Zoom F6 are the digital limiters which do a sort of digital look ahead to avoid clipping while recording a 24 bit audio file. This is in comparison to the analog limiters in the Sound Devices recorders. Which type of limiter is better is a matter of opinion. The analog limiters are more expensive to implement, and have a theoretical noise floor advantage, but the noise floor is so low in the F6 that I don’t believe it actually matters.

If you use a Zoom F6 in 24 bit mode with the digital limiters, you can be a bit braver with your input levels than you normally would with a 24 bit recording. You can also do that with the analog limiters of the Sound Devices recorders of course, but for more money, especially when you factor in the extra battery holder attachment that you need with the Mix Pre 6 ii.

[jim1960]

I think this guy explains the 32bit thing in terms that are fairly understandable. In part two, he explains who would benefit from 32 bit and who would derive no benefit from 32 bit.

[Doug Young]

Originally Posted by jim1960

I think this guy explains the 32bit thing in terms that are fairly understandable. In part two, he explains who would benefit from 32 bit and who would derive no benefit from 32 bit.

I just want his -102db noise floor!

[lkingston]

Originally Posted by Doug Young

I just want his -102db noise floor!

Me too! Lol!

[lkingston]

Actually, it’s a bit more complicated than that. I don’t understand it well enough to explain it properly, but on the Zoom and Sound Devices recorders, they are using two AD converters set to different scales of dynamic range. When one is overloaded, it automatically goes to the other. This happens throughout the dynamic range.

The practical result is that the noise floor varies along with the level.

Here’s how I look at it. Sound pressure levels are exponential. To record something very soft, you need to be measuring very small increments of air pressure changes. Whereas to record something very loud, your increments of air pressure changes become orders of magnitude larger.

The way this is done with linear 16 or 24 bit recording is to have an adjustable preamp going into an ADC (analog to digital converter). The sound is scaled in the analog domain to something that fits within the range of the ADC.

With the 32 bit float design, the approach is a little different. It is designed to work over a very broad dynamic range. With low levels of audio, the increments of volume differences recorded are very small. At loud volumes, the increments of volume differences are very large. This is the key.

The two ADC circuits give you a wide dynamic range where there still is a noise floor, but it fluctuates along with the volume.

This same principle is also being used with light in some of the new camera designs. For example, I have a new Panasonic GH6. It has the same kind of dual sensor design that can handle a much wider range of light levels, from dark shadows to bright sunny skies. Yes, there is still ISO noise, but it scales up and down with the amount of light, so it is much less apparent.

It is the same with the F6 or Sound Devices recorders working with 32 bits. When you record at 32 bit float, you don’t set the input levels. You might be recording a whisper. You might be recording gun shots. It doesn’t matter. There is still the same white noise floor limitations, but the cascading ADC design means that that floor is scaled up and down along with the audio levels, and not anywhere near the problem traditional audio design would lead you to expect.

If you record something really soft, it is going to sound like you set your levels to record something really soft. If you record something really loud, it is going to sound like you set your audio levels to record something really loud. It is amazing, but it does seem to work.