The Zoom F6 and 32 bit recording

[lkingston]

Originally Posted by Doug Young

Also, depending on your DAW, you may not even be working with 32 bits, it may get converted to 24 when you import the file.

Yes, but it usually normals first before it does this conversion, thus it is still an advantage.

[TBman]

Originally Posted by Chipotle

Don't confuse the number of bits used in the *Operating system* with the number of bits used to record.

Thanks, but I wasn't. Just specifying what version I used.

[Doug Young]

Originally Posted by lkingston

Yes, but it usually normals first before it does this conversion, thus it is still an advantage.

Right, all I was addressing was the issue of the DAW not being able to handle the larger files (which seems unlikely - a single stereo file shouldn't be a problem for any reasonable computer). If the file's converted first, then the DAW doesn't have to work with the larger file anyway.

[DupleMeter]

Originally Posted by lkingston

Let’s say you record the same audio on two channels of an F6 or Mix Pre, but one is at mic level and the other is at line level. Then you normalize both so that they are played back at the same level.

The differences between the very large numerical jumps in the line level recording vs the much smaller numerical jumps in the mic level recording will be the same size after normalization of both tracks. There will be neither more more less precision between the two. This is a factor of the mantissa being 24 bits in both cases. Either track could have been recorded at 24 bits after the input was scaled to the proper level.

The advantage of 32 bit float isn’t an increase or decrease in precision. That is the same as it would be had it been recorded at 24 bits with the level properly scaled in the analog domain first.

Rather, the advantage is in being able to record at this 24 bit precision WITHOUT knowing where this range would fall exactly ahead of time.

Again, not quite accurate.

Those 24 bits are only moveable to account for the “limitless headroom” of 32-bit float through logarithmic scaling, and that comes at a cost. It’s basically the equivalent of riding a fader. As you move down to keep from distorting the converter, you are moving across the logarithmic scale. So each move makes your smallest change a little larger than the last. If you really screwed up a set your gain horribly wrong, it will get noticeable really fast.

Go back to that video I posted. BJ explains it very well. He’s one of the smartest guys in audio today.

And again, all this is useless if your analogue stage overloads before the converter. And, it’s far more likely that you’re analogue path will have less headroom than your converter (which is 144dB at 24/32 bits).

I’m not saying not to use 32 bit. Just understand the reality vs. the marketing of the technology. Liked most marketing, it’s being over sold.

[lkingston]

Originally Posted by DupleMeter

And again, all this is useless if your analogue stage overloads before the converter. And, it’s far more likely that you’re analogue path will have less headroom than your converter (which is 144dB at 24/32 bits).

But 32 bit float recorders don’t use a single ADC. They use dual ADC converters set to different volume ranges. When the first ADC converter hits the top volume, the sound is sent to the second ADC. This sound which goes to the second ADC is going through a second preamp with a gain structure set for that volume range. This is why the noise floor is so low on both low and high volume signals in spite of not being set for a particular recording.

I agree that BJ is way ahead of me on almost everything audio, and I admire his work, but on this one particular topic, I am absolutely sure of my position.

[lkingston]

Originally Posted by Doug Young

Right, all I was addressing was the issue of the DAW not being able to handle the larger files (which seems unlikely - a single stereo file shouldn't be a problem for any reasonable computer). If the file's converted first, then the DAW doesn't have to work with the larger file anyway.

Here is my experience with this. Before I bought the Zoom F6, I downloaded some test 32 bit float audio files from the Sound Devices website. On the test files, there was audio recorded in three volume ranges within that same file. When I imported it into my iPad, it imported, but it did so as though it had been recorded at 24 bits. The soft recording was almost inaudible. The mid level sounded right. The high level hopelessly clipped.

On my PC, I could scale this, but I couldn’t on the iPad. I have been doing most of my editing on the iPad Pro because I like the form factor and touch screen so much, and because for the simple edits I do, it mostly works fine.

I exchanged a bunch of emails with the developer of the iPad apps, telling him that he should fix his apps so that you could change the multiplier component of the 32 bit audio files. He replied that it should work. That his apps (Ferrite Pro and Hokusai Audio Editor) calculated that number on import, and that it should show up as though it was normaled 24 bit audio.

I bought the F6, thinking I would use it mostly as a 24 bit recorder. But when I tried importing 32 bit float audio I had recorded myself, it did exactly what he had said: it imported as though it was normaled to 24 bits. No, I didn’t have control of the exponent component, but I didn’t need it.

It turned out that Sound Devices had somehow set the exponent part of the file so that it would read as though the mid level recording was the primary one. That way you could have the experience of bringing up the soft recording from the noise floor, and rescuing the loud recording from it’s initial clipped sound.

In real life, that isn’t how it works. When the file is imported, the exponent part gets set so that the audio reads like it has already been normaled. This is incredibly convenient!

As it stands, 32 bit float is very workable with the iPad Pro and almost every audio or video app. It works in the Luma Fusion video editing app as well. The Apple import routine figures out the exponent component, then imports the audio as though it was perfectly recorded 24 bit audio.

I like using 32 bit float just because it makes setting levels so easy and risk free, and because I really just want to concentrate on my performance. No it doesn’t sound better than 24 bit audio with the record gain structure properly set. It’s just an extra tool that can make life easier when you don’t know exactly how loud something is going to be, and you want to make the recording process easier.

[Doug Young]

It sounds like, from your description that there is a potential issue with converting 32 to 24, depending on how the conversion is being done. I assumed that wasn't a problem. I'm confused from your experience tho. Importing to a 24 bit app works right, or doesn't work right?

Keep in mind that all Barry is doing is recording solo acoustic guitar, there isn't a lot of dynamic range. It's pretty easy to predict how loud a fingerstyle guitar instrumental of the type Barry plays will be.

Me, too. I set the gain on my MixPre once, when I got it, and haven't touched it again. I actually haven't tried the 32-bit mode, I set mine to 24 since that's what I'd be working with anyway - Logic doesn't support 32 bits, so I've not bothered. I can see that not having to worry about gain levels would be nice for field recording or working with a lot of different sources, but for my home use, making You Tube videos of my own playing, it's never been an issue.

[lkingston]

Originally Posted by Doug Young

It sounds like, from your description that there is a potential issue with converting 32 to 24, depending on how the conversion is being done. I assumed that wasn't a problem. I'm confused from your experience tho. Importing to a 24 bit app works right, or doesn't work right?

It works right. It was only in the test files off the Sound Devices website where there was a problem, and I believe that this was intentional in order to give the potential buyer the experience of being able to save a recording recorded too soft or too loud.

Keep in mind that all Barry is doing is recording solo acoustic guitar, there isn't a lot of dynamic range. It's pretty easy to predict how loud a fingerstyle guitar instrumental of the type Barry plays will be.

Me, too. I set the gain on my MixPre once, when I got it, and haven't touched it again. I actually haven't tried the 32-bit mode, I set mine to 24 since that's what I'd be working with anyway - Logic doesn't support 32 bits, so I've not bothered. I can see that not having to worry about gain levels would be nice for field recording or working with a lot of different sources, but for my home use, making You Tube videos of my own playing, it's never been an issue.

I agree. Setting levels on my own recording isn’t that difficult either.

I had an experience a few years ago however which really rattled me. I was recording a concert at a church. The guest artist was one of Ray Charles’s daughters. She did a sound check at a modest level when I set the record levels. Then during the concert, she completely blew past those levels. I brought them down on the audio console for the live show, but I completely forgot to change them on the recorder.

Later when I checked the recorder, it was completely distorted. Fortunately I was able to save it using iZotope RX’s peak restore (which is another story entirely), but it really scared me and made me wary of setting levels manually for live recordings…especially when you are also doing several other things at the same time.

iZotope RX saved the day, fixing the sound to the point where nobody complained or noticed that anything was wrong, but it was still a compromise that I absolutely do not want to have to rely on!

I have been paranoid of this since then. Hence my interest in 32 bit float audio.

Had I had my F6 and 32 bit float back then, it would have been a complete non-issue!

[Doug Young]

Originally Posted by lkingston

It works right. It was only in the test files off the Sound Devices website where there was a problem

Ah, OK, I get it now. I just did a quick test, recording in 32 on the MixPre and importing to Logic. Worked fine, Logic converted to 24 automatically and levels are fine. Opened with Adobe Audition, which does support 32 bits, and that was also fine, preserving the 32 bits.

[KevWind]

Interesting discussion
I guess I will have give 32 bit float a try, since PT does support it And I am guessing I have plenty of computer to handle it .
I just set up some 32 bit session templates so my next session will be good to go ....

[sdelsolray]

Another subtopic concerning 32 v 24 v 16 bit involves processing the audio after recording. Many DAWs and plugins are capable of doing their internal processing at the higher bitrates and that higher bit processing renders higher quality files, at least according to many who know much more than I do. I've never noticed a difference between 24 bit and 32 bit FP processing myself, so I continue to use 24 bit.

[Fran Guidry]

I've used an F6 in 32 bit mode for a couple of years now. I've found three situations that make it useful to me.

The first is my own fallibility. I'm usually shooting "look at me play guitar" video so I'm camera op, audio tech, directer/producer, and talent. I've never shot a clip without making a mistake in at least one of those jobs. Eliminating the need to carefully set levels gets rid of one error path.

Second is the "sandbagging" issue. My buddy nearly always holds something back when we're setting levels and hits it a bit harder during the take. This is especially a problem if we're DI-ing a pickup where clipping sounds _really_ nasty. The F6 makes that problem disappear. It also saves time and energy that used to go into setting levels, making for less stress and possibly a better performance.

Third is event recording. I've shot a number of hula show videos and before the F6 I always had issues with levels. The audio in these events consists of miked narration, unamplified chant, amplified chant, recorded music, amplified live music, drum ensemble, and audience reaction. Before the F6 I just did the best I could and accepted the issues that arose. With the F6 I can get a decent recording of every aspect of the show.

If I were exclusively recording solo acoustic guitar in a controlled environment I might not have gone to the F6 and 32 bit, but now that I have I'm not going back.

Fran

[jim1960]

Originally Posted by KevWind

I guess I will have give 32 bit float a try, since PT does support it

I didn't even know we could record in 32 bit with our Carbons. It didn't get mentioned in any of the reviews I watched before buying and I never bothered to look at Avid's product page until I read your post. But there it is...

https://www.avid.com/products/pro-tools-carbon

[lkingston]

Originally Posted by Fran Guidry

I've used an F6 in 32 bit mode for a couple of years now. I've found three situations that make it useful to me.

The first is my own fallibility. I'm usually shooting "look at me play guitar" video so I'm camera op, audio tech, directer/producer, and talent. I've never shot a clip without making a mistake in at least one of those jobs. Eliminating the need to carefully set levels gets rid of one error path.

Second is the "sandbagging" issue. My buddy nearly always holds something back when we're setting levels and hits it a bit harder during the take. This is especially a problem if we're DI-ing a pickup where clipping sounds _really_ nasty. The F6 makes that problem disappear. It also saves time and energy that used to go into setting levels, making for less stress and possibly a better performance.

Third is event recording. I've shot a number of hula show videos and before the F6 I always had issues with levels. The audio in these events consists of miked narration, unamplified chant, amplified chant, recorded music, amplified live music, drum ensemble, and audience reaction. Before the F6 I just did the best I could and accepted the issues that arose. With the F6 I can get a decent recording of every aspect of the show.

If I were exclusively recording solo acoustic guitar in a controlled environment I might not have gone to the F6 and 32 bit, but now that I have I'm not going back.

Fran

Thanks for this post. These are EXACTLY the best reasons for using 32 bits!

[Doug Young]

Interesting. So I gave this a try, comparing 24 bits vs 32 both with normal gain, and with the gain all the way up for total overdrive.

The sensible-gain channels were basically identical, no advantage in terms of noise or anything.

However, loading the 32 bit overdriven track into Audition, I was able to totally recover the distorted file by just reducing the track volume. This did not work in Logic, the 32 bit track could not be salvaged. Checking the Sound Devices site, they call this out - Apple's products don't support this, but Audition, Audacity, Reaper, iZotope RX do. Interesting that they don't mention ProTools that I could find. I tried RX as well, and it worked fine. Certainly a benefit for cases where you can't set the levels right for whatever reason.