Taperssection.com
Gear / Technical Help => Recording Gear => Topic started by: dallman on September 30, 2019, 02:48:31 PM
-
This started as a recording thread but that has been moved to a new thread, and I leave this for those that wish to debate defend complain or whatever about their gear or the 32bit float process.
I thought with a few new decks having 32bit float recording and more coming, it might be nice to have a thread dedicated to recording with this process. There are plenty of threads debating the merits and features of the decks but many of us are interested in focusing on how these decks record in 32bit float and what the results are, so I thought this might be useful.
I used my MixPre6 II on Saturday to record Gov't Mule. I ran hotter than I ever had, and I did lower the levels about 2db after the first song which was in the red the entire time. My goal was to run hot, but not super in the red at all times hot. I ran 4 channels, Schoeps MK41's Hypercardioid, and Superlux S502 Cardioid. Here are my thoughts:
I started recording in 1971. It was really hard, really hard to watch levels run regularly in the red after 48 years of avoiding the red. I did it, but I figured how hot to run will be a gradual process.
The results though were excellent! There was nothing even close to distortion. I opened in Soundforge which is what I like to use nest. I transferred the poly file to a stereo file for the Schoeps and a stereo file for the Superlux. On both files I lowered the gain on the first song 2 db and then the entire file another 2 db. The files looked perfect and sounded great. I see obvious advantages , especially when the levels at the start of the show are much hotter than expected.
The learning curve might just be does anything matter? Is any gain too much? Do I care that I have to lower the gain in post if it sound great? Will it at any point with high hot gain sound compressed? These seem to be the details that we will all be looking at, and there probably will be a few differing opinions just like in all of our recording techniques. However there is much more freedom. After 48 years I do not make too many mistakes and I know my mics and most local rooms, but...this was really cool, I totally look forward to using the deck more. Tonight I will record 6 channels at a DSO show with different mics and while I do not plan to run superhot, will push my levels into the red with confidence.
-
Do you have any screen shots of the 'before & after' during Post Edit?
-
Will it at any point with high hot gain sound compressed?
in absence of using limiters it shouldnt
-
Do you have any screen shots of the 'before & after' during Post Edit?
check this out! this is from
https://www.sounddevices.com/noise-in-32-bit-float/?fbclid=IwAR1gJr5l4Bfo1GMNZXvLMqa_hWC2k1Sf9vk2tZoD7tdcz1K0GTsWcgYpRUY
I opened the one that says "all 3" in iZotope and saved the clipped one by itself. As you can see it's +31db and is full of distortion etc. Lowered the gain to just below clipping and it looks and more importantly sounds perfect! I'm sold.....
-
The only things you need to worry about with the MixPre-II in 32-bit mode are:
1) The microphone itself clipping!
2) Not exceeding the +12dBv (+14dBu or 11 volts p-p) maximum input on the MixPre's mic input. Good luck with that!
-
2) Not exceeding the +12dBv (+14dBu or 11 volts p-p) maximum input on the MixPre's mic input. Good luck with that!
^ This.
A lot of hot modern condensers will clip that, then 32 bit float does nothing for you.
-
i suppose for every recorder there is an optimum gain level that will offer maximum S/N
at some point adding extra gain just to remove it later can only add noise unnecessarily, right?
-
2) Not exceeding the +12dBv (+14dBu or 11 volts p-p) maximum input on the MixPre's mic input. Good luck with that!
^ This.
A lot of hot modern condensers will clip that, then 32 bit float does nothing for you.
explain like I'm an idiot please ;)
-
2) Not exceeding the +12dBv (+14dBu or 11 volts p-p) maximum input on the MixPre's mic input. Good luck with that!
^ This.
A lot of hot modern condensers will clip that, then 32 bit float does nothing for you.
DPA MMP-A says maximum output voltage is >5V rms/ >16V peak but youd prob have to be recording jet engines to get it to output that hot
-
2) Not exceeding the +12dBv (+14dBu or 11 volts p-p) maximum input on the MixPre's mic input. Good luck with that!
^ This.
A lot of hot modern condensers will clip that, then 32 bit float does nothing for you.
explain like I'm an idiot please ;)
If it's clipped....it's clipped.....you can turn it down.....but it's still clipped.....
A TLM103 will exceed that input level anywhere close to a drum kit. As will any Sennheiser MKH mic. A loud show with a lot of subs blowing out the bass will make a lot of mics output that kind of voltage if you are near the stacks. You might still need a pad on the mic or the preamp. Wind noise or vibration will definitely make levels like that, and if it clips it's harder to clean in post. You still have to set levels with an eye on not turning up too much, there's still a ceiling, it's just (confusingly) higher than it was, and past a monitor-able level.
The benefit to me seems to be more from the other end, not having to turn things up a lot in quiet ambient scenarios, especially when you might have unexpected hot transients (thunderstorms?), you can leave it low knowing you have a ton of headroom, without paying a digital penalty for that low level. Recording loud sources that are very even in volume like most people here do, there's not as much to be gained. The guy next to you who starts yelling and exceeds 0dBFS can be turned down if your were recording at -6 or -12 dBFS and he pegs the meter, he might not be distorted. He's a different sort of problem....
-
I guess I just don't understand the specs listed to watch out for. I looked up the tech specs of my Gefell's and don't see the same things. I've never been good at the tech aspect of all this!
-
I guess I just don't understand the specs listed to watch out for. I looked up the tech specs of my Gefell's and don't see the same things. I've never been good at the tech aspect of all this!
I wouldn't even bother looking at theoretical possibilities with mic specs, just know this 32 bit thing is not a free pass to ignore gain, it can't save everything. Most mics give you a max SPL for a 1% distortion, but that doesn't tell you what level they will output in a clear manner.
-
From what I have been able to gather, none of these machines actually use 32-bit floating point converters. They use several 24-bit integer converters at different gain levels and then the data is combined and converted to 32-bit float. So I guess it will depend on the implementation to some extent. Sound Devices patented their method, I think, so it will likely be different in other brand's recorders.
I guess I just don't understand the specs listed to watch out for. I looked up the tech specs of my Gefell's and don't see the same things. I've never been good at the tech aspect of all this!
Mostly mic sensitivity and SPL. See the table at Sengpiel (http://www.sengpielaudio.com/calculator-transferfactor.htm). It won't be easy to do with most mics, but still possible!
The guy next to you who starts yelling and exceeds 0dBFS can be turned down if your were recording at -6 or -12 dBFS and he pegs the meter, he might not be distorted. He's a different sort of problem....
That's what analog limiters are for! You'll have to do something to him in post regardless, right?
-
The guy next to you who starts yelling and exceeds 0dBFS can be turned down if your were recording at -6 or -12 dBFS and he pegs the meter, he might not be distorted. He's a different sort of problem....
That's what analog limiters are for! You'll have to do something to him in post regardless, right?
Yeah, or RX7 type cleanup. If it's splattered with a bunch of distortion harmonics, it's a lot harder to clean.
-
Here are a couple of interesting posts from DSatz on this topic from the F6 thread: http://taperssection.com/index.php?topic=190161.msg2304803#msg2304803 (http://taperssection.com/index.php?topic=190161.msg2304803#msg2304803) and http://taperssection.com/index.php?topic=190161.msg2309381#msg2309381 (http://taperssection.com/index.php?topic=190161.msg2309381#msg2309381).
-
Yeah, or RX7 type cleanup. If it's splattered with a bunch of distortion harmonics, it's a lot harder to clean.
This is why I have been using the limiter on my MixPre-6 first version. I set the levels pretty conservatively, so I tend not to hit them often, but occasionally there is that guy doing the "missile whistle" under the mics...
-
This thread, dedicated to 32bit Float recording in itself but not addressing the details of how it is implemented in specific recorders won't really be of much practical interest to tapers.. other than being useful to dispel some academic misunderstandings about what it can and can't do. In other words it will be mostly academic because what really matters is how its implemented in each specific recorder in question.
That's where the rubber meets the road and where all the current confusion lies!
-
The constraint is not 32bit floating point storage (which is ridiculously immense), but real-world bottle-necks prior to it.
One needn't adjust input trim/gain on a recorder if:
1) The total dynamic range of the signal will fit within the available dynamic range of the storage format..
(check, 32-bit floating point does this easily)
AND
2) The total dynamic range of the signal will fit within the available dynamic range through the recorder's signal path, up to the point where the signal is stored in that format..
(this reflects the most critical part of any specific implementation- the preamp and ADC performance of the specific recorder)
AND
3) The actual upper and lower dynamic range values of the signal fit comfortably within the upper and lower dynamic range limits of the recorder's signal path, up to the point where the signal is stored.
(this is about shifting level of the source if necessary to comfortably fit requirement 2, by making sure the signal does not overlap or exceed either end of the available range)
32-bit float storage solves the first problem. The available storage range of 32-bit float is effectively unlimited for audio purposes. Practically, it will depend on how well the other two things are handled in any particular recorder touting the benefit of 32-bit floating point storage.
The problem is accommodating the total output range of a source to the available input range through the recorder (which has been steadily increasing as tech advances, yet remains limited). Most of the hoopla from manufacturers and reviewers thus far has focused on the "wow factor" of the second part of that statement, while not talking much about the "real world" constraints of the first part. Those of us calling for temperance understand that the first part is really what is going to matter. And that will boil down to a performance measure of each individual recorder which is rarely specified clearly by manufacturers.
-
This thread, dedicated to 32bit Float recording in itself but not addressing the details of how it is implemented in specific recorders won't really be of much practical interest to tapers.. other than being useful to dispel some academic misunderstandings about what it can and can't do. In other words it will be mostly academic because what really matters is how its implemented in each specific recorder in question.
That's where the rubber meets the road and where all the current confusion lies!
True. I think that we will just have to figure it out ourselves. I have an F6 on the way, and will get to put it through its paces very soon. One difference in implementation right off the bat has to do with that max level. The MixPre II units have that incredible +14 dBu max mic input level, whereas on the F6 it's only +4 dBu. I don't record anything that would get my mics near that lower spec, but some others here might need to consider it if they don't have inline pads.
-
^ Yes! Exactly the problem I have currently with the F8.
-
2) Not exceeding the +12dBv (+14dBu or 11 volts p-p) maximum input on the MixPre's mic input. Good luck with that!
^ This.
A lot of hot modern condensers will clip that, then 32 bit float does nothing for you.
That's never been an issue for me. I'm recording with the same exact gear that worked at 24 bit, I can just run much hotter without fear. I have never had any of my mics distort and the only variable for me is recording at 32bit float instead of 24bit.
I'm recording right now a little hotter than I did at Mule. I'll try to post some pictures tomorrow.
-
I am pretty sure DSD recording allows for this same behavior as well just not as much of it - The Korg MR-1 could go 2 db over 0 before it clipped - you would just need to lower the levels after the recording to remove the clipping above 0db.
-
Considering the SoundDevices MixPre-II implementation in this light-
Requirement 1) The total dynamic range of the signal will fit within the available dynamic range of the storage format..
(check, 32-bit floating point does this easily)
Requirement 2) The total dynamic range of the signal will fit within the available dynamic range through the recorder's signal path, up to the point where the signal is stored in that format..
Paul reports a total dynamic range through the SoundDevices MixpreII's of 142dB (difference between Mic EIN @ -130dBV and maximum mic input level before clipping @ +12dBV). This is very impressive and reflects the clever design by SD using parallel preamps and ADC's stages of more limited range combined with a way of shifting between their outputs in real time. That will also shift the noise floor simultaneously, so technically the recorder cannot represent the full 142dB range at any particular moment, instead it automatically adjusts constantly to provide that overall range in a way which is presumably audibly transparent (and I expect it is). More on this in panatrope's post at GS. (https://www.gearslutz.com/board/showpost.php?p=14190912&postcount=526) The noise-floor shifting thing is probably not consequential for music recording - we don't hear or care about the noise floor of current recorders when signal is nearing 0dBFS - but it would be consequential if this were a data recorder rather than an audio recorder, which actually needed that full range at any moment. Anyway, this represents more than sufficient dynamic range for me.
Requirement 3) The actual upper and lower dynamic range values of the signal fit comfortably within the upper and lower dynamic range limits of the recorder's signal path, up to the point where the signal is stored.
This is gets to what paul and EmRR posted above:
The only things you need to worry about with the MixPre-II in 32-bit mode are:
1) The microphone itself clipping!
2) Not exceeding the +12dBv (+14dBu or 11 volts p-p) maximum input on the MixPre's mic input. Good luck with that!
^
A lot of hot modern condensers will clip that, then 32 bit float does nothing for you.
One will need to attenuate any input hotter than +12dBv (+14dBu) prior to the recorder. This is probably the biggest real world limitation of this particular implementation.
-
^ Yes! Exactly the problem I have currently with the F8.
Hmm, specs on that say mic input will do +14 dBu but with limiters on. I wonder what it is without limiters that they felt the need to pad that spec. (Ha!)
-
One will need to attenuate any input hotter than +12dBv prior to the recorder.
Do any recorders have metering that accurately shows this? You have to know at what point the meter is tapping the signal also. It could be showing +14, but that could be post-preamp, at the ADC. By then, the input has already been overloaded.
-
Hmm, specs on that say mic input will do +14 dBu but with limiters on. I wonder what it is without limiters that they felt the need to pad that spec. (Ha!)
F8 with Advanced Limiter switched on (implemented post ADC) attenuates input by 10dB prior to the ADC, so presumably +4dBu (that's not enough!)
F8N allows for phantom power on line-input, effectively increasing that to +24dBu with limiters on instead of +14dBu with limiters on (which should be sufficient)
-
One will need to attenuate any input hotter than +12dBv prior to the recorder.
Do any recorders have metering that accurately shows this? You have to know at what point the meter is tapping the signal also. It could be showing +14, but that could be post-preamp, at the ADC. By then, the input has already been overloaded.
Any of them that meter the input stage. Most probably meter the ADC though. Certainly the small hand-helds we use do.
Consider the "brickwalling" problem with many hand-held recorders. We post about the minimum input level setting which can be used with various recorders without clipping distortion occurring from the an overly hot input even though the meters are not indicating clipping. This indicates a design problem within the recorder for which we are finding a work arounds and sharing them with each other. If the recorder was designed correctly, brickwalling could not occur. The recorder's available range of input trim would be set up such that it could not be adjusted to do so. The minimum input-trim setting which could be used without brickwalling would equal "0" on the scale (equating to the maximum available input attenuation).
MixPre-II does appear correctly designed in that way:
This is achieved with multiple ADCs with more than one preamp in front of each one.
The MixPre-II Mic EIN is -130dBV and its maximum mic input level before clipping is +12dBV thus you have a dynamic range of 142dB. The 142dB dynamic range is mapped such that the maximum mic input signal (+12dBV) does not exceed 0dBFS in the multistage ADC circuit. The 32-bit output of the ADC is then converted to 32-bit float.. (https://www.gearslutz.com/board/showpost.php?p=14190646&postcount=523)
-
Here's something that's been nagging me from the start about all 32bit float audio recorders.
Taking the MixPre-II as example- Ignoring the input overload limits, what is important is the wide 142dB dynamic input range without requiring user adjustment. A 24 bit fixed file format can accommodate a 144dB signal which exceeds the throughput of this recorder..
32-bit floating point maybe helpful to accommodate the internal mixing aspects with appropriate processing headroom, but for anyone recording ISO tracks for later mix-down (most tapers), storing the larger 32-bit floating point files is a pointless waste of storage space.. and of bandwidth, which might be a real problem given the recently reported SDcard issues.
We should be able to harness the full power of that 142dB dynamic range when writing 24bit ISO files.
-
so what yore saying is with proper circuit design, the clipping point of the analog and digital stages are all well matched, so 0dB=0dB across the board and users can just set their levels anywhere above -40 dB and still have 100 dB of dynamic range and 40 more dB of headroom
-
A 24 bit fixed file format can accommodate a 144dB signal which exceeds the throughput of this recorder.
I think that 144 dB is a theoretical maximum. In reality, no ADC comes close. As Dan Lavry put it, "There is no such thing as true 24-bit conversion and there won't be in my lifetime." The first iterations of the MixPre, as well as the new Scorpio and 833, for example, specify a dynamic range of 120 dB for their 32-bit ADCs. I suppose that is why the 32-bit float machines need to use multiple ADCs and combine the data with, as jerryfreak memorably termed it, a little "digital fuckery"...
-
Hmm, specs on that say mic input will do +14 dBu but with limiters on. I wonder what it is without limiters that they felt the need to pad that spec. (Ha!)
F8 with Advanced Limiter switched on (implemented post ADC) attenuates input by 10dB prior to the ADC, so presumably +4dBu (that's not enough!)
F8N allows for phantom power on line-input, effectively increasing that to +24dBu with limiters on instead of +14dBu with limiters on (which should be sufficient)
I didn't notice it until you pointed it out, but the F6 does phantom with line-in also, specifying +24 dBu max level but no mention of limiter. I would have never thought to connect a mic directly to a line input. I wonder how the additional 2000 ohms of input impedance changes things on the low end of the signal spectrum. In other words, I question whether running straight to phantom-powered line in really gains you 20 dB dynamic range (probably not). You would only do that for constantly loud stuff.
-
This thread on JWSound gets a bit into the implementation of SD MixPre II vs. Zoom F6:
https://jwsoundgroup.net/index.php?/topic/34205-zoom-f6-vs-sound-devices-mixpre-ii-series-dynamic-range/ (https://jwsoundgroup.net/index.php?/topic/34205-zoom-f6-vs-sound-devices-mixpre-ii-series-dynamic-range/)
-
One question I have is would a 32 bit float recording sound better than a properly captured 24 bit recording with all other things being equal?
My sense is that the answer is no, but that doesn't take away from the benefits of having a larger margin of error in the field.
Dolby B and later Dolby C expanded the margin of error for tapers using compact cassette in the early 1980s. But skill and experience continued to be a requirement to make the best recordings. PCM and later DAT reduced the need to run hot to reduce noise since tape saturation was no longer a factor. 20 bit, and later 24 bit A/D chips that dithered to 16 bit improved the lower level signal quality by reducing the impact of quantization noise, slightly increasing the margin of error setting levels a little more. Still, precision in level setting was a key factor separating the best recordings from the rest, exposing the less skilled and experienced unless they got lucky.
Moving from 16 bit to 24 bit recording offered the improvement of allowing for higher sampling rates, but primarily freed the taper from having to "nail" a recording in 16 bit. The skill level and amount of experience needed plummeted with 24 bit.
From everything I've read, 32 bit float recording is just another step in this evolution, making it almost stupid easy to make a good recording in the field. The notion of "set it and forget it" is realized to the point that only operator errors like forgetting to hit the record button can get in the way.
With all of these advancements, isn't it ironic that there were far more active tapers back when it was harder and required more skill to get a worthy result than are active now?
-
Hmm, specs on that say mic input will do +14 dBu but with limiters on. I wonder what it is without limiters that they felt the need to pad that spec. (Ha!)
F8 with Advanced Limiter switched on (implemented post ADC) attenuates input by 10dB prior to the ADC, so presumably +4dBu (that's not enough!)
F8N allows for phantom power on line-input, effectively increasing that to +24dBu with limiters on instead of +14dBu with limiters on (which should be sufficient)
I didn't notice it until you pointed it out, but the F6 does phantom with line-in also, specifying +24 dBu max level but no mention of limiter. I would have never thought to connect a mic directly to a line input. I wonder how the additional 2000 ohms of input impedance changes things on the low end of the signal spectrum. In other words, I question whether running straight to phantom-powered line in really gains you 20 dB dynamic range (probably not). You would only do that for constantly loud stuff.
It does not gain you 20dB unless the limiter is on, and it's not entirely clear what's happening with the limiter. As I reported in the F8 thread, you can overdrive the F8n line input (set to lowest gain) into square waves with it metering -5.5dBFS. If you get a board feed, you may clip the input without an inline pad; I had a board feed drive it into clipping and ran the limiter just last week). I frequently run MKH mics at shows with it set to line rather than mic, lowest mic gain setting not low enough. I don't run the limiter, I run gain low enough to avoid it.
-
Well I have had the night to cool down and I am glad I did not post last night. I support everybody's right to an opinion and their ability to share it, but this thread was to be all about recording with 32bit float. It is new, it is different, the approach is different, and it would be great to have a place here to discuss as we use and learn this new feature on a few existing decks and more on the way. That is what I asked for in the thread, so I was a bit (or more) annoyed to see that after a nice start, the thread was totally hijacked with the same "stuff" that to me has been stated and restated. There are plenty of long rambling threads where people have already decided what they like or do not like about 32bit recordings or the company's that so far are selling them. I have no interest in that debate and I know I am not alone. I had a MixPre6 and loved it, but when the version II was announced, I sold that deck to a very good friend and decided to dive in to the new deck because when you record on the fly in different venues with different mics and different types of music to me what could be bad about not needing to worry about your gain settings? Again, everyone has a right to their opinions, but jeez, there are pages and pages of whining about cards and talking about why 32bit is no good. I just have no interest in thumbing through pages and pages of that to learn what those who are using 32bit are doing or learning or discovering. I just want a spot for that to be the focus as I think it is a pretty big deal. I have been doing audience recordings for 48 years in every conceivable form and deck in every conceivable format and I just only rely on my ears. I love the technical stuff, but audience recording has its own set of variables and rules. The only thing that never changes is that everything changes, and this concept, 32bit float on a portable deck is pretty amazing so far in my 2 outings. Whether I run super-hot or barely a bit hotter, I feel we are onto something here and those who do embrace it or those who want to learn more about how it is working in the field deserve one thread that talks about it. So, I hope that we can let the next thread not be the thread of anything more than what was originally stated.
I orginally locked this, but all of those that want to discuss what this thread became are welcome to. 8)
-
Thanks dallman,
Apologies for my part in taking this thread in a direction contrary to your intent. Your idea to open another dedicated to hands-on user reports and experience with 32-bit float recorders is a good one. I also think it wise to continue the overview and technical discussion concerning 32bit floating-point recording in this thread, where the technical aspects have been laid out pretty up to this point. The other threads where this has been discussed thus far are more broadly dedicated to specific recorders and other aspects about them. Hopefully keeping this overview and technical discussion about 32bit-float going here will keep it from inevitably spilling over into your new thread. I don't intend to let that happen on my part, but I can easily see it happening otherwise.
~regards
-
A 24 bit fixed file format can accommodate a 144dB signal which exceeds the throughput of this recorder.
I think that 144 dB is a theoretical maximum. In reality, no ADC comes close. As Dan Lavry put it, "There is no such thing as true 24-bit conversion and there won't be in my lifetime." The first iterations of the MixPre, as well as the new Scorpio and 833, for example, specify a dynamic range of 120 dB for their 32-bit ADCs. I suppose that is why the 32-bit float machines need to use multiple ADCs and combine the data with, as jerryfreak memorably termed it, a little "digital fuckery"...
I'm referring to the file storage format written to memory, not the analog to digital conversion. The multiple-ADC switching scheme is a clever work-around of the real-world practical limits imposed by traditional single-ADC conversion which Lavry was referring to, extending DR to 142dB.
My point is that 32-bit float is useful internally for summing and DSP (just as it is in a DAW), yet is unnecessary as a storage format because 24bit fixed provides sufficient storage capacity (just as it does in a DAW). The 142dB of actual throughput will fit within the 144dB of dynamic range storage capacity of a 24bit PCM file. Storing in 24bit fixed files loses nothing useful, reduces the throughput burden of writing data to the SDcard (a currently reported problem), is more universally usable and playable, and significantly reduces file size with high channel counts.
In the current MixPre-II implementation there is some quantization noise way down there due to SD not applying dither in the conversion from 32bit-float to 24bit-fixed (Discussion about that here (https://taperssection.com/index.php?topic=191640.msg2310715#msg2310715)). Paul mentions SD has up to this point not seen the need to do so, which indicates how insignificant the problem is (even though they are now highlighting it as example of superiority for 32-bit float recording - see here (https://www.sounddevices.com/low-signal-32-bit-float/) - erroneously in my opinion, for the reasons stated in the link above). If SD were to apply dither in the conversion from 32bit-float to 24bit-fixed there would be no quantization noise, yet the noise floor would increase by a few dB. So maybe we'd get ~140dB total range or something instead of 142dB, which is still more than enough. And if someone really needed to amplify so much that the noise-floor of the recorder became evident (and it isn't completely buried deeply beneath microphone self-noise and the environmental noise-floor, which is almost certain) they'd find a dither noise-floor rather than quantization artifacts.
For those reasons, I'd like to see Sound Devices implement dithering to 24-bit in combination with the new 32bit-floating point conversion, in addition to direct 32bit floating point storage.
[edit- SD page link fixed, thanks voltronic]
-
Gutbucket - your embedded link to SD's site is broken:
https://www.sounddevices.com/low-signal-32-bit-float/ (https://www.sounddevices.com/low-signal-32-bit-float/)
Here is another relevant JWSound thread:
https://jwsoundgroup.net/index.php?/topic/34174-floating-point-v-fixed-point-wav-files/ (https://jwsoundgroup.net/index.php?/topic/34174-floating-point-v-fixed-point-wav-files/)
-
One question I have is would a 32 bit float recording sound better than a properly captured 24 bit recording with all other things being equal?
not for concert recording purposes. same for 16 bit. if you are careful with levels and make a recording that peaks near zero your room noise is still 30-40 dB above the noise limit of the digital data
-
On the F8/F8N overload point and Advanced limiter-
It does not gain you 20dB unless the limiter is on, and it's not entirely clear what's happening with the limiter.
Not sure about the 20dB figure. I understand the Advance limiter to reduce sensitivity by 10dB, providing that much additional headroom before clipping should otherwise occur, but that 10dB includes the action of the limiter and presumably increases the noise-floor by the same 10dB. Threshold is not set directly nor clearly defined, with only the not to exceed level being set by the user. I've speculated that the Advanced limiter increases ratio as the signal rises higher above the threshold, reaching the stated infinity:1 ratio just below the do not exceed setting - basically a 10dB soft knee. It doesn't sound like inf:1 when lightly engaged. It has worked to keeep the F8 out of clipping for me, but I'd rather it be just a safety measure rather than the only way I can avoid overload when using high-sensitivity mics - and that's basically how I'm currently running it.
Zoom rep states- When the limiter of the F8/F8n is activated for a channel it automatically cuts 10 dB of gain from the analog preamp. That 10 dB is automatically added back after the AD converter. This is a way we build an extra 10 dB of headroom to avoid clipping over the already high dynamic range (A/D Converter has 120 dB dynamic range, with limiter effective dynamic range is 130 dB)
As I reported in the F8 thread, you can overdrive the F8n line input (set to lowest gain) into square waves with it metering -5.5dBFS. If you get a board feed, you may clip the input without an inline pad; I had a board feed drive it into clipping and ran the limiter just last week). I frequently run MKH mics at shows with it set to line rather than mic, lowest mic gain setting not low enough. I don't run the limiter, I run gain low enough to avoid it.
Hoping I can do the same as you to avoid clipping if I switch to F8N by using line-in and the input sensitivity turned all the way down. What is the sensitivity of the MKH mics so that I might compare to better confirm this? I'd forgotten the -5.5dBFS overload point behavior you found, which doesn't inspire confidence. Are you able to keep your recordings using MKH > line-in lowest gain > no limiter peaking below -5.5dBFS?
-
I have not had an F8n problem with MKH into line input, definitely too hot for mic input in many cases with a lot of what I do. 4060 not far behind it.
Senn. MKH 800 TWIN 40mV/Pa = -28dBV
Senn. MKH 20/30/40 25mV/Pa = -32dBV
Neumann TLM 103 23mV/Pa = –32.5 dBV - they quote a max output voltage +13dBu
DPA 4060 20mV/Pa = -34dBV
Neumann KM184 15 mV/Pa - they quote a max output voltage +10dBu
Neumann KM131 12mV/Pa = -38.4dBV
-
I have not had an F8n problem with MKH into line input, definitely too hot for mic input.
That's good to hear that the phantom+line is a useable option for really hot mics. I've never had a device with that capability, so have never tried it. I'll have to borrow some hot mics to try it on my F6.
It's worth noting that the Zoom recorders have lower input impedance than the SD MixPres. The F8n is 2 kΩ mic; 2.6 kΩ line. F6 is a bit higher at 3 kΩ mic; 5 kΩ line. MixPre II is 4 kΩ regardless of setting. That may not be the whole story when it comes to how hot a level it will deal with, but it's part of it.
-
From everything I've read, 32 bit float recording is just another step in this evolution, making it almost stupid easy to make a good recording in the field. The notion of "set it and forget it" is realized to the point that only operator errors like forgetting to hit the record button can get in the way.
Perhaps I'm at risk of picking on something you didn't quite mean the way you worded it, but there's a danger all the way through this discussion that we're saying that a recording which does not suffer from clipping at one end or noise at the other is a good recording.
In recent years, I doubt whether any appraisal of the real merits of a recording by any competent recordist will have included comments about poor level setting. A good recording in the field is still, and always will be, dictated first by the choice of microphones and their placement relative to the sound source. Next, these days, comes the choice of preamp, and last in importance is the choice of recording device, and the format used to record in (bits and sample rates etc). That's not to say it is of no consequence, and of course all items in the signal chain have to be matched at a technical level and at an overall quality level, but it's dangerous to risk any statement that devalues the fundamental skill that separates a competent field recordist from a beginner, that of mic selection and placement. We have to be careful that we don't make statements about recordings that are the audio equivalent of "that's a great photo, you must have a fantastic camera".
From what I'm reading above, choice and placement of mic is even more important with these devices. This now becomes the equivalent of level setting - which has been moved from a knob to the mic, and if you've got it wrong your amazing new recorder can't help you. As always, put a hot mic close to a loud source and you'll still risk clipping - or an unhot mic too far from a low level source, and you'll still risk noise (apart from all the other considerations affecting mic placement).
However, I do agree that we're seeing an important new phase in the development of audio recorders - we started with 14 bits (effectively, as I recall it), then 16, then 24, and now 32 bit float, and then... perhaps we're at the end of the road.
-
From everything I've read, 32 bit float recording is just another step in this evolution, making it almost stupid easy to make a good recording in the field. The notion of "set it and forget it" is realized to the point that only operator errors like forgetting to hit the record button can get in the way.
Perhaps I'm at risk of picking on something you didn't quite mean the way you worded it, but there's a danger all the way through this discussion that we're saying that a recording which does not suffer from clipping at one end or noise at the other is a good recording.
In recent years, I doubt whether any appraisal of the real merits of a recording by any competent recordist will have included comments about poor level setting. A good recording in the field is still, and always will be, dictated first by the choice of microphones and their placement relative to the sound source. Next, these days, comes the choice of preamp, and last in importance is the choice of recording device, and the format used to record in (bits and sample rates etc). That's not to say it is of no consequence, and of course all items in the signal chain have to be matched at a technical level and at an overall quality level, but it's dangerous to risk any statement that devalues the fundamental skill that separates a competent field recordist from a beginner, that of mic selection and placement. We have to be careful that we don't make statements about recordings that are the audio equivalent of "that's a great photo, you must have a fantastic camera".
From what I'm reading above, choice and placement of mic is even more important with these devices. This now becomes the equivalent of level setting - which has been moved from a knob to the mic, and if you've got it wrong your amazing new recorder can't help you. As always, put a hot mic close to a loud source and you'll still risk clipping - or an unhot mic too far from a low level source, and you'll still risk noise (apart from all the other considerations affecting mic placement).
However, I do agree that we're seeing an important new phase in the development of audio recorders - we started with 14 bits (effectively, as I recall it), then 16, then 24, and now 32 bit float, and then... perhaps we're at the end of the road.
Sorry, I thought when I said in my semi-rhetorical question at the top of my post "with all other things being equal" that removed those obvious factors from the recording device equation. Of course location, source quality, and mic placement come before the recorder in order of importance. I've actually made the point about skill and experience being more important than gear cost several times over the years on this forum. And while it's true that a naturally gifted or massively practiced golfer can beat a guy with the best clubs made using an old crappy set due to more advanced skill, I'm starting to pull back on that notion somewhat with photography and audio recording with the advancements in technology.
Take photography, when I started out using a 35mm SLR in 1980, it had a manual light meter called "match needle", and was significant because before they were around, light meters were separate hand held devices that photographers had to know how to interpret in order to set f-stop and shutter speed using one. You had to know how things like depth of field were directly related to aperture for portrait work, for example, when you couldn't preview it. And while advancements in camera technology can't improve poor composition and balance, something only skill and training can do, many of the aspects of cameras that also had to be mastered no longer require the same long learning curve. Just like with analog tape saturation, film exposure to maximize the amount of silver left on the negative was very important to understand in the past, but no longer applies in the digital era. It's like knowing how to navigate a boat with paper charts and dead reckoning turned into a quaint ability no longer needed once GPS technology became well established. I guess the point of my remarks was that 32 bit float removes one more skill set from the equation of requirements for a good recording, knowing how to set proper levels to maximize the performance of the recorder.
It means that skills around factors other than location and mic placement are being eliminated, just like portrait mode on modern phone cameras eliminates the need to understand why the subject is in focus and the background is out of focus when it's used, but they do need to know when to use the setting for composition, which absolutely reduces the learning curve just like automatic exposure meters in 35mm did a couple decades earlier. Hope that clarifies my comments from before.
-
A 24 bit fixed file format can accommodate a 144dB signal which exceeds the throughput of this recorder.
I think that 144 dB is a theoretical maximum. In reality, no ADC comes close. As Dan Lavry put it, "There is no such thing as true 24-bit conversion and there won't be in my lifetime." The first iterations of the MixPre, as well as the new Scorpio and 833, for example, specify a dynamic range of 120 dB for their 32-bit ADCs. I suppose that is why the 32-bit float machines need to use multiple ADCs and combine the data with, as jerryfreak memorably termed it, a little "digital fuckery"...
I'm referring to the file storage format written to memory, not the analog to digital conversion.
Gotcha. But further to the conversion itself, I think it is kind of curious that SD doesn't offer 32-bit float on their new pro-level recorders (the Scorpio and 833). Perhaps they were already too far down the development/testing pipeline to add that feature and there will be II versions of those in the near future as well. Or maybe it is not in demand by professional customers. Or some other reason. In any event, I would be interested in hearing why this highly-touted feature isn't on those recorders...
-
It's like knowing how to navigate a boat with paper charts and dead reckoning turned into a quaint ability no longer needed once GPS technology became well established.
You would have to be an idiot as a sailor (or pilot) not to learn old-fashioned navigation skills and have the correct charts and gear around in case your GPS fails. The sea (and sky) are unforgiving masters. Definitely some worse potential outcomes than a clipped recording...
-
one thing to consider, from a marketing perspective, is that its a relatively easy-to-implement feature that can be hyped
it costs a lot of money to improve analog front ends significantly
it likely costs a lot less to use multiple (relatively) cheap ADC chips in parallel
even the $200 sony A10 uses dual ADCs for higher dynamic range
realistically while 32bit float may save a recording or two, for most people who are familiar enough with the their gear to get levels within 20 dB of where they should be, it actually offers zero improvement to the end product
-
Gotcha. But further to the conversion itself, I think it is kind of curious that SD doesn't offer 32-bit float on their new pro-level recorders (the Scorpio and 833). Perhaps they were already too far down the development/testing pipeline to add that feature and there will be II versions of those in the near future as well. Or maybe it is not in demand by professional customers. Or some other reason. In any event, I would be interested in hearing why this highly-touted feature isn't on those recorders...
I thought I'd read somewhere that 32-bit float was going to be available on the 833 with a future firmware upgrade.
-
^ Interesting. Do you recall where you saw that? On SD's FAQ, it says, "The ability to record in 32-bit float is a future possibility." Whether or not that can be done in firmware was left unsaid (i.e. are all of the necessary ADCs already in there)...
-
^ Interesting. Do you recall where you saw that? On SD's FAQ, it says, "The ability to record in 32-bit float is a future possibility." Whether or not that can be done in firmware was left unsaid (i.e. are all of the necessary ADCs already in there)...
It's annoying me because I'm sure I've read it somewhere, but can't find it now. :shrug:
-
It's like knowing how to navigate a boat with paper charts and dead reckoning turned into a quaint ability no longer needed once GPS technology became well established.
You would have to be an idiot as a sailor (or pilot) not to learn old-fashioned navigation skills and have the correct charts and gear around in case your GPS fails. The sea (and sky) are unforgiving masters. Definitely some worse potential outcomes than a clipped recording...
I wasn't talking about celestial navigation, I was talking about Dead Reckoning which is all worked out on the chart itself, and is absolutely an obsolete skill. It was dropped from the USCG Master and Pilot commercial license requirements around 2014, about a year after NOAA issued the last printed navigational charts. I don't know about where you are, but if you relied on a paper chart last updated in 2014 sailing near the Eastern Shore or at the mouth of the York or Rappahanock Rivers, you would run hard aground in no time.
-
^ I didn't say "dead reckoning" specifically; I said "old-fashioned navigation skills and have the correct charts and gear". In aviation, it is required to have the right charts and a magnetic compass and both pilotage and dead reckoning are a part of the tested body of knowledge. I am less familiar with sailing, but I am pretty sure a magnetic compass and charts are required for anything more than a few miles from the coast almost everywhere. Also, it is kind of disingenuous to say that charts aren't printed anymore, as both aeronautical and nautical charts are constantly updated and can be obtained electronically and used either on an electronic device or printed and, at least in the case of FAA charts, can be obtained on paper from "approved print providers". This is pretty off-topic, though, so if you would like to further discuss the wisdom of relying solely on GPS without some old-school redundancy, feel free to PM me...
By the way, I say this as someone who has hundreds of hours of pilot in command time and has private, commercial, instrument, multi-engine and seaplane ratings.
-
^ Interesting. Do you recall where you saw that? On SD's FAQ, it says, "The ability to record in 32-bit float is a future possibility." Whether or not that can be done in firmware was left unsaid (i.e. are all of the necessary ADCs already in there)...
It's annoying me because I'm sure I've read it somewhere, but can't find it now. :shrug:
Looking at the spec sheets for the Scorpio and the 833, both state that they have 32-bit D/A converters, but only record in 16 and 24 bit depths. 32-bit A/D converters have been out and implemented for a bit now, but it's only very recently we're seeing portable recorders doing 32-bit floating point recording. So the question is: do these top-line SD units have multiple A/D converters in place already?
-
I believe any 32-bit A/D must use multiple overlapping converters to achieve that level of performance.
-
I believe any 32-bit A/D must use multiple overlapping converters to achieve that level of performance.
Exactly my point. Having a 32-bit A/D, but only one of them, means you'll never get 32-bit floating point recording. I mean, I guess you could, but it wouldn't be worth the trouble.
These 32-bit A/D chips have been out for a while now, but it's only these very new MixPre II units and the Zoom F6 that are using them with the multiple-overlapping structure.
There was a poster on GS who pointed out that the Stagetec has been around for a long time doing gain-ranging A/D. Their Truematch technology in their current converters have been out for a few years now, and they do 32 bits with a 158 dB dynamic range. I do not know if they ever integrated this with a recording media system though - just outboard converters, consoles, and routers.
-
The Scorpio and 833 have the hardware architecture to support 32-bit float in the future. This includes >1 ADC
As pointed out by someone earlier, use of more than 1 ADC to achieve wider dynamic range is not a new idea - its been around for decades.
We patented our method of multistage ADC because it is a new unique approach which we believe greatly improves handling the transition between ADCs compared to other implementations. If you want to know more about it you can read our patent online. Warning - its complicated and much of the math goes over my head so don't expect further insights from me.
-
The Scorpio and 833 have the hardware architecture to support 32-bit float in the future. This includes >1 ADC
As pointed out by someone earlier, use of more than 1 ADC to achieve wider dynamic range is not a new idea - its been around for decades.
We patented our method of multistage ADC because it is a new unique approach which we believe greatly improves handling the transition between ADCs compared to other implementations. If you want to know more about it you can read our patent online. Warning - its complicated and much of the math goes over my head so don't expect further insights from me.
Thanks for that info, Paul. I'm sure people who bought Scorpios just before the MixPre II came out are reassured to know this.
I did try to read the patent a couple weeks ago, but it's way above my comprehension level.
-
Genuine question - in case anyone can be bother to answer what may be a silly one - what is the difference between using one of these new recorders, compared with recording in 32 bit float on a laptop (with a decent rig in front of it)? Is the breakthrough that it's all in one box? Was I actually doing this years ago in Audition?
-
The routing bus and capture are 32 but the converters are 24. My Motu 16A shows up as 32 but that’s just the bus depth.
-
Genuine question - in case anyone can be bother to answer what may be a silly one - what is the difference between using one of these new recorders, compared with recording in 32 bit float on a laptop (with a decent rig in front of it)? Is the breakthrough that it's all in one box? Was I actually doing this years ago in Audition?
The constraint is not 32bit floating point storage, but real-world bottle-necks prior to it.
One needn't adjust input trim/gain on a recorder if:
1) The total dynamic range of the signal will fit within the available dynamic range of the storage format..
AND
2) The total dynamic range of the signal will fit within the available dynamic range through the recorder's signal path, up to the point where the signal is stored in that format..
AND
3) The actual upper and lower dynamic range values of the signal fit comfortably within the upper and lower dynamic range limits of the recorder's signal path, up to the point where the signal is stored.
The laptop is handling part 1
The rest of the rig is responsible for parts 2 & 3
-
Any info how 32Bit Float will work with compressors and software limiters in real life?
If your source have a lot of really loud splashes and you bring back loudest part to visible range in one click, other parts of track gain became too low. Simple normalize tool for selected regions only or manual gain adjust with curve is ok for simple situations, but for instant random loud splaches you need some kind of automatic limiter. So will those software plugins see that hidden dynamic range in 32Bit mode? Is it possible that software limiters in 32Bit mode will produce same result as analogue limiters or even better?
UPDATE: I downloaded samples from SoundDevices website and play with very simple AU limiter plugin in Cocos Reaper. All you need do is set attack/release time. Seems works great. Artifacts-free real time recovery! :headphones:
-
When a recording on the unit is going well into the red, what happens to the monitoring? Do you hear distortion that won't be heard after normalising, or... how does it work? I guess it's down to where in the chain the monitor signal comes from.
-
When a recording on the unit is going well into the red, what happens to the monitoring? Do you hear distortion that won't be heard after normalising, or... how does it work? I guess it's down to where in the chain the monitor signal comes from.
My guess you will hear distortion that won't be heard after normalising. MixPre limiters are disabled in 32 bit mode.
Here are some processed examples (original, normalized only, non normalized with software limiter plugin):
https://www.dropbox.com/sh/rm6c9sdgimeoy98/AACXdJ3tpd5Mi5QJYyO7gKG7a?dl=0
It is also interesting what strategy will work better for sound "highlights" recovery?
a. normalize all track and then use compressor to boost silent parts withiout clipping loudest parts.
b. don't normalize track and recover hidden loudest parts with limiter.
-
DSatz, Gutbucket, and a few others raised some excellent points about tempering expectations regarding 32-bit float recording. This is valuable information we all need to pay attention to, but as has been said before: the real test is in the implementation.
To that end, I would like to hear from other 32-bit float recorder owners if they have run into anything of concern.
So far, recordings from my Zoom F6 have been flawless. These recordings have swung very wide dynamic ranges, and I do not hear any noisefloor modulation, artifacts, or anything else that shouldn't be there. If there is anything nasty going on, it is buried way down in a level that I cannot hear it, nor can my spectral analysis show it.
-
DSatz, Gutbucket, and a few others raised some excellent points about tempering expectations regarding 32-bit float recording. This is valuable information we all need to pay attention to, but as has been said before: the real test is in the implementation.
To that end, I would like to hear from other 32-bit float recorder owners if they have run into anything of concern.
So far, recordings from my Zoom F6 have been flawless. The recordings I have made have swung very wide dynamic ranges, and I do not hear any noisefloor modulation, artifacts, or anything else that shouldn't be there. If there is anything nasty going on, it is buried way down in a level that I cannot hear it, nor can my spectral analysis cannot show it.
I agree, my recordings have sounded excellent and very much like the 24bit recordings I have been making for years. I have played with running hot and I have played with recording conservatively and it does not seem to make a difference. I have seen very positive tests where things have really been pushed but I am not looking to make a point when I record, I am looking to grab as good a recording as I can, so I am not pushing into anything crazy. I have used both the F6 and the MixPre6II and I am happy with both decks performance and sound.
-
Excellent feedback, thanks!
Good to hear if I had kept my order with Zoom that all would end well. But I switched on the delay because I couldn’t be sure it would be ready for my late September and October recordings where I used the MixPre-6 II successfully.
-
got into a discussion on another board where it was asserted that 32-bit float offers similar 'precision' to 24-bit. their discussion was in regard to DSP.
heres the bulk of my post. Whycome i got what look like rounding errors if the container can represent 1500 dB of data to the sample?
----------------------------------------------------------------------------------------------------------------------------------------
im not really mathematically equipped to discuss it at that level
though it seems they are talking about precision of calculation of 32-bit float in DSP
i could kind of see this in a way. for a given mantissa value, there is a finite 'precision' those numbers can represent. and then the exponent in a way 'slides' that scale of representable numbers up and down, but it doesnt go any 'deeper' or more precise in calculations
seems like almost two different discussions. (DSP vs representing large differences in headroom)
im not super concerned about how the LSB in a DSP is handled, as much as i am about not clipping my signal by running out of data 'container capacity' for a better recording.
i did a test in soundforge. i generated a 1 sec 1khz sine at FSD and saved that file
then i used 'volume' effect in soundforge to adjust the volume by -120.00dB. saved that file
then i used volume again to adjust the volume +220.00 dB, which took it to an expected +100 dB signal on playback , saved that file
then i used volume to reduce it exactly -100.00 dB which put it at an expected 0dB peak
when i inverted this over the original 0dB generated file, it canceled it.... almost. there is some resulting random noise with peaks at -135 dB
so it can be argued that i indeed represented this waveform at ~24 bit precision at volume levels ranging over 200+ dB difference. however at the end of the day, the DSP precision of that file after multiple computations was indeed right around 24-bit.
could be multiple accumulated rounding errors at the 32-bit noise floor summing up to make some random low-level noise?
______________________________________________________________________________________________________
responses from that forum:
Right, it's about what you could call instantaneous dynamic range. 24 bit signed integers effectively have 23 bits plus sign but any integer gain adjustment could either clip or throw away bits. Each operation in 32 bit floating point essentially normalizes the result to keep 23 bits plus sign in the result. There numerical noise issues when doing a large number of re-scales or effects but these are usually minor.
and another response, emphasis mine
This interactive demo should be self-explanatory:
https://www.h-schmidt.net/FloatConverter/IEEE754.html
An implied bit when exponent bits are non-zero (1), (23) bits of mantissa and (1) sign bit.
The benefit of floating point is it has a near constant precision at different scales, that makes digital processing and gain staging much safer and consistent.
-
jerryfreak, nice experiment. Thank you for it.
I don't agree with the last part of the final statement (in boldface); it even seems propagandistic to me--an attempt to make the shifting digital noise floor seem like a virtue when it isn't one. I don't mean that it's necessarily a defect, either, because if it's low enough at all times, no one will hear it shifting. But "it successfully evades detection" is the best that can be said about it if so.
--best regards
-
DSatz, Gutbucket, and a few others raised some excellent points about tempering expectations regarding 32-bit float recording. This is valuable information we all need to pay attention to, but as has been said before: the real test is in the implementation.
To that end, I would like to hear from other 32-bit float recorder owners if they have run into anything of concern.
So far, recordings from my Zoom F6 have been flawless. These recordings have swung very wide dynamic ranges, and I do not hear any noisefloor modulation, artifacts, or anything else that shouldn't be there. If there is anything nasty going on, it is buried way down in a level that I cannot hear it, nor can my spectral analysis show it.
Bumping this thread for anyone considering a 32-bit float dual-ADC recorder...
6 months in, and I can report zero glitches to date with my Zoom F6 running in this mode. While I can't speak for the specifics of the exact implementation on other units, the input gain on the F6 is fixed in this mode. I tend to turn up the faders a bit just so I can see levels popping up during setup, +20 to +40 depending on which mics I'm using. Keep in mind that this is only affecting post-ADC fader level as the gain is fixed. This means that the resulting file may have areas significantly beyond 0 dBFS, but in this format you can just pull it down and post; there really isn't any clipping.
For me, I am never going back to 24-point fixed unless another 32-bit float dual-ADC device I am using down the road doesn't work as well as this one does. I have posted on some other threads that I am often both performer and recordist for my concerts, so I cannot monitor my recordings which have wide dynamic ranges, and peak levels can be unpredictable. Because of this, this new implementation truly is a game-changer for me, because I no longer have to rely on safety tracks and can use all of my inputs without have to think about levels.
Yes, one still needs to be careful not to overload the input stage, as has been discussed. I don't own very sensitive mics so it's a non-issue for me. If it is a concern for you, the Zoom F6 has a Line + Phantom mode which has a much higher max input level, or you can go with the Sound Devices MixPre-II series which has a much higher max level in all recording formats.
I am eagerly awaiting to see how the Tentacle Track E recorders perform once released.
-
ive come across some good recent posts from other forums with respect to bit-depth and fixed vs float, that i thought i would share here
It appears the most state of the art gear uses 32-bit fixed, not float
here is an example of a modern dual-ranging ADC, the Merging ADA8S, with some astonishing specs, using its 'dual gain' topology. it literally is utilizing close to 24bits of resolution through its entire signal path which would put it quite a bit above the specs of the 32-bit float recorders we are discussing here
https://youtu.be/IMkUcWr2-cg?t=629
heres a direct demo of a sine wave recorded peaking at -110dBU and boosted 80dB in post. the new ADA8S is the one on the left and is pretty astonishing:
https://www.merging.com/products/interfaces
some comments ive culled on fixed vs float re:this gear and its implementation
audiosciencereview:
Basically the floating point approach is just an array of ADCs (or a multichannel ADC) recording at different levels. The old Sony PCM-D100 used a similar approach -- recording two copies at the same time, but with a difference of 12dB. Of course, it only supports 24-bit integer and DSD.
While 32-bit float "only" has 25 bits of precision, don't forget it wins 32-bit integer when sample values (yes, sample values, not the overall level of the whole file) are below -42.1442dBFS (7 bits below full scale) and above 0dBFS (integers will clip). Floating point on the other hand always store 25 bits of data within the ~1500dB range. Since no combination of ADC, mic and preamp can achieve 25 bits of instantaneous dynamic range, the floating point approach is much more sensible.
from Gearslutz, someone asked Merging tech support the following question
Will the new ADA8 also be 32-Bit like Anubis? And speaking of 32-Bit conversion, Anubis specs doesn't say 32-Bit float, so it's fixed correct? And if that's the case, what made you choose 32-Bit fixed instead of float, I'm not saying one is better than the other, I'm just curious as to why you made the decision to use 32-Bit Fixed, and not Float?
response:
In terms of raw resolution, 32 bit fixed point is much better than 32 bit floating point (which only gives the equivalent of 25 bit resolution).
In other words, if converting a 32 bit fixed point signal to a 32 bit float signal and then back again to 32 bit fixed point, one would lose the 7 least significant bits from the original 32 fixed point number.
In terms of Dynamic range however (headroom, noise floor), 32 bit float is of course incomparably better (about 1500 dB) than fixed point (about 186 dB) but in most real situations 1500 dB is uselessly large.
And with respect to using a full 32 bit fixed point path in the Anubis it does make most sense, since the AD chip is able to deliver up to 32 bit fixed data and the DA chip is able to eat up to 32 bit fixed data as well.
So making sure to preserve all those 32 bit during the whole signal processing chain including through the mixing engine offers maximum transparency in the Anubis (and soon in the new ADA8 as well).
at the end of the day, we are stuffing 12-16bits of information from our microphones into a box. at the end of the day it doesnt practically matter if the box you put it in has 24 bits, 25 bits, or 32 bits of space. All that matters is that you get it in the box ;D
as mentioned above when your sample data is below -42dB, 32bit-float offers theoretical advantage. Youd be hard pressed to hear any quantization noise over the noise floor of the room, the mics, and the analog input of your recorder, however.
-------------
here is an example of a modern ADC chip, the 8-channel AK5578
you can see how multi-channels of a chip are typically combined to boost S/N
S/N: 121 dB (8-to-4 mode: 124 dB, 8-to-2 mode: 127 dB, 8-to-1 mode: 130 dB)
https://www.akm.com/content/dam/documents/products/audio/audio-adc/ak5578en/ak5578en-en-datasheet.pdf
-
ive come across some good recent posts from other forums with respect to bit-depth and fixed vs float, that i thought i would share here
It appears the most state of the art gear uses 32-bit fixed, not float
here is an example of a modern dual-ranging ADC, the Merging ADA8S, with some astonishing specs, using its 'dual gain' topology. it literally is utilizing close to 24bits of resolution through its entire signal path which would put it quite a bit above the specs of the 32-bit float recorders we are discussing here
https://youtu.be/IMkUcWr2-cg?t=629
heres a direct demo of a sine wave recorded peaking at -110dBU and boosted 80dB in post. the new ADA8S is the one on the left and is pretty astonishing:
https://www.merging.com/products/interfaces
some comments ive culled on fixed vs float re:this gear and its implementation
audiosciencereview:
Basically the floating point approach is just an array of ADCs (or a multichannel ADC) recording at different levels. The old Sony PCM-D100 used a similar approach -- recording two copies at the same time, but with a difference of 12dB. Of course, it only supports 24-bit integer and DSD.
While 32-bit float "only" has 25 bits of precision, don't forget it wins 32-bit integer when sample values (yes, sample values, not the overall level of the whole file) are below -42.1442dBFS (7 bits below full scale) and above 0dBFS (integers will clip). Floating point on the other hand always store 25 bits of data within the ~1500dB range. Since no combination of ADC, mic and preamp can achieve 25 bits of instantaneous dynamic range, the floating point approach is much more sensible.
from Gearslutz, someone asked Merging tech support the following question
Will the new ADA8 also be 32-Bit like Anubis? And speaking of 32-Bit conversion, Anubis specs doesn't say 32-Bit float, so it's fixed correct? And if that's the case, what made you choose 32-Bit fixed instead of float, I'm not saying one is better than the other, I'm just curious as to why you made the decision to use 32-Bit Fixed, and not Float?
response:
In terms of raw resolution, 32 bit fixed point is much better than 32 bit floating point (which only gives the equivalent of 25 bit resolution).
In other words, if converting a 32 bit fixed point signal to a 32 bit float signal and then back again to 32 bit fixed point, one would lose the 7 least significant bits from the original 32 fixed point number.
In terms of Dynamic range however (headroom, noise floor), 32 bit float is of course incomparably better (about 1500 dB) than fixed point (about 186 dB) but in most real situations 1500 dB is uselessly large.
And with respect to using a full 32 bit fixed point path in the Anubis it does make most sense, since the AD chip is able to deliver up to 32 bit fixed data and the DA chip is able to eat up to 32 bit fixed data as well.
So making sure to preserve all those 32 bit during the whole signal processing chain including through the mixing engine offers maximum transparency in the Anubis (and soon in the new ADA8 as well).
at the end of the day, we are stuffing 12-16bits of information from our microphones into a box. at the end of the day it doesnt practically matter if the box you put it in has 24 bits, 25 bits, or 32 bits of space. All that matters is that you get it in the box ;D
as mentioned above when your sample data is below -42dB, 32bit-float offers theoretical advantage. Youd be hard pressed to hear any quantization noise over the noise floor of the room, the mics, and the analog input of your recorder, however.
-------------
here is an example of a modern ADC chip, the 8-channel AK5578
you can see how multi-channels of a chip are typically combined to boost S/N
S/N: 121 dB (8-to-4 mode: 124 dB, 8-to-2 mode: 127 dB, 8-to-1 mode: 130 dB)
https://www.akm.com/content/dam/documents/products/audio/audio-adc/ak5578en/ak5578en-en-datasheet.pdf
The way I interpret the two bolded statements is that if I'm exporting from my DAW in 24bit, it seems that 32bit float is the best of everything and there might not be a need to ever upgrade a recorder after this.
-
well, it depends on what youre happy with. technically im sure somebody said the same thing when good HQ pre/AD combos like the V3 came out. As easily as digital technology can become dated, 20 years later you can still go make world-class recordings with a V3, in a studio, yet alone a relatively noisy live show environment.
Heck a sax>SBM setup from the 90s was putting down phenomenal dynamic range for our purposes esp with the noise-shaping of the SBM
that said, perhaps you could have highlighted this part as well:
Since no combination of ADC, mic and preamp can achieve 25 bits of instantaneous dynamic range
it seems that the ADC portion is ahead of the rest of the signal chain, for now (forever?). However, unlike previous improvements in digital technology, the 32-bit float architecture only improves the sound when compared with a less-than-ideal setup of lesser bitrate. In other words there is no sonic advantage to a 32-bit file vs a 24-bit file at proper levels. Which depending on setup (background noise, mic noise, amount of gain used, etc) could mean anything where your peaks dont hit -12dBu to where your peaks dont hit -48dBu. Again, its a container for your data, if you are carefully putting your data in the container there will be no difference. if you are missing the container with your data its a huge problem. and an easy fix in this case, with the auto-ranging ADCs which "move the container for you" as needed.
will you want to upgrade someday? sure... when a recorder that sounds as good as your mixpre is the size of your phone. that will be another 20+ years though, and will likely only offer diminishing returns, if any, over your current setup in terms of overall sonic quality for our purposes
next stop is microphone signal path, it seems that paused for a bit with the digital technology, but i would imagine thats what we see next. Every manufacturer has their own way of powering their mics, from the permanently polarized DPAs that take a trivial amount of voltage/current, to the schoeps who have cut the power requirements of their preamps by nearly an order of magnitude in a few decades
couple these advances with digital microphone preamp technology (advantage being less power required to generate a hot signal to compete with noise), as well as the ever evolving ADC and op-amps, and, basically its only going to get better as far as tech.
so i guess to circle back, you asked "there might not be a need to ever upgrade a recorder after this."
was there a need before this? :bigsmile:
-
^ Good summary!
auto-ranging ADCs "move the container for you" as needed
-
Regarding the mic signal path:
Remember AES42 digital mics, like the Neumann D line? For those that don't know, these mics have a 28-bit ADC right in the mic body. The specs on them are amazing, and the few recordings I have heard made with them are extremely transparent sounding. It doesn't seem that standard ever got wide adoption though - I hear Neumann is discontinuing that series. Sound Devices always supported AES42 back through the 7-series.
The silver bullet for audio recording in my mind would be AES42 mics with the type of 32-bit multi-ADC converters we are discussing built right in. Now you have the mic manufacturers doing all of the "container moving" onboard, and you can simplify your recording box to a bit bucket that only has to supply 10V phantom to the mics. Lower noise all around, less chance of interference on your cable runs, lower power requirements, etc.
-
yes schoeps digital version of CMCs (CMD i beleive?) just disappeared as well about 7 years ago
i’m sure they could do it better with today’s tech. those came out in like 07-08 IIRC. that’s forever ago in digital
-
Thoughts on this? I have not heard this in my recordings.
https://youtu.be/y3431uljZ2k?si=I7I7qhsKK7TUo49s
-
Thoughts on this? I have not heard this in my recordings.
https://youtu.be/y3431uljZ2k?si=I7I7qhsKK7TUo49s
Wow. There's more information in the comments, including how Sound Devices (mostly?) fixed this with a firmware update.
Now I'm going to have to dig into some of my F6 recordings and see if I can locate this issue.
What I don't understand is why the low-gain ADC is 30 dB noisier than the high-gain one. I would expect it to be the reverse.
-
It's an unpleasant issue, indeed. Thing is, I am pretty sure that Zoom F serie recorders still have dual ADC working even in 24 bit mode, so no escape unless the manufacturer finds a solution via firmware ( if any solution is possible)
-
Thoughts on this? I have not heard this in my recordings.
https://youtu.be/y3431uljZ2k?si=I7I7qhsKK7TUo49s
Wow. There's more information in the comments, including how Sound Devices (mostly?) fixed this with a firmware update.
Now I'm going to have to dig into some of my F6 recordings and see if I can locate this issue.
What I don't understand is why the low-gain ADC is 30 dB noisier than the high-gain one. I would expect it to be the reverse.
From what I understood, the recorder engages the low-gain ADC when it detects a transient - and it does so before the signal gets strong enough that it has the chance to clip. This means that there's some amount of guessing involved, and it might engage the low-gain ADC for sudden sounds that end up not getting anywhere near clipping level. And it's precisely because it has engaged the low-gain ADC for a sound that was actually fairly quiet that the device needs to make up for it by amplifying the signal digitally, raising the noise floor of the recording. It isn't that the low-gain ADC is noisier than the high-gain one in absolute terms (you're right that this would make no sense): it's that the high-gain one has cleaner analogue gain than simply boosting the digital signal produced by the low-gain one by however much is necessary so that the levels remain consistent throughout the recording.
You might have noticed he was recording tiny sounds (they sounded like water dripping?), not a gunshot or a drum kit. I imagine that these artefacts wouldn't appear in a live music recording in any noticeable way.
Edit: actually, I paid a bit more attention this time, and I don't think it's supposed to detect transients? It just routes the signal based on the voltage the microphone produces. So the situations when the issue would happen are a little different, but the same principle still applies. Basically, say the high-gain ADC is a +30dB one. Any signal strong enough to clip with that amount of gain is automatically routed to the low-gain ADC. If it's a healthy signal, you won't hear any noise. But if it's JUST loud enough to cross the clipping threshold of the high-gain ADC (say, by 1dB), then clearly the low-gain ADC is going to need to be boosted digitally so that the levels are consistent with rest of the recording. You're not going to see an issue when the sound falls into the best operational range of each ADC. You get some noise when a sound is just loud enough that you can't use cleaner gain, but still quiet enough that the new noise floor is noticeable.
-
^ That explanation makes sense. I was thinking of the signal being amplified in the analog domain before it's converted, but that's not how these devices work. I should have realized that because I frequently correct other people on this assumption. Thanks for setting me straight.
-
It's an unpleasant issue, indeed. Thing is, I am pretty sure that Zoom F serie recorders still have dual ADC working even in 24 bit mode, so no escape unless the manufacturer finds a solution via firmware ( if any solution is possible)
Yes, and that was discussed in the comments. One poster mentioned that it's the same deal for the Sony D100.
-
I noticed my Mike Stern F3 recording from last week was really noisy. Almost analog hiss sounding.
-
I noticed my Mike Stern F3 recording from last week was really noisy. Almost analog hiss sounding.
Which mics were you using? How loud was the show?
-
I noticed my Mike Stern F3 recording from last week was really noisy. Almost analog hiss sounding.
Which mics were you using? How loud was the show?
Its this one. I figured it was because of the low pro situation. Mics have never made any similar hiss.
https://taperssection.com/index.php?topic=204608.0
-
It just shouldn't be anywhere near where it would be using the low gain adc if the show wasn't very loud. And that band doesn't strike me as one that would be inordinately loud.
-
It just shouldn't be anywhere near where it would be using the low gain adc if the show wasn't very loud. And that band doesn't strike me as one that would be inordinately loud.
Unfortunately the only reason I got it was to record >:D these particular bands. Think Im going back to mk2e's>BB>A10
-
is the hiss always there? if you can isolate it, RX can remove it automatically
-
Just the quiet parts. It's not terrible, but I might try and fix it up.
-
That is strange. I love my F3s because the preamps are so darn quiet.
-
is the hiss always there? if you can isolate it, RX can remove it automatically
Watch the video. This is addressed. It's only around the louder transients.
-
Just the quiet parts. It's not terrible, but I might try and fix it up.
I've just listened to your recording! It isn't the same issue described in the video. I hear the noise, but it's everywhere during the quiet parts, not just around the transients. It's either the microphone's or the preamp's noise floor, and I don't imagine you'd get a lot less noise with a different recorder unless your F3 is faulty. There could be other weird things going on too - I know for example that I get a lot more noise out of my Church Audio preamp if I put it down on a church pew or my desk. If I keep it on my body, it's fairly quiet (I imagine this has to do with the grounding).
-
This idea is making waves on various recording forums/blogs I read:
https://www.youtube.com/watch?v=y3431uljZ2k
tl;dr, on a 32-bit recorder, the multiple ADC's kick in depending on how loud the source is. The problem is that the different ADC's have vastly different noise floors, so hit a loud transient and your recorder will choose the low-gain ADC, which adds a lot of noise right around the transient.
Now, because the noise is different/worse only around the loud transient, it's also extra hard to profile and remove with software (versus a consistent noise throughout the whole recording).
May not apply to you all. For me, I record sopranos who go into glass-breaking-mode sometimes after a gently piano intro, so this is indeed a problem.
-
so can you post an example of where it happened on one of your recordings?
also: https://taperssection.com/index.php?topic=192100.0;prev_next=next#new
-
Having spent a happy hour reading about the subject (in documents dated prior to this video), it seems to be... complicated. One writer suggested that indeed the noise involved in 32 bit float recordings is variable, not fixed, but it's variable in the area of -144dB so it is unlikely to be audible. I'm too old to hear the noise apparently demonstrated in the video, but I did rather wonder whether the source being used was not a good choice. It sounded like some kind of sound effects file with a lot of constant low frequency background noise, plus artificial reverb (but I could be wrong) - is it the reverb noise that's happening? A much better test would be to simply use a tone generator, surely?
Anyway, I would be much more concerned about distortion (clipping) than noise for real-world examples, and the whole concept of having no level controls and no meters appeals to me immensely. The ability to walk into a classical chamber recital hall and position a 32 bit float recorder/mic at the sweet spot, turn it on and leave it to its own devices uncablednd unmonitored, and go to the bar for the performance, appeals to me immensely. Would the client complain about the subtle variable noise floor? I doubt it very much. Audience coughing and phones ringing are likely to be a greater source of irritation.
But I do concede that ultimate perfection should be sought by ultimate perfectionists
-
So has anyone heard this transient noise in practice in the field?
-
The voltage ranges for the ADCs overlap, at least in the Sound Devices implementation. That means that the sound right around the "switch" is likely to be in more than one post-ADC stream. SD's algorithm, based on regression models, enables rapid shifts from one ADC to another and since there is overlapping content, this can also be done very accurately. I would guess this can be fine-tuned to eliminate or, at least, minimize this problem. Maybe different in Zoom's implementation. I don't know and this is just speculation anyway...
-
Having spent a happy hour reading about the subject (in documents dated prior to this video), it seems to be... complicated. One writer suggested that indeed the noise involved in 32 bit float recordings is variable, not fixed, but it's variable in the area of -144dB so it is unlikely to be audible. I'm too old to hear the noise apparently demonstrated in the video, but I did rather wonder whether the source being used was not a good choice. It sounded like some kind of sound effects file with a lot of constant low frequency background noise, plus artificial reverb (but I could be wrong) - is it the reverb noise that's happening? A much better test would be to simply use a tone generator, surely?
Anyway, I would be much more concerned about distortion (clipping) than noise for real-world examples, and the whole concept of having no level controls and no meters appeals to me immensely. The ability to walk into a classical chamber recital hall and position a 32 bit float recorder/mic at the sweet spot, turn it on and leave it to its own devices uncablednd unmonitored, and go to the bar for the performance, appeals to me immensely. Would the client complain about the subtle variable noise floor? I doubt it very much. Audience coughing and phones ringing are likely to be a greater source of irritation.
But I do concede that ultimate perfection should be sought by ultimate perfectionists
Imho the question is: do you want audible noise around some transients (not all of them, just the ones that are too loud for the high-gain ADC, but a bit too quiet for the low-gain one), or do you want noise everywhere? Without multiple ADCs, if you're expecting the music to have louder passages, the one option you have is to be conservative and use low gain for everything, which means the quiet parts will have a higher relative noise floor. Sure, it's easier to profile the noise in post that way, but I'd much rather have less noise during the quiet parts in the first place. I could hear the noise around the transients, but I personally don't find it distracting at all.
-
Foreshadowing back when..
This thread, dedicated to 32bit Float recording in itself but not addressing the details of how it is implemented in specific recorders won't really be of much practical interest to tapers.. other than being useful to dispel some academic misunderstandings about what it can and can't do. In other words it will be mostly academic because what really matters is how its implemented in each specific recorder in question.
That's where the rubber meets the road and where all the current confusion lies!
jerryfreak, nice experiment. Thank you for it.
I don't agree with the last part of the final statement (in boldface); it even seems propagandistic to me--an attempt to make the shifting digital noise floor seem like a virtue when it isn't one. I don't mean that it's necessarily a defect, either, because if it's low enough at all times, no one will hear it shifting. But "it successfully evades detection" is the best that can be said about it if so.
--best regards
(Bolding above is my emphasis) Of course back then we were speculating on where potential issues could arise. The current hullabaloo about noise-floor windowing artifacts during the hand-off between multiple ADC's is a specific aspect of this. DSatz essentially called out the noise shift issue quite specifically. Going back and reading his comment about successfully evading detection brings to mind a another thought on a somewhat deeper level..
There is a deeper fundamental change going on in this shift to multi-ADC designs, which is a shift away from a recording system that is fully agnostic/isotropic in respect to the recording of whatever content fits within its bandwidth, to one which gives that up in achieving some alternate aspect deemed more valuable for the intended use case, even though doing so introduces problems for less-common uses. If evading detection of the ADC switching artifacts can be successfully arranged for the recordists who value not having to set levels far more than a truly isotropic data set, such a trade off is likely to be a welcome one for them - its good enough for their purposes by definition and they gain a quite welcome new feature, even if the scheme may not be good enough for outlier uses where a more fully isotropic data set is required for more esoteric applications such as recording ultrasonic signals, dramatically pitch shifting content, or whatever.
The machine has become specially optimized for certain uses, at the cost of being less optimized for others. Of course many other features of any recorder are optimized for its intended use, but the digital recording itself previously has not been.. unless recording to a lossy format.
I'm reminded of the development of things like noise-shaped dither, and psychoacousticly tuned lossy data compression. Those are useful tools, the successful implementation of which required careful determination of how much fidelity to the source is needed, prior to sacrificing what lies beyond the limits of perception. Its an optimization for certain use cases, which makes for a tricky game in determining where those limits might be, how they might be different for different uses, and how close one is willing to get to them in seeking to leverage human hearing perception to advantage. Noise shaped dither moves the bulk of the noise to where it is perceptually less obvious. Lossy codecs minimize storage requirements in part by discarding data deemed perceptually irrelevant. The fundamental shift is from a complete data set that contains extraneous information to one that is perceptually equivalent yet not fully complete. Fundamentally this is the same philosophical calculus based upon making a decision about what maters and what doesn't. If well implemented, it's not going to be s problem for most, but may for some. Its a sacrifice of true fidelity for all use cases, for easier use for intended target use cases.
-
Following up:
- How many ADC's do different recorders have (MixPre-3/6 ii, Zoom F3/F6, etc.)?
Presumably, one fix is to just add more ADC's to these little recorders, so instead of a 30dB noise jump on passages where it chooses the lo-gain ADC, you could have 5dB increments of background noise increase. I have heard that some 32bit recorders have 3 or more ADC's, but not sure which ones those would be.
-
The recording in that video that's getting folks talking about this is a recording of a water droplet.
Recordists making Foley or SFX recordings often use 60-70 dB gain to get those types of sounds loud enough to be usable.
I'm not surprised at all that after that type of gain adjustment that you are going to be able to hear system noise. Most people recording music or even dialog are not going to hear the low level interference from the ADC.
I'll keep staging my gain the old fashioned way using 24 bit.
-
Following up:
- How many ADC's do different recorders have (MixPre-3/6 ii, Zoom F3/F6, etc.)?
i believe the zooms have 2 and the the mixpres have 3, (according to mixpre patent filings)
-
Curious, do any the Zooms (F3/F6) or SD Mixpre II's have the option to still record 24 bit instead of 32 bit float?
-
Curious, do any the Zooms (F3/F6) or SD Mixpre II's have the option to still record 24 bit instead of 32 bit float?
MP-6 II can record either 32 or 24. F6 can record 32 or 24 or both and the F3 is 32-bit only.
-
Following up:
- How many ADC's do different recorders have (MixPre-3/6 ii, Zoom F3/F6, etc.)?
i believe the zooms have 2 and the the mixpres have 3, (according to mixpre patent filings)
^ I think so. The grist is how the switching between the multiple ADCs is handled. Can it be managed in such a way that the inherent change in noise-floor at the switchover (which is unavoidable unless some sort of noise reduction is done ahead of the switchover) is made imperceptible. To complicate things further I believe many multichannel ADC and DAC chips can be configured to combine their multiple internal units in parallel to incrementally increase dynamic range without switching between them. That avoids the switching problems but only gains a couple dB for each additional unit in parallel. Rather than switching between units set to target different levels, I assume that strategy does so by way of the signal being correlated across the multiple units while noise is not. I think that's how the strategy works anyway.. this goes well beyond my knowledge comfort zone. Perhaps both of these schemes are being used.
-
Curious, do any the Zooms (F3/F6) or SD Mixpre II's have the option to still record 24 bit instead of 32 bit float?
MP-6 II can record either 32 or 24. F6 can record 32 or 24 or both and the F3 is 32-bit only.
Keep in mind that the switching ADC architecture and the recorded file format are two separate, distinct things.
The relevant follow up question is: When the 32bit recorders that can do so are set to record 24bit files, does that mean only one ADC is used? Or is the machine still switching between ADCs in the same way as when in 32bitFP mode but writing a 24bit file? I suspect the second might be the case. If so it will not be a work around for the problem.
-
I'll keep staging my gain the old fashioned way using 24 bit.
I've stuck with that too so far. It does everything I need. That said, I don't expect this particular issue would be problematic for my use and what/how I record if I were to use a recorder which does multiple ADC switching.
The tl;dr of my post yesterday about the deeper fundamental issue: An audio recorder for music is not the same as a data-logger.
-
I am pretty sure that Zoom F series, and probably also SD, are still using multi ADC in 24 or 16 bit mode. IF anything can be done by Zoom it could be either: 1) allow a single ADC functioning when set to 24 bits mode or 2) provide a "smoother transition"between ADC with a sort of noise reduction or whatever they can figure out.
I have no technical knowledge to imagine any solution, but the posted YouTube video description it seems that Sound Devices has pretty fixed the issue with a firmware upgrade, so it is in theory possible to do something?
-
Since I am not recording water droplets and can hear absolutely no trace of noise during live music transitions, I will continue using my F3s with their great pre-amps and leave gain staging in my past along with my past fears of not demagnetizing my cassette deck heads often enough (or too often) or burning out my clutch in shifting my gears too hastily. YMMV
-
I am pretty sure that Zoom F series, and probably also SD, are still using multi ADC in 24 or 16 bit mode. IF anything can be done by Zoom it could be either: 1) allow a single ADC functioning when set to 24 bits mode or 2) provide a "smoother transition"between ADC with a sort of noise reduction or whatever they can figure out.
I have no technical knowledge to imagine any solution, but the posted YouTube video description it seems that Sound Devices has pretty fixed the issue with a firmware upgrade, so it is in theory possible to do something?
This point is addressed in the video. All of these recorders that have multiple ADCs still have all of them in the signal path even if you are reporting in 24-bit fixed mode.
Your possible solution #2 is what I understand Sound Devices did in their fix. I don't think we should assume the same will be possible in Zoom recorders. Not because of the different number of ADCs, but because the DSP hardware and software is likely to be significantly different between the two manufacturers.
-
This point is addressed in the video. All of these recorders that have multiple ADCs still have all of them in the signal path even if you are reporting in 24-bit fixed mode.
Your possible solution #2 is what I understand Sound Devices did in their fix. I don't think we should assume the same will be possible in Zoom recorders. Not because of the different number of ADCs, but because the DSP hardware and software is likely to be significantly different between the two manufacturers.
I think the biggest reason is that zoom doesn't care, and neither do most of their customers. If enough people complain, the f3n or f8x pro or whatever will have a different strategy.
-
Since I am not recording water droplets and can hear absolutely no trace of noise during live music transitions, I will continue using my F3s with their great pre-amps and leave gain staging in my past along with my past fears of not demagnetizing my cassette deck heads often enough (or too often) or burning out my clutch in shifting my gears too hastily. YMMV
Sure, you can do that as long as you are always and only running your mics direct in to the recorder. If you ever use an external preamp, or run any other signal chain consisting of multiple units into your F3, the gain staging ahead of the recorder will matter.
Gain staging will always remain relevant and important. Its not tied to any specific technology. It matters just as much inside the F3, the thing is Zoom has engineered it so you needn't be aware of it. In fact, the issue we've discussing is how the internal gain-staging across multiple ADCs is being managed somewhat imperfectly!
OT- Long live the manual transmission and drivers who know how to work them.
-
OT- Long live the manual transmission and drivers who know how to work them.
Most cars in Europe have manual transmissions and diesel engines. Cars are as disposable as bic lighters in America, they're most people's primary life investment in Europe, since people are way less transitory, and homes are oftentimes handed down from generation to generation.
-
Since I am not recording water droplets and can hear absolutely no trace of noise during live music transitions, I will continue using my F3s with their great pre-amps and leave gain staging in my past along with my past fears of not demagnetizing my cassette deck heads often enough (or too often) or burning out my clutch in shifting my gears too hastily. YMMV
Sure, you can do that as long as you are always and only running your mics direct in to the recorder. If you ever use an external preamp, or run any other signal chain consisting of multiple units into your F3, the gain staging ahead of the recorder will matter.
Gain staging will always remain relevant and important. Its not tied to any specific technology. It matters just as much inside the F3, the thing is Zoom has engineered it so you needn't be aware of it. In fact, the issue we've discussing is how the internal gain-staging across multiple ADCs is being managed somewhat imperfectly!
OT- Long live the manual transmission and drivers who know how to work them.
Thanks for the blessing. I drove manual transmission cars for decades. Modern automatic transmissions are now more efficient than manuals. Thank goodness. I am so old that I remember manual choke knobs on my first cars.
-
Most cars in Europe have manual transmissions and diesel engines. Cars are as disposable as bic lighters in America, they're most people's primary life investment in Europe, since people are way less transitory, and homes are oftentimes handed down from generation to generation.
At least in the Netherlands, this isn't really true. Most cars are manual (75%?), but the shift has been large and the rate of change is increasing as well. The vast majority of passenger vehicles run on gas, though. Probably ~80% gas. Of the other 20%, diesel is a substantial, but quickly decreasing, chunk. Pretty large increases in hybrids and electric the past few years. I also think people move/travel at least as frequently as Americans. I know more people from back home that never crossed the PA state line than Dutch that haven't traveled overseas...
[EDIT TO ADD] Manuals are considerably cheaper, so that is probably a reason they continue to sell. It is more expensive to own/operate a car here.
And, on topic, I would wager the reason SD was able to address this issue more easily is because their patented method basically estimates the instantaneous rate of change (i.e. first derivative) by running regression on a small number of samples. Any curve is essentially linear over a tiny interval, so they can leverage that to accurately estimate the shifts over very short time spans.
Also, there is no such thing as a free lunch. Maybe cheap, but not free. I suspect this is the reason that SD doesn't offer 32-bit float on their pro level recorders.
-
Continuing the OT-
I'm curious if a particular driver behavior change I notice here might be the same in the Netherlands and elsewhere, which may be partly due to the shift away from manual transmissions but I think is far more driven by mobile phone usage-
Decades ago, drivers would generally tend to coast to a red traffic light, then take off quickly as soon as the signal changed. That behavior seems to now be completely inverted- Drivers tend to rush to a red light, seemingly braking as late as possible, then slowly take off after the signal changes.
Anyone else notice this? I assume its because they are impatient get back on their phone while stopped at the light, and are slow to quit it after the change.
It's somewhat annoying as I'm still coasting, timing the light changes to preserve inertia, shifting manually. The new behavior folks speed past and swerve in front during the coast, then sit there after the light change, screwing up the well preserved inertia and timing. [shakes old man fits at cloud]
I do wonder if my current manual 6-speed (2013 era, its so good - manual transmission finally perfected) may unfortunately be the last of its breed. Its been a very long time since I was adjusting valves and timing, and longer still since I was working a choke-knob and shifting "3 on the tree"! Where's that old wire-recorder?
-
At least in the Netherlands, this isn't really true. Most cars are manual (75%?), but the shift has been large and the rate of change is increasing as well. The vast majority of passenger vehicles run on gas, though. Probably ~80% gas. Of the other 20%, diesel is a substantial, but quickly decreasing, chunk. Pretty large increases in hybrids and electric the past few years. I also think people move/travel at least as frequently as Americans. I know more people from back home that never crossed the PA state line than Dutch that haven't traveled overseas...
Yeah, my Euro experience is from 2007-2009 from living in Slovenia, so I shouldn't have assumed it was still the same today. Somewhat surprised though about the shift to gasoline. Not surprised at all about the shift to electric/hybrids.
I'd be curious to know more about transitory lifestyles in the US versus Europe. I know that in Slovenia, with only 2 million people, they're much more internally focused on Slovenia first...15 years ago anyway. There weren't alot of specialty manufacturers and/or industry and people were so proud of Slovenia...in general they had absolutely no desire to live anyplace else. The majority of jobs are people supporting other people. There's not nearly as much of a profit and loss focus by multi-national mega-company's. So people aren't as subject to economic cycles as in the US, where virtually nobody goes through their lives anymore without experiencing 2 or 3 or 4 layoffs. In Slovenia, employers were much more about providing a company that supports the lifestyles of its workforce than they were about making profits for their shareholders. At least that was my perception.
I hope it hasn't changed, honestly. I loved the fact that, in Europe, entire countries shut down for 6 weeks for summer holidays because, for six weeks anyway, the focus is on the people and the family and not the company.
-
Thanks for the blessing. I drove manual transmission cars for decades. Modern automatic transmissions are now more efficient than manuals.
about as efficient perhaps, but its impossible to be more efficient than a direct clutch connection (with a capable driver of course)
-
All this car talk has me thinking, is 32 bit float the CVT of recording formats? It's the most efficient transmission made, yet real driving enthusiasts hate them.
-
I think a dual clutch automatics (which are essentially manual transmissions operated automatically) are more efficient than a CVT, due to less friction loss through a direct clutch connection. They compensate for the lack of the CVT's infinite gear ratio by adding additional gear sets - eight or nine I in some cases I think. They are more efficient than a true manual for three reasons: There are additional gear sets to choose between. The dual clutch arrangement means any two adjacent gear sets are always simultaneously engaged and dual clutches switch between them, so there is no "down time" when the engine is disconnected from the drive train during a shift. And the shifting is computer managed, cyber vigilantly paying attention to the situation without distraction to optimize efficiency to the ability of its programming.
I just like doing the left foot right hand dance. Paddles just aren't the same. And operate the CD player at the same time. ; )
-
I have not read all of the above, but as someone who has been taping since 1971, I love the 32 bit world we have entered. Why? Because I have made great recordings. I have had no issues at all. The biggest difference is that I do not have to occasionally spend a inordinate amount of time leveling the first 10 or 20 seconds of a recording because it is too hot or too low. I have a few decks and do not at all mind going 24 bit when one of those makes sense, but using the same gear I have used for years, 32 bit is my choice. For me, how the end result sounds is what matters most. That said, I respect and admire all the dialog dissecting the respective differences in how we approach our passion.
-
I have not read all of the above, but as someone who has been taping since 1971, I love the 32 bit world we have entered. Why? Because I have made great recordings. I have had no issues at all. The biggest difference is that I do not have to occasionally spend a inordinate amount of time leveling the first 10 or 20 seconds of a recording because it is too hot or too low. I have a few decks and do not at all mind going 24 bit when one of those makes sense, but using the same gear I have used for years, 32 bit is my choice. For me, how the end result sounds is what matters most. That said, I respect and admire all the dialog dissecting the respective differences in how we approach our passion.
exactly.
-
I skimmed the Sound Devices 2016 patent for 32 bit (most of which is way above me) and note that it does describe 3 and 2 converter layouts, but noting it could be any number. I did not recognize any comment about the converter handoffs and noise control, but that could be me. I note that other manufacturers have to avoid infringing on their patent, so maybe SD understands the issue better and has the secret sauce in their patent. It would be nice to know if SD uses multiple converters in 24 bit mode, I haven't heard any issues and have not done much 32 bit.
On the clutch, I just prefer manuals as more fun. Today's automatics get better gas numbers and track lap times, but on the street that's not everything for me. I like road and machine feel, although most Americans apparently prefer automatics. I also don't use the phone while driving, even hands-free (legal requirement here).
As to the anti-theft aspect, I had an Integra manual for many years with a Club steering lock that was stolen 3 times, the last time driven 100 miles away and totaled. Integras were theft targets, so the thieves certainly were not stopped by a clutch or Club.
-
I have not read all of the above, but as someone who has been taping since 1971, I love the 32 bit world we have entered. Why? Because I have made great recordings. I have had no issues at all. The biggest difference is that I do not have to occasionally spend a inordinate amount of time leveling the first 10 or 20 seconds of a recording because it is too hot or too low. I have a few decks and do not at all mind going 24 bit when one of those makes sense, but using the same gear I have used for years, 32 bit is my choice. For me, how the end result sounds is what matters most. That said, I respect and admire all the dialog dissecting the respective differences in how we approach our passion.
exactly.
Ditto, from a guy who started taping in 1972.
-
Every car I owned from 1979 to 2011 was a manual. By that point years of driving in stop & go traffic keeping my left foot pressing the clutch most of the time had wiped out all the romance of a manual.
As for taping, the last deck I bought from Doug Oade is the Marantz PMD-561 with concert mode and the way he set the gain stages make it "almost" set and forget. He set the preamp gain so the overload point of the analog circuit and the clipping level of the A/D are matched perfectly. This allowed me to find a sweet spot to set the gain at the same starting position for 90% of shows that I never have to adjust again, or just once at the very beginning. Not the true "set & forget" convenience of 32bit float, but for a guy who used to make a dozen unecessary micro-adustments at every show I'm now comfortable only checking that the deck is running every now and then, which for me is huge.
-
Every car I owned from 1979 to 2011 was a manual. By that point years of driving in stop & go traffic keeping my left foot pressing the clutch most of the time had wiped out all the romance of a manual.
As for taping, the last deck I bought from Doug Oade is the Marantz PMD-561 with concert mode and the way he set the gain stages make it "almost" set and forget. He set the preamp gain so the overload point of the analog circuit and the clipping level of the A/D are matched perfectly. This allowed me to find a sweet spot to set the gain at the same starting position for 90% of shows that I never have to adjust again, or just once at the very beginning. Not the true "set & forget" convenience of 32bit float, but for a guy who used to make a dozen unecessary micro-adustments at every show I'm now comfortable only checking that the deck is running every now and then, which for me is huge.
Concur with the point of about losing the romance of shifting in traffic, especially when stopping uphill is involved. My German-built six speed automatic is wonderful. As for taping, yeah, my Oade-modified Sony M-10s are also great in the regard that you mention. No need for an external pre-amp and the sweet spot always seems to be at about three on the gain wheel, except in the quietest situations. That said, i still prefer the total ease of 32 bit Zoom F3, along with the security of the tight XLR connections.
-
Driving a four speed manual transmission around SanFrancisco for a few years was fun. Parallel parking on the steep hills was actually made somewhat easy by gravity. Other drivers pulling up too close behind at a light on really steep inclines was worrisome though. For I while to activate the starter I had to crawl under, so I'd almost always just park on an incline, roll out and pop the clutch. My current manual has a hill-hold feature which makes it almost too easy- if the car senses it is on an incline, the brakes remain engaged for 5 sec after releasing them or until the clutch is engaged. Almost no hill-skills required.
Here's an idea, the next round of 32bit float recorders could implement a feature that normalizes the 32bit float file after recording and writes a 24bit file from that. No loss, and everything fits in a smaller, standard fixed-point format file. Sort of a 24bit hill-hold manual in comparison to a 32bit-float auto-trans? Yeah, reaching a bit there.
-
My current manual has a hill-hold feature which makes it almost too easy- if the car senses it is on an incline, the brakes remain engaged for 5 sec after releasing them or until the clutch is engaged. Almost no hill-skills required.
I know this is still off topic but my GFs current ride is a 6 speed manual Mini Cooper and it has the hill brake thing. The first time we took a road trip and I was driving I had to stop on a really steep hill so I did the hand brake with the button pushed in thing to keep me from rolling back or stalling out and she just laughed at me. I learned in a three on the tree.
-
Back on topic....Given that the drawback of ADC switching might be a rise of noise floor, the point is that for some applications it may be audible, for others not.
Whatever the final delivery might be, i.e. 24bits 48khz, is there at your opinion a difference between recording 48, 96 or even 192 khz in order to minimize the potential issues?
-
I just pulled up the last 32-bit float recording I had made to find out if I could see and hear the problem being described.
Short answer: See? Definitely. Hear? Absolutely not.
See the attached screenshots from RX. This recording was 32fp/96kHz using my F6.
1. Hand Claps - You can clearly see the windowing as the noisefloor increases around the louder transients.
2. A Cappella - The three transients just right of center are accented syllables in the choir. You'll also see stray moments before and after where the converters switch.
3. Noise Zoomed - As it says on the tin. This shows why the added noise is inaudible - it's only showing up above 35 kHz or so.
So, is this still a concern from a sound design standpoint when captures will be pitched down, as shown in the video? Sure.
Is it a concern for studio or live music recording? Well, the noise floor is constantly shifting with the auto-ranging ADCs, at least on a Zoom F6. It's not just on the stray loud transients; it's happening frequently. But the noise is in a place that no one can hear. While I don't like that this is happening at all, it's not causing any problems in the audible range.
If anyone has a Sound Devices MixPre-II or one of the newer Tascam recorders that have auto-ranging ADCs, it would be helpful to compare their behavior.
EDIT: A user on GS has tested a Stagetec Nexus Compact and it does not do this. Maybe it's a Zoom implementation issue?
https://gearspace.com/board/all-things-technical/1425826-problems-32-bit-float-sound-design-pitching-down-white-noise.html#post16963512
-
Thanks @voltronic for posting your results. Most probably Stagetec or others have a better implementation…it is interesting though to read that Sound Devices seems to have fixed the issue via firmware update. This means that the issue was there before and that something has been possibly done. Don’t know wether Zoom will bother finding an improvement, should it be technically possible. Given the huge number of F series users- among which probably a lot of people recording and manipulating sounds- I can expect that some of all these complaints will eventually reach them and maybe lead to a fix.
The better behaviour with a SD788 is maybe due to a better hardware? It would be interesting to try the Zoom with an external preamp to tell how the overall tone of the recording changes.
-
the 788 works in a fundamentally different way.
-
From the manual and other statements made by Zoom it seems that the F serie uses two 16 bit converters; Zooms says that both in 32 bit and 24 bit mode the dual ADC is engaged.
This leaves to think that in 1 bit mode only one is working? If so it would be interesting to compare the spectral view of a 16 bit recording with a 32 bit one. Not to conclude it's better to go back at 16, just out of curiosity.
-
From the manual and other statements made by Zoom it seems that the F serie uses two 16 bit converters; Zooms says that both in 32 bit and 24 bit mode the dual ADC is engaged.
This leaves to think that in 1 bit mode only one is working? If so it would be interesting to compare the spectral view of a 16 bit recording with a 32 bit one. Not to conclude it's better to go back at 16, just out of curiosity.
I'm not sure where you got this impression, but neither of the ADCs are limited to 16-bit. One is optimized for high gain; the other for low gain.
Both ADCs are engaged no matter what recording mode is selected. You can't choose to use only one, and that's a point addressed in the video.
-
I’m sure I read it somewhere but it’s probably not true. I was maybe tricked by the fact that Zooms says that both 32 bit and 24 adopt the dual adc, but no mention about 16 bit.
-
YouTube poster has taken his video "private" for some reason, so hard for newcomers to see what the fuss was about;
https://youtu.be/y3431uljZ2k?si=I7I7qhsKK7TUo49s