32bit floating point recording question

I’m genuinely enjoying this discussion. As long as you are comfortable let’s continue. In the case of Linn, they are recording at 192k with pristine microphones, preamps, beautiful spaces and the like. It’s not like they are recording at 44.1 and upscaling to 192 to pull a fast one. I don’t think marketing plays a huge part in this, honestly. They were one of the first to offer it so it wasn’t as if they were needing to compete. Physics says that 44.1/16 is plenty for dynamic range and frequencies that the best human ear can respond to but it does not address the “rate” part. 96k captures complex tails better because there are more sample points than at 44.1k. Has anyone studied at what rate the human ear can discriminate sample points? In down-sampling this has a better effect than simply recording straight at 44.1k. There are plenty of folks in the business with “golden ears” who can reliably tell the difference a good deal of the time. I’m not one of them unfortunately but something “feels” different when I switch between SACD and regular CD mode on my Onkyo player. As someone wise recently told me, music isn’t just about frequency…it is also about time.

did you read the link posted above? it adresses almost all the topics of this discussion in a scientific way

We had that discussion a lot here and in other places, so I m not sure how far this still makes sens… in the case of Linn, I dont know, they surely have great sounding recordings and let me be provocative: I think they would even sound better when recorded at 44kHz/24bit and dithered beautifully to 16bit.

" I’m not one of them unfortunately but something “feels” different when I switch between SACD and regular CD mode on my Onkyo player. As someone wise recently told me, music isn’t just about frequency…it is also about time."

I get that totally. And whatever sounds better subjectively IS better for your personal case. Noneoftheless there are various issues about verifying if it is “objectively” better, on the recording side, and on the playback side: Where the SACD and the CD mixed and mastered exactly the same? maybe the internal SRC of your onkyo player is better than the one they applied in the studio? would the SACD still sound better if the CD would be recorded and mixed exactly the same way but at44kHz with no SRC? if the SACD is just mastered 0,1 db louder, it might appear to sound better. no audio system can reproduce 24bit, so what is happening at that stage in your hardware? does the dac in your system has the exact same circuitry? and there are a lot of more questions. …

Check also the nyquist-shannon theorem the digitisation process relies on.

but then: imo whatever sounds and feels good is fine on the music part, because it is about the fun of it. It s just not right to use myths and marketing to justify a supremacy of a recording chain no one can scientifically proof it is better (and that for the time being is proofed to be worse). The reason for that I think is also there is just so much money in marketing for products that is misleading and no one will spend millions in ads on the nyquist-shannon theorem, too bad :slight_smile:

1 Like

Yes, I’ve actually had that bookmarked in my browser and read it a good number of times – it makes very convincing reading! I want to believe it is true, I really do, and came close to releasing everything I do purely at 44.1/16 and I still might :wink: In terms of “beautiful dither” what do you use? Good ol’ triangular, shaped, “non-dither” a la airwindows?

To clarify, you would consider the whole higher samplerate / complex tails phenomenon nonsense too? I may be overthinking things but I’ve tried to use the Xiph article in discussions previously and it has been shot down as not telling the whole story. They say that phase and aliasing are the most important factors when listening to music as we do it both in the frequency and in the time domains. 44.1, they claim, is not good enough to avoid phasing issues and why perhaps people still enjoy vinyl and tape because it is more relaxed.

I appreciate the time you have taken to discuss this and what you say does make a lot of sense. I hope you agree that it’s healthy to question the various viewpoints and what Xiph has going for it, obviously, is that it isn’t trying to market anything :wink:

I have some real estate on moon you may be interested in. I can get it for you at quite a good price.

1 Like

Brilliant…so generous. But an immediate concern – I wouldn’t be able to hear 32-bit float dither :frowning:

@anon60445789 I usually use the shaped noise dither within ardour after some reading I did. Not that I remember any of the arguments exactly but it sounds good to my ears and I m happy with it! I m a fan of some of airwindows plugins and would like to test how to dither with his method… how would i do that? apply his dither plugin while exporting, without applying the internal dither option in ardour on export?

The phasing issues: what are they? I have not heard that before…

And yes imo the analog <–> digital discussion is a whole other thing better not to get into in this thread. but within the digital realm, things can be pretty binary so to speak :slight_smile:

Exactly this. On the master bus I use NJAD24 and NJAD16 as part of his StudioTan plugin but his new monitoring plugin will also allow by selecting either Out24 or Out16. Either way, disable any DAW dither and you will be good and be sure to leave master fader at unity so that dither (or “non-dither” in this case) is the very last calculation before export. If you watch Chris Johnson’s videos on NJAD, StudioTan etc you’ll get a great idea about the math behind it (Benford realness calculations) and his claim that even NJAD16 will give a CD a sense of infinite resolution.

1 Like

Actually for a design like this it is more (excuse poor ASCII diagram)

           /|----- preamp gain 1 -----> ADC1 --|\
audio in -<-|----- preamp gain 2 -----> ADC2 --|-> DSP to selectively blend between three inputs
           \|----- preamp gain 3 -----> ADC3 --|/

The thing to watch out for is that the noise floor can change with signal level as the DSP fades between the three (in the case of Sound Devices; four in the case of Stagetec) different preamp gain/ADC combinations. You have to be aware of that fine print when looking at the specs, and understand what the measurements mean. For example, Stagetec claims that their input stage has a 158dB SNR (with 32 bit integer output, 143dB when converted to 24 bit output). I don’t believe that a 158dB SNR is physically possible, I suspect that if you looked at the noise floor and signal at any one signal level you would see something more like 120dB SNR, but because of the way the output is derived from multiple input stages with different gain settings the 120dB of SNR is moved around to different places in the available code points of the output word.

I have not heard one of those gain ranging converters myself, reportedly it sounds OK when done well. It obviously gets pretty expensive since you are effectively taking three or four normal input channels and then adding on additional DSP. I’m also not sure how the input clamping is handled, the stages with higher gain that get selected for low level signals must get overloaded when the signal level goes higher, so something has to take that into account in the circuit design so that when the signal goes below the overload point those high gain stages recover back to full performance very quickly.

It does sound very appealing to just plug in a signal source and not have to worry about gain staging. Maybe calimerox can report back in a couple of months and let us know how it works out in practice.

1 Like

Well that doesn’t look right at all. Something ate the extra spaces I put in to try to make the ascii bars line up. Hopefully you can figure out what I was trying to do, else I can try to paint up a png file and see if discourse will let me add that in a forum post.

Fixed it for you, utilize three backticks to create blocks of preformatted text:


See Code Block

1 Like

Awesome, thanks. I was looking back and thought hmmm…pretty sure it didn’t look that nice a few minutes ago. :wink:

I now have my MixPre-6 II in hand and will hopefully have a chance to do some live concert recordings in the near future. The nice thing is that even if the 32-bit float doesn’t produce excellent results, there’s always standard 24-bit recording with lovely Kashmir preamps. The test files from both SoundDevices and Zoom do seem like these will be game-changers, particularly for wide dynamic field recordings and last-minute concert captures.

1 Like

excellent! I m curious about your experiences! I think so too, for set recording this will make a huge difference, and for very dynamic music to be recorded too…

Just adding this to the discussion:

I had read the wordy version as posted by the OP but this gives some nice graphical analysis as well as access to the sound files for null-testing. It seems like they deliberately recorded at the two extreme levels of gain knob turned all the way down and all the way up. I think the lesson here (provided null-tests even reveal slight differences that wouldn’t be noticeable during normal music playback) is to just set the gain approximately and just sit back and relax knowing capture will be great. And who is to say that any slight null-test differences are not due to Reaper 32-bit/64-bit float fader calculations when truncated to 24-bit output? Unclear why Reaper is being used actually when this could all have been done in RX…

1 Like

I need to add one further post because, based on reading a thread with Paul Isaacs, my latest understanding is that the Sound Devices new Mixpres operate in the exact same way as the Zoom F6. In both brands, 32-bit float mode uses the gain knobs purely as post-converter digital gain faders (i.e. they introduce zero noise). In other words, you cannot set traditional gain levels in this mode. The only difference is that the Zoom uses 2 preamp/convertors combined in a clever way and the SoundDevices, 3. So, my earlier thinking that I would set a healthy approximate gain on sound check is completely unnecessary (and impossible) other than to get a decent monitoring level. You really can leave the “gain” knob at its lowest setting and bring it up in post with no downsides. Someone jump in and correct me if I’m wrong!

If it is true that any gain/level control is post digital conversion and post conversion to float, then that is correct. There would be no sound reason for setting gain before converting to float so if there is no possibility of setting analog gain then the record level is not relevant. In fact the level control may not even affect the output digital level at all only the monitor level. I think for such a method of conversion to work right, the input level may require a factory setting or auto setting… though that doesn’t sound right either. The mic type would make a difference too. There is a difference between dynamic, condenser and ribbon mics for sure. Maybe there is a switch instead of a continuous level for that though.

This begs the question: If some sort of auto level is included, does that reduce the dynamics of the audio being recorded? Or is there a setup before recording to get it right before recording so the level(s) can be locked? In theory, it should be possible to get a better setting by doing this step manually but even more complex than setting up a normal single conversion level. If there is no auto analog level setting, then the mic used matters (note that the zoom comes with it’s own mic, at least the ones I have seen). Of course with 24bits of resolution (for 32 bit float), which is probably 8 more than needed, the extra 48dB of range may cover this quite well. The three converters may be as much as 20dB apart. The level range from my ribbon to my condenser mic is 25 to 30 dB and would fit into that range with no problem.

I believe that the level control will affect how the 32-bit float file is printed or at least that’s how both the SoundDevices and Zoom are presented when showing recordings at lowest, middle and highest “gain” settings. i.e. moving the knob all the way to the right will print a file with the waveforms looking like they are clipping given enough input. Of course, the beauty of the float file is that reducing the gain in the DAW will reveal the unaffected peaks.

There’s no “auto” level, simply a set gain structure that has a wider dynamic range than any currently-made microphone. All dynamics would be available.

Not true on the F-series models. The F6, currently the only model that includes 32-bit float recording, has 6 xlr connections. The 32-bit stuff is only dependent on the preamp/converter stage.

My understanding is that the unknown “magic” is the way the companies glue together the final 24 bits (23-bits + sign) with the 8 bits of exponent for the digital gain by taking, say, 12 bits from one preamp/converter level and 11 bits from another (Zoom) and a three-way equivalent, say 7+8+8 for the SoundDevices. I suppose it isn’t that simple but that’s how I’ve seen other people talk about it. Perhaps the distribution of bits is not as equally distributed, for example.

EDIT: I should also add that it really is impossible to clip the internal circuitry of both the Zoom and SoundDevices as the uppermost limit of the gain structure aligns with (or is a hair under) that clipping point.

That does not make sense to me. In my mind there would be three ADCs running in parallel, each with a different gain. ADC 1 would be set at 0 dB, ADC 2 would be -20 dB and ADC 3 would be set at -40 dB (for some suitable value of 0dB). When quiet sounds are encountered, ADC 1 would have the greatest number of bits representing the audio and so it’s output would be used and converted to 32bit float and then it’s gain reduced by 20dB. When the input got loud enough that ADC 1 was over loaded then ADC 2 would be used converted to float with no gain change. Then when the input got even louder and ADC 2 is now clipped, the input from ADC 3 would be used and of course after being converted to float the level would be increased by 20 dB.

This setup would require that the 20 and 40 dB pads in the analog section were precise (do-able) and that the ADCs used are matched (all 6 of them for stereo). Also note that the ADC output would be chosen on a per sample basis so that even on loud sounds, a sample that is closer to a “0” value (not talking dB here but rather 24bit int) would still use ADC 1. It would be much easier to explain with a drawing, but I am not that kind of artist :slight_smile:

Please also note that my choices of dB steps is only for illustration purposes and may be wildly different from the reality. I do expect they would use “even” amounts of gain so that the 24bit part of the float could remain the same and only the exponent would need to be changed… well actually it would not be changed but rather a specific exponent would be added to the output of each ADC when converting it to float so that there would be no math needed. Switching would be achieved by simple “less than” “greater than” which could probably be done with simple register compare rather than real math. Not magic really, but more than 3 times more costly. I do not know if the gain is really worth the cost… yes you could use real canons … but in the end the output will be restricted to 16bits or maybe 24bits and the speakers would need to have the available power output equal to said canons. Hardly front room listening levels.

Not having to set levels in a high paced film recording situation would be another use. As for the actual sound, one would have to do double blind AB testing. Better if you have one to just enjoy…

My head hurts. Diagram, please :wink:

In all seriousness, yes I plan to just enjoy it. But my brain does like to figure out what is going on. Paul Isaacs refuses to go into details other than about the “gain” being digital post-preamp/ADC.

Nah, that is the canons going off…