I was wondering if anyone is using 32bit float audio interfaces with Gig Performer and would it be possible or there is a limitation to this? There are field recorders that can also work as audio interfaces with 32-bit floating point data streaming but I can’t find anything regarding this topic on this forum.
Zoom has released today their new UAC-232 audio interface and it looks very promising to me. I like the appeal of not thinking about gain and ADC clipping. For those who wonder how this works, they use dual ADCs with different gain levels and stich the two signals into one 32bit float value.
I know that Gig Performer runs audio streams internally in floating point (not sure if it’s single precision or double precision) so for conventional 24bit audio interfaces there needs to be a conversion from integer to float. If this is done by Gig Performer and not by the driver or OS, then Gig Performer could be expecting integer values and not work with floating point data streams.
My first thought is (in a live setting) that the mixing engineer will probably still use the gain, just to have all levels more or less in the same range.
In a recording context, I agree, it might prove useful, because getting the levels right can be (and I think will be) done afterwards.
The dynamic range it offers does (live) not seem very useful to me, especially given the number of compressors that is being deployed to manage the dynamic range.
Of course it is not bad to have such an interface, but I’m not sure how much DAWs and of course GP have to be altered before they can benefit the greater bit-depth. That’s a question @dhj could answer better.
Maybe in 10 years all interfaces are 32bit. Someone (supposedly) said about personal computers: 640KB RAM is enough for anyone …
I could not find that, at least not referring to using to smaller ADCs to create one 32bit. If they do, then it is susceptible for having distortion at the levels between the two ADCs, the same way the traditional R2R resistor networks in the older ADCs had distortion, because of the tolerance of the resistors. If I remember well, the solution was to revert to 1 bit ADCs with very high oversampling. (This statement is highly questionable and may be just plain stupid, so don’t take my word for it! I do not see how this might work, although I can’t rule it out either)
These are just a few thoughts that come up in me. This isn’t meant to keep anyone from buying such a device…
I must add that to me there is other gear I would be after, as long as it’s about live performance
Perhaps there are imperfections to combining signals but that is a worthy tradeoff in my opinion. There are certainly ways to mitigate some of these issues. This interface should sound exactly the same as the F6 field recorder, and as far as I can tell people are very pleased with those devices.
Sound Devices MixPre recorders use triple ADC circuits with extremely good results…
The fact that there is no gain to adjust makes that there is a constant 1:1 correlation between the analog and digital world. And this bring the digital much closer to having a real analog device.
The question here was just about support for 32-bit floating point in Gig Performer. There is a fallback mode by running the audio interface in 24bit if necessary but that would be counterproductive.
I’m not at all familiar with how this is or might be implemented in Zoom devices, but at the moment I’m not seeing a meaningful practical value for what (I think) the vast majority of us (GP users) are doing.
My thinking - there are really two parts to this that can’t be completely separated. There are the physical limits of your recording and A/D conversion equipment, and the encoding format of what they capture.
Sound happens in the physical world with sound waves. Those are generally captured by microphones, which have their own physical limitations on the dynamic range they can capture. If the diaphragm on your microphone hits a physical limit, you’re physically clipped. If the sound you are trying to capture is below the noise threshold, it’s lost.
If you’re trying to record a whisper from 1000 yards and then a scream from 5 feet, I don’t think the encoding format (32 bit float or 24 bit) is going to be your problem. Maybe somewhere out there there’s a microphone that can handle that, but I’ve sure never seen one.
This is really just the inverse process of playing sound back. If somebody told you that your speakers would sound better if you used 64 bit or 128 bit audio, would you believe it? That you could play those whispers from 1000 yards and then the bomb explosion without ever touching the volume control, do you believe it? If so, you have better speakers and amps than I do.
I see the point of not needing more than 24 bit for sound ADC conversion and I agree with that. But that is not the point here. The point is about convenience and consistency. Not having to worry about clipping and always expecting the same sound.
The first thing that came to my mind is one of my friends that is not very knowledgeable with regards to digital audio recording or processing. A device like this is perfect for him because there is very little way for him to mess up while recording or playing live.
I have no plans to specifically look at this in the near future - I agree with @Vindes that it’s just not that meaningful for the intent of GP.
GP receives audio from the underlying OS (possibly by specialized drivers depending on the actual audio interface) as 32 bit floating point and sends it back that way.
I’m waiting for the reviews to see if I should get one myself. I am in touch with someone that will be reviewing this device in the near future so I’ll ask him to test the compatibility with GP.
…
