Zoom UAC-232 32-Bit Float Interface

[joeyluck]

https://zoomcorp.com/en/us/audio-interf ... s/uac-232/

This is the first dedicated 32-bit float audio interface, which isn't counting their 32-bit float recorders which can work as interfaces.

• Dual AD circuitry with 32-bit float recording technology
• High-quality mic preamps, same as the Zoom F6
• 2x XLR/TRS combo inputs
• Headphone output and 2x TRS main outputs
• High-power headphone outputs (100 mW)
• Records up to 192 kHz sample rate
• Compatible with Windows, Mac, and iOS devices (Android not supported)
• Zoom Mix Control software mixer
• MUSIC and STREAMING modes available via the software to optimize I/O
• Loopback available via the software
• Direct Monitor switch for latency-free recording
• Vertical or horizontal use
• MIDI IN/OUT for connection to external MIDI equipment
• Bus-powered operation (separate power required for Lightenting-enabled iOS devices)

[bxbrkrz]

Hmmm

[O1B]

Nice find.
Quasi-Compact. $200. Stylish. 32 bit.
“The future is near!”

[DaveyG]

Interesting stuff and hopefully the first of many such interfaces across the industry. Gain controls? Who needs 'em...

@O1B - that clip made me laugh. It's been a long time since I watched that movie. A classic that they probably wouldn't be allowed to make these days.

[MuttReason]

Interesting bit of gear. This may be ignorance on my part, but if you record an audio signal at super low level and then massively crank up the gain of the recorded track when you’re mixing, you’re still going to bring up the noise floor too, right? As in unwanted background ambient noise, hiss and chatter from within the audio circuit itself etc. 32bit doesn’t change that, does it?

I get how setting the input gain just high enough to get a good clean signal but not so high as to risk clipping is easier with the dynamic range of 32bit vs 24 or 16. I’m just not sure it’s true to imply that gain staging doesn’t really matter at all when, at very low input levels, I’d have thought it still matters a lot. This may be down to my misunderstanding, though…

[avasopht]

I wonder, ... is this just doing roughly the same thing as balance (additional converters set at different gain levels).

And surely there must be a limit. Pretty sure that if I supplied a signal that went straight from -1000dBVU to +1000dBVU, you'll be able to tell

[DaveyG]

If you google it you'll find some good explanations of how these things work and Zoom have a decent section on it here:
https://zoomcorp.com/en/gb/news/32-bit- ... d-to-know/

It's usually two overlapping converters giving a huge dynamic range with the result converted to 32 bit float. So the practical limit of what you can usefully record comes down to the microphone physics. If you are recording something very, very quiet you are still subject to the residual nose of everything but that's also true if you have a gain control. The real benefit of 32 bit float is that you don't have to set the input level because very loud sounds can't clip the signal until you overload the mic itself and very quiet sounds don't suffer from lack of bits to store each sample so you can amplify/normalise them in an editor after the event without "pixelating" them (yeah, this is a simplification...)

Of course, you do have to ensure that all the software tools you are likely to use also support importing and editing in 32 bit float or better. I had a few false starts trying to work with these files on an iPad, even with some editors that boast floating point editing.

[jam-s]

And surely there must be a limit. Pretty sure that if I supplied a signal that went straight from -1000dBVU to +1000dBVU, you'll be able to tell

Sure, if you give it too much input voltage, you'll be able to fry the input electronics of the interface.

[joeyluck]

Interesting bit of gear. This may be ignorance on my part, but if you record an audio signal at super low level and then massively crank up the gain of the recorded track when you’re mixing, you’re still going to bring up the noise floor too, right? As in unwanted background ambient noise, hiss and chatter from within the audio circuit itself etc. 32bit doesn’t change that, does it?

I get how setting the input gain just high enough to get a good clean signal but not so high as to risk clipping is easier with the dynamic range of 32bit vs 24 or 16. I’m just not sure it’s true to imply that gain staging doesn’t really matter at all when, at very low input levels, I’d have thought it still matters a lot. This may be down to my misunderstanding, though…

32-bit float helps with low levels as well. It's really awesome It's like magic.

https://www.sounddevices.com/low-signal-32-bit-float/

https://www.sounddevices.com/sample-32- ... wav-files/

Reason supports 32-bit float import in Reason, but just doesn't currently support 32-bit export. So at this time, Reason has to be the last piece of software in the workflow if working with 32-bit float files and wanting to maintain them as 32-bit float files before exporting them as 24-bit from Reason...

I used a Zoom F6 on a film, recording in both 32-bit float and 24-bit, with scenes where there was whispering and shouting and when raising the the low levels of 32-bit float, it was certainly cleaner than raising the level of the 24-bit files.

[avasopht]

Interesting bit of gear. This may be ignorance on my part, but if you record an audio signal at super low level and then massively crank up the gain of the recorded track when you’re mixing, you’re still going to bring up the noise floor too, right? As in unwanted background ambient noise, hiss and chatter from within the audio circuit itself etc. 32bit doesn’t change that, does it?

I get how setting the input gain just high enough to get a good clean signal but not so high as to risk clipping is easier with the dynamic range of 32bit vs 24 or 16. I’m just not sure it’s true to imply that gain staging doesn’t really matter at all when, at very low input levels, I’d have thought it still matters a lot. This may be down to my misunderstanding, though…

32-bit float helps with low levels as well. It's really awesome It's like magic.

https://www.sounddevices.com/low-signal-32-bit-float/

https://www.sounddevices.com/sample-32- ... wav-files/

Reason supports 32-bit float import in Reason, but just doesn't currently support 32-bit export. So at this time, Reason has to be the last piece of software in the workflow if working with 32-bit float files and wanting to maintain them as 32-bit float files before exporting them as 24-bit from Reason...

I used a Zoom F6 on a film, recording in both 32-bit float and 24-bit, with scenes where there was whispering and shouting and when raising the the low levels of 32-bit float, it was certainly cleaner than raising the level of the 24-bit files.

24-bit PCM gives you a 144dB dynamic range, and 32-bit PCM gives you a 192 dB dynamic range.

The noise floor of the Zoom UAC-232 is -129dbV, which means that even with 24 bits, you can represent everything up to +0 dBV (or +48dBV with 32-bit PCM) with 15dB headroom.

Now, ... ... ... the maximum input level for the UAC-232 is 21.8dbV for TRS (9.3dBV with hi-Z), and 3.8dBV for XLR, which means the floating point adds nothing over 32-bit PCM in terms of the ability to represent the signal (beyond the extra 8-bits of precision offered by 32-bit PCM).

DISCLAIMER ... there might be issues with db conversion ...

[bxbrkrz]

32-bit PCM gives you a 192 dB dynamic range.

How about 32 bit Float?

[avasopht]

32-bit PCM gives you a 192 dB dynamic range.

How about 32 bit Float?

Floats give extremely high dynamic range ... but with 23 bits of precision.

This basically means if, for whatever reason, a signal is biased by a few volts, you will actually hear bit-crushing effects with a float where the 32-bit PCM format would sound exactly the same and you could just subtract the bias.

At the same time, 32-bit floats won't offer any gains. Though I guess it does make things simpler as you can make any value equal 0dBV (like 1.0).

[RobC]

ASR didn't have much good to say about the F6

https://www.audiosciencereview.com/foru ... iew.15668/

[avasopht]

ASR didn't have much good to say about the F6

https://www.audiosciencereview.com/foru ... iew.15668/

Well, he does mention you could be seeing USB noise while giving no details about his setup.

But his results are the sort of results you should expect anyway, as you can never escape the fact you're navigating a pretty narrow dynamic range anyway.

Still, this is great. Propellerhead were well ahead of their time as they were already doing this with balance (but the gain difference might have been much smaller).

For recording, a knob is still useful for direct monitoring

[RobC]

ASR didn't have much good to say about the F6

https://www.audiosciencereview.com/foru ... iew.15668/

Well, he does mention you could be seeing USB noise while giving no details about his setup.

But his results are the sort of results you should expect anyway, as you can never escape the fact you're navigating a pretty narrow dynamic range anyway.

Still, this is great. Propellerhead were well ahead of their time as they were already doing this with balance (but the gain difference might have been much smaller).

For recording, a knob is still useful for direct monitoring

As far as I know, he uses some really expensive devices for measuring.

To be honest, I expected way better results, given the 32 bit support and all.

[joeyluck]

Yeah that guy has no idea what he is talking about.

[joeyluck]

Here's more info about 32-bit float:
https://www.sounddevices.com/32-bit-flo ... explained/

[RobC]

Yeah that guy has no idea what he is talking about.

Well, he has quite the background:

https://www.audiosciencereview.com/foru ... rm.2/about

[selig]

Yeah that guy has no idea what he is talking about.

To be fair, if you go to the last few pages of that thread you will see there was an update on the zoom that addressed one of the issues he found. So I’m not sure he has NO idea what he is talking about.
Maybe you could clarify that comment with some additional context?

[joeyluck]

Yeah that guy has no idea what he is talking about.

Well, he has quite the background:

https://www.audiosciencereview.com/foru ... rm.2/about

What is he even doing with it? ASIO on Windows? That's a rocky start... Can't take that seriously. Is he even using it as a recorder? Is he trying to use it as an interface on Windows? Sounds like he has a bad, noisy USB port on his computer.

[bxbrkrz]

Here's more info about 32-bit float:
https://www.sounddevices.com/32-bit-flo ... explained/

Yep.

[QVprod]

The lack of input control scares me a bit. Sure it’s unlikely you clip, but either it’s really good at setting auto levels or your gain stage work afterwards may be pretty tedious.

[avasopht]

As far as I know, he uses some really expensive devices for measuring.

He does. But he said that the USB could be responsible for a lot of the noise and was not able to eliminate its impact (so there is more work to do).

He has stated clearly (as a professional should) the limitations of his test.

That doesn't mean his test is bad. He is just communicating the scope of his experiments.

And these findings are important to know, but he is warning us that you may yield different results without a USB connection, or with a more stable and clean USB connection.

[joeyluck]

As far as I know, he uses some really expensive devices for measuring.

He does. But he said that the USB could be responsible for a lot of the noise and was not able to eliminate its impact (so there is more work to do).

He has stated clearly (as a professional should) the limitations of his test.

That doesn't mean his test is bad. He is just communicating the scope of his experiments.

And these findings are important to know, but he is warning us that you may yield different results without a USB connection, or with a more stable and clean USB connection.

No somebody else had to point him to that suggestion that it was his USB connection.

We had a post here in the forum with someone experiencing a mysterious hum in Reason. Our community helped get to the bottom of that and that user was very acceptable of the answer. If it was this guy reporting, he'd declare Reason has a bad hum issue and show you pictures of the hum frequencies and tell you that his findings don't lie...

I'm not a fan of vague reporting. I'm the kind of person who wants to see the setup, I want to know what you are trying to do, hear the examples, see a video...

Anyways, this thread is about the UAC-232 Not about a 32-bit float recorder not being used as a recorder or in 32-bit float ‍♂️

[RobC]

As far as I know, he uses some really expensive devices for measuring.

He does. But he said that the USB could be responsible for a lot of the noise and was not able to eliminate its impact (so there is more work to do).

He has stated clearly (as a professional should) the limitations of his test.

That doesn't mean his test is bad. He is just communicating the scope of his experiments.

And these findings are important to know, but he is warning us that you may yield different results without a USB connection, or with a more stable and clean USB connection.

Guys, I can't really add too much useful info, since I suck at the electrical engineering part of music.

Honestly, I saw this post, and kind of hoped, 'what if this might be better to get for just my microphone and recording, than a Focusrite Scarlett, where I don't need the DAC, Line Out, Instrument In, and HP Amp (except direct monitoring) at all'.

All I could do, was looking the specs up, but I didn't find not even for the F6, what the dynamic range and SNR is. Let alone this new device, that the topic is about.

Still, ASR most of the time has good posts, hence I trust them.

It would be easier if manufacturers would just share all the specs, so that we can match our microphones.

(And of course, I DO get what the original purpose of these particular devices are.)