Upcoming 32-bit float Audio Interfaces in 2024

[Tranasaurus Reks]

Does anyone have any rumors or inside information on any companies releasing a 32-bit float Audio Interface anytime in the next year? I understand the "no clipping" is more gimicky than necessary in the majority of use cases, however it would definitely benefit the industry to adopt 32-bit float as a new standard (as long as it doesn't take too much resources to add onto an upcoming product/idea). As far as I understand, one of the ways 32-bit float is able to salvage clipped recordings is due to a combination of factors like a high dynamic range, so obviously software updates can't fix the missing hardware like multiple Audio Digital Converters.

Funnily enough, the Zoom UAC-232 has a dynamic range of 136 dB, which is the same as the Neumann MT 48 and the Merging Anubis (the audio interface the MT 48 is based upon), as well as many other audio interfaces greater than 120 dB like the Focusrite Red16 Line, Antelope Audio Goliath HD Gen 3, Merging Horus, Merging Hapi, Zaxcom Deva 24, Zaxcom ZMT4, Zaxcom ZMT4-X, etc.

Zaxcom claims to have introduced something extremely similar back in 2010 called Neverclip, where they used dual Analog Digital Converters exactly like what is happening right now, and in 2017 their Zaxcom ZMT4 was released with the Neverclip Technology. The Zoom F3 Field Recorder released in ‎February 18, 2022. The Zoom UAC-232 Audio Interface released in ‎February 16, 2023. Most recently, The Rode NT1 5th Generation released in February 20, 2023 with dual outputs USB-C and XLR and a self-contained 32-bit float mechanism in their PCB transmitted through the USB-C port.

So seemingly the technology has existed for a while, with the Rode NT1 5th Gen managing to integrate it into their own microphone, and can definitely trickle down into affordable options (The Zoom F3 is $399, the Zoom UAC-232 is $179, and the Rode NT1 5th Gen is $249). However there are flaws with many of the current products like the self noise and monitoring problems of the Zoom UAC-232, and the Rode NT1 5th Gen sounding different than the 4th Gen with a downgrade in microphone build quality like it's weird body. With these as some of our only current options for 32-bit float, it's probably best to wait for something in the future. I thought that the new Focusrite 2i2 and 4i4 4th Gen released back in August 31, 2023 would have 32 bit float, but alas it's more waiting.

I don't know if 32-bit float technology is currently patented or is able to be be patented (perhaps Rode and Zaxcom has something), but if anyone knows if any Audio Interface manufactures like Focusrite, Solid State Logic, Universal Audio, etc. is planning on releasing any upcoming 32-bit float Audio Interfaces soon, please do share. And if any representatives are reading this, 32-bit float is definitely a welcome addition worth pursing in a product (as long as it does't take too much resources of course).

 

[ccaudle]

as long as it does't take too much resources of course)

The way 32 bit is implemented in hardware involves multiple A/D converters (probably two, but in principle could be three) and DSP to fade between the different converters at different signal levels. The analog front end also has to be a little more complicated, because you need a way to make sure the higher gain converter doesn't get saturated for too long when it clips, and doesn't crosstalk or disturb the other lower gain converter in some way.
Stage Tec had a patent on their gain ranging converter design, that was out in at least 2010 so should be expired in another five or six years I would think. Someone would have to look up the patent to confirm.

 

[Gold]

As I understand it 32 bit float is a data format only. There is no such thing as a 32 bit float converter with 1500dB dynamic range in the analog world. The practical limits of conversion are still about 20 bit.

The idea behind using multiple AD’s is to keep a full 24 bit resolution even at low signal levels.

Anyone who can’t figure out how to avoid clipping with 144dB dynamic range should probably look for a new hobby.

They could be 32 bit fixed converters. They would offer little to no practical advantage over 24bit converters.

 

[Tranasaurus Reks]

The way 32 bit is implemented in hardware involves multiple A/D converters (probably two, but in principle could be three) and DSP to fade between the different converters at different signal levels. The analog front end also has to be a little more complicated, because you need a way to make sure the higher gain converter doesn't get saturated for too long when it clips, and doesn't crosstalk or disturb the other lower gain converter in some way.
Stage Tec had a patent on their gain ranging converter design, that was out in at least 2010 so should be expired in another five or six years I would think. Someone would have to look up the patent to confirm.

I found two patents cited by Stage Tec US9654134B2 and US4999628. But I don't think these are the patents you described, since they date back to 1996. Nevertheless, I found a Gearspace forum with a comment by David Rick on 16th September 2021 describing "the expiration of a Stage Tec patent describing how to switch between the two A/D streams in the digital domain depending on the signal level."

 

[Thomas49]

The current range of the portable mixers/recorders by "Sound Devices" operates at 32 bit.
The "Mixpre" series can record 32 bit float on SD card and as USB audio interface.
The "8-series/Scorpio" combines 32 bit AD-conversion with 64bit internal processing and 64bit wav recording.

I just quoted the data of the latter from the manufacturers website. I frequently use their 833 mixer for german public tv stations, but we do not record 32bit. The "official workflow" keeps us at 24bit, which is fine. however, we even keep the (pre AD analog) limiters switched in.

 

[cyrano]

Anyone who can’t figure out how to avoid clipping with 144dB dynamic range should probably look for a new hobby.

I agree with that in a studio setting.

Once you think about all these places where the sound guy has absolutely no control on [accidental] sound levels, like outside interviews, nature recording, or even just news gathering, 32 bit float is a godsend.

BTW, there was an Italian 32 bit interface years ago. Can't remember the brand, but it was very, very expensive. The analog part was suited for mic and line level, just set by digital gain. Imagine a mic preamp with that kind of headroom...

 

[emrr]

You can always still clip the analog receiver in front of the converter. The 1st gen Zoom recorders for instance don't have that much headroom, so a loud source with many hot condensers could still clip it at unit gain....or a hot line source, which they claim to take. YMMV.

 

[cyrano]

I found two patents cited by Stage Tec US9654134B2 and US4999628. But I don't think these are the patents you described, since they date back to 1996. Nevertheless, I found a Gearspace forum with a comment by David Rick on 16th September 2021 describing "the expiration of a Stage Tec patent describing how to switch between the two A/D streams in the digital domain depending on the signal level."

A while back I came across a representative of Sound Devices. I asked him about their patent and how it was different from previous designs. His answer didn't make much sense. It seems everybody is using the same approach: send low level to one AD, high levels to another and combine the data. So that puts patents in another light.

 

[Brian Roth]

Does anyone sell an A to D with two or more analog input circuits...each operating at different gains...feeding into a "smart" digi convertor, which somehow "cross fades" between the two analog inputs? Beyond my skill set after asking that question!

Bri

 

[OneRoomStudio]

It seems like the biggest problem is that most engineers still insist on hitting the converters with hot levels (regardless of headroom available). Avoiding clipping is easy with 24 bit converters if you track with 0 VU = -18 dBFS (or even easier with 0 VU = -24 dBFS), assuming you keep your average levels around 0 VU. The problem is that everyone still wants to "bury the needle," and still worries about noise floor (even though it's vanishingly low in modern studios). Just track with lower levels and you'll have all the headroom you need.

 

[Brian Roth]

It seems like the biggest problem is that most engineers still insist on hitting the converters with hot levels (regardless of headroom available). Avoiding clipping is easy with 24 bit converters if you track with 0 VU = -18 dBFS (or even easier with 0 VU = -24 dBFS), assuming you keep your average levels around 0 VU. The problem is that everyone still wants to "bury the needle," and still worries about noise floor (even though it's vanishingly low in modern studios). Just track with lower levels and you'll have all the headroom you need.

Indeed! I recall the "Bad Old Daze" when I argued with the session engineer re. operating levels. At 16 bits, I argued for "desk 0VU at -12 dBFS. He insisted for -16 dBFS! egads

Bri

 

[Gold]

Indeed! I recall the "Bad Old Daze" when I argued with the session engineer re. operating levels. At 16 bits, I argued for "desk 0VU at -12 dBFS. He insisted for -16 dBFS! egads

Bri

The engineer was arguing for 16dB of headroom. You were arguing for 12dB headroom. I use -16dBfs = 0VU in my setup. With 24 bit digital if you expect to be recording cannon shots 144dB dynamic range should still be sufficient. Just lower where the 0 is if it needs to look good on the meters.

 

[Gold]

Does anyone sell an A to D with two or more analog input circuits...each operating at different gains...feeding into a "smart" digi convertor, which somehow "cross fades" between the two analog inputs? Beyond my skill set after asking that question!

Bri

That’s how the best converters have always worked. The Prism AD2 is over 30 years old and that’s how it was designed. Still one of the best. Single input though. It takes care of that for you.

 

[Brian Roth]

The engineer was arguing for 16dB of headroom. You were arguing for 12dB headroom. I use -16dBfs = 0VU in my setup. With 24 bit digital if you expect to be recording cannon shots 144dB dynamic range should still be sufficient. Just lower where the 0 is if it needs to look good on the meters.

Sorry...I was thinking "upside down" in my memory/comments. Negative numbers! I was merely recalling how he wanted to "slam" the convertor while I suggested a lower operating level.

Bri

 

[Gold]

Sorry...I was thinking "upside down" in my memory/comments. Negative numbers! I was merely recalling how he wanted to "slam" the convertor while I suggested a lower operating level.

Bri

Well, you can smash the converter no matter where the blinky red light is. 12dB headroom doesn’t seem out of bounds. That was the analog EBU standard. Makes sense with a +24vdc rail.

 

[Tubetec]

There seems to be this way of recording music nowadays where if something has out lying peaks well above the average , you simply slam those peaks into digital zero and hope for the best .
Who thought people that bullshit ? where did they learn sound engineering , in the computer studies class ?

 

[Brian Roth]

<g> At least with analog tape there is a "soft overload".

Bri

 

[k brown]

I need 32 float for all those times I record gunshots without being in the room to watch levels . . .

 

[gyraf]

I hear ya', but over here in Europe we rarely have shootings in recording studios

 

[zamproject]

12dB headroom doesn’t seem out of bounds. That was the analog EBU standard

12dB headroom over Vu is burning hot ! the 12dB over PPM, which is probably the EBU standard you are referring? is a comfortable margin