32 bit recording with Aaton CANTAR

[borjam]

44 minutes ago, Constantin said:

While you are making valid points I am curious why you think that 32-bit float of all things would negate the need for mixing? 

 

My post was rather poorly worded. I wrote on the iPad which is not exactly pleasant

 

When I said "mixing" I meant "using files from different sources in your workflow", not mixing as in audio  mixing. So well, it's obvious that if your other sources use 32 bit FP audio files it is more convenient to avoid conversions.

 

From that point of view, even if they use a single converter it makes sense to generate floating point files. It just simplifies your workflow.

 

28 minutes ago, Constantin said:

As I understand it the converters themselves don’t need dynamic range greater than 24 bits, as they basically shift the reference point around to accommodate louder sounds. So the range stays the same, but the 0dBfs moves around. 
Still the converters do need to be able to facilitate this

I always thought the Cantar did have dual ADCs. But Aaton has never been very open about their specs

 

Again, my poor writing I said "32 bit capable architecture" meaning dual converters with different gain or whatever. There are several approaches used by different manufacturers.

 

I wonder about that as well. One problem with how 32 bit floating point has been marketed is, it looks like they suggest a full 32 bit dynamic range which is insane. The truth is, the dual ADC converter will add enough dynamic range so that you can really relax about gain settings. But do not expect it to record at a nuclear blast zero and not clip!

 

A cost advantage is that you can implement the limiters in the digital domain.

 

Indeed, seems they are not very open about their specs. And actually I guess it would be pretty easy for Zaxcom with their neverclip architecture to implement floating point files. Just they do not like the marketing and irrational expectations about it at all.

 

I remember many years ago (1990 maybe) I attended a course on DSP programming based on the Motorola DSP56000. Being a fixed point processor you had to be careful with your calculations because, well, fixed point mathematically "clips" as well. The DSP96000 was different, based on floating point, so you could somewhat relax about that. Your data came from 24 bit converters at most of course.

 

 

[The Documentary Sound Guy]

10 minutes ago, borjam said:

I guess it would be pretty easy for Zaxcom with their neverclip architecture to implement floating point files. Just they do not like the marketing and irrational expectations about it at all.

I'm thoroughly in agreement about irrational expectations and all, but I feel like it's only a matter of time before I get asked to deliver 32-bit because post wants that workflow (or someone higher up has decided it's technically superior).  Once that happens, it will become necessary to have it, regardless of its technical merits.

I feel like that is primarily what this new "feature" for the X3 is really about.  If we are going to be asked to deliver 32-bit, our recorders might as well be able to supply it, whether it is technically advantageous or not.

I do wonder whether Zaxcom's strange mirroring algorithm would be able to keep up with the data rate ... I've heard bad things about trying to record at higher data rates (mainly, 96kHz), which I think is mainly due to asking the recorder to convert MARF to WAV on the fly.  But, I'm wandering off-topic...

[Dejan Ceko]

Response from Aaton support via email , as a manufacturer, they don't usually express themself on public forums .

 

Pascal Grillère : "To try to clarify, as it is in the announcement, the 32 bits float audio file format is now available in the audio settings. It only concern the recording file format, it doesn't change the way you are working on the Cantar. The use Aaton Cantar limiter to prevent audio signal overload get the best signal dynamic is still true. This is the state of the art for version 3.532.C12."

 

Cheers .