Reason 32bit floating point recording

[DNGmaestro]

I’m thinking about buying the Rode NT1 5th generation, that can record at 32bit floating point, which is awesome for voice overs and people that have a lot of dynamics in their singing. I’m just not sure if reason supports 32bit recording. Logic still doesn’t.

Does someone have experience with 32bit recording on Reason or is it not possible?

Thank you

[stillifegaijin]

I'd also like to know if anyone has an answer to this. I know Reason can import 32 bit float audio, but I'm still not sure if it can record it. I'm very interested in that Rode mic too.

[selig]

It MIGHT work since it’s coming into Reason via USB and Reason’s internal path is 32 bit.
But remember, the dynamic range of 24 bit covers from dead silent to well over the threshold of pain. Not sure any voice needs THAT much dynamic range.
AND, since 24 bit interfaces gives you such a wide range (well over 100 dB, often 120 dB or more) before you get into the noise floor, you can easily record ANY signal with peaks 10-20 dB or more BELOW clipping and still capture the entire dynamic range (and then some) of the original source.
That said the human voice can range from 50 dBSPL for soft conversation to 85-90 dBSPL for loud yelling (taking an average of the available data). So 40 dB dynamic range, maybe 50/60 dB in the extreme (if you count whispers to screams).

Even a 16 bit recording with 96 dB dynamic range can faithfully capture a human voice with room to spare…
Main takeaway: there is much gear that exceeds the needs of the source - always consider the source!

[joeyluck]

I was just looking at this and trying to figure out where the USB connection is.

Then I looked at the XLR connection and zoomed in... clever design.

[joeyluck]

My only experience with 32-bit float recording is with field recorders, in the field capturing environmental sounds and on films. I haven't tried to record directly into Reason or done any 32-bit recording onto a computer.

I think 32-bit float is most useful in the field with much more unexpected sound, but it's still nice to have in any situation. People didn't have many objections to Reason's approach with their clip-safe feature on Balance and understand the need. And people have accepted for years the idea of back-up recordings that capture sound at a lower level. 32-bit float is just so much better and flexible.

Even when I do record people in a controlled environment and do a level check, asking them give their quietest and loudest, some people aren't very consistent with that. Many first takes that follow a level check, I end up discovering that they will in fact be louder or quieter than they were during the check lol. What I like about 32-bit float in those situations is that we can just go...the first take which could be amazing, doesn't have to be a level check or a discovery of needing to adjust levels for the following takes or adjusting during takes for a wide range of dynamics (or doing separate takes for those big differences in dynamics). So now I just tell people that we'll do a rehearsal take and I'll capture it just in case, and that take is always usable.

If looking into recording 32-bit float onto a computer, I would also consider the Zoom UAC-32, because then you could use an NT1 or any mic with a 32-bit float interface.

[raveled]

On the subject of 32-bit floating point recording:

[Murf]

I think the idea is that the microphone itself uses the whole 32 bit range, with 24 bit in reason and a normal microphone you are not getting the whole dynamic range of 24 bit, rather just a 'window" within it.
The Rode Youtube on the Rode NT1 5th generation explains it very well.
The selling point is that you dont have to do a soundcheck first to set your levels, it can all be done in post.

[avasopht]

You can do this with two preamps 36-42dB apart, ...

When the signal for the lower gain preamp is below 0dBFS, the floating point value will be produced by the lower gain preamp, otherwise the value will be produced by the higher gain preamp.

Floating points consist of the signal (known as the mantissa, and has 24 bits of precision with a 32-bit float) and a scale.

This means the hardware can effortlessly switch between the preamp-source on a frame-by-frame basis while merely adding the fixed scale factors for whichever preamp the signal was sourced from in that frame.

Now, the microphone would never use the full 32-bit range (since that also includes the positive exponents).

This can all be simulated in software by applying the same scheme to generated tones (if someone wanted to confirm the logic).

[Mataya]

What about capturing signal in 32bit for later processing, like time stretching and pitch shifting etc? Would such recording benefit from it?.

tx
M

[RobC]

It MIGHT work since it’s coming into Reason via USB and Reason’s internal path is 32 bit.
But remember, the dynamic range of 24 bit covers from dead silent to well over the threshold of pain. Not sure any voice needs THAT much dynamic range.
AND, since 24 bit interfaces gives you such a wide range (well over 100 dB, often 120 dB or more) before you get into the noise floor, you can easily record ANY signal with peaks 10-20 dB or more BELOW clipping and still capture the entire dynamic range (and then some) of the original source.
That said the human voice can range from 50 dBSPL for soft conversation to 85-90 dBSPL for loud yelling (taking an average of the available data). So 40 dB dynamic range, maybe 50/60 dB in the extreme (if you count whispers to screams).

Even a 16 bit recording with 96 dB dynamic range can faithfully capture a human voice with room to spare…
Main takeaway: there is much gear that exceeds the needs of the source - always consider the source!

NT1s, USB-free, had 90 dB SNR; new ones are said to be the same; but bhphotovideo says some specs are better now. Maybe it's 96 dB SNR now. So the noise floor is covered by 16 bit for sure; max SPL is 142 dBSPL, which doesn't even cover the dynamic range of 24 bits. What's the point of 32 bits? And the mic price is about the same if not cheaper.
Do manufacturers give away built-in, micro sized recording systems with 1528 dB dynamic range for free? : ) Kidding!

By all means, I'm still 100% for the NT1, it's a reachable kind of holy grail, and really hope that bhphotovideo just misprinted something, cause I might grab yet another, hopefully still matching NT1 one day.

[jlgrimes]

It MIGHT work since it’s coming into Reason via USB and Reason’s internal path is 32 bit.
But remember, the dynamic range of 24 bit covers from dead silent to well over the threshold of pain. Not sure any voice needs THAT much dynamic range.
AND, since 24 bit interfaces gives you such a wide range (well over 100 dB, often 120 dB or more) before you get into the noise floor, you can easily record ANY signal with peaks 10-20 dB or more BELOW clipping and still capture the entire dynamic range (and then some) of the original source.
That said the human voice can range from 50 dBSPL for soft conversation to 85-90 dBSPL for loud yelling (taking an average of the available data). So 40 dB dynamic range, maybe 50/60 dB in the extreme (if you count whispers to screams).

Even a 16 bit recording with 96 dB dynamic range can faithfully capture a human voice with room to spare…
Main takeaway: there is much gear that exceeds the needs of the source - always consider the source!

I agree. Recording between -20 to -10 db is pretty much what I do. The main purpose of 24 bit recording was to be able to record at lower levels to not have to worry about clipping and yet still be well above the digital noise floor. As with 16 bit recording, it was possible to when recording weak signals to introduce noise (very similar to using tape).

Even if I stray a bit above -10 db or below -20 db, no spilled milk or detriments to the audio fidelity.

I'm all for new technology but from a practical standpoint, we haven't even made a perfect 24 bit ad/da converter yet as the dynamic range is still about 20 to 21 bits at best (with many budget 24 bit ad/da converters about 16 to 17 bits from a dynamic range perspective). Even if it was possible to create a perfect 24 bit recorder, the dynamic range would be "dangerous", as you would get to the threshold of pain. So as you are saying it is even possible to get great recordings with just 16 bits of dynamic range (as many people in the past did) or even lower. As after doing tests even 12 bit audio can be surprisingly good from a noise perspective.

Regarding 16 bit recording, in 2009, I bought a Yamaha Audiogram3 while vacationing in NYC pretty much just to buy something. Several years later, I got a good opportunity to use it (to record a vocalist at her house). I was following the -20 to -10 db below full scale principle and the audio was sounding good and clean except for the fact that I could hear what sounded like "tape noise" in my headphone monitors (still at a pretty low level but audible when listening via headphones). I couldn't wrap my brain around what was going on. I later found out that the Yamaha Audiogram3 was only a 16 bit recorder (which I wasn't expecting as I thought by 2009 16 recorders were no longer made anymore). But that (and a Xiph .org YouTube video) gave me a good practical lesson on how bit rate affects your signal. I used to record with 16 bit adats all the time in the 90s early 00s, but forgot that I was recording vocals with compression and almost right before clipping (which was a common practice then) to stay above the noise floor which is why I never really heard any noise in the old 16 bit days (and that I probably didn't listen to music with headphones much then).

So mainly 24 bit recording is practically the "limit" as far as having all the dynamic range you need to make great recordings.

[selig]

16 bit recording: @ nearly twice the dynamic range of 24 track analog machines I never once heard noise coming from the recording. That said, I did little ADAT recording (less than with DA88 format), but the ‘big’ 16 bit machines I worked with had very low noise.
Just like Fairlight CMI only had 8 bit sampling, yet matched/beat the sound of 12 bit samplers like Akai because of the build quality of the convertors, the pro level 16 bit recorders of the 80-90s had very low noise compared to the far cheaper ADAT machines.
My guess is the ‘support’ electronics have as much or more affect on things like noise floor than the actual recording/storage format.
Meaning, yes you cannot have MORE than 96 dB dynamic range with a 16 bit recording, but you CAN have far less than 96 dB DR with a 16 bit recording. It’s not the bit depth alone that determines noise levels.
Bottom line - you must compare ALL specs, not just one (bit depth in this case) to know what to expect from any audio system. And even then, there are qualities that may cause you to prefer the end effect of a ‘lesser’ quality device, such as microphones with more noise but better overall sonics than others.

[joeyluck]

Curious if anybody has tried recording 32-bit float into Reason, since that is the topic of this thread.

We have other threads discussing 32-bit float and the benefits. The OP here is asking if it's possible to record 32-bit float into Reason. Reason can load 32-bit float files, but can it record in 32-bit float?

[jlgrimes]

16 bit recording: @ nearly twice the dynamic range of 24 track analog machines I never once heard noise coming from the recording. That said, I did little ADAT recording (less than with DA88 format), but the ‘big’ 16 bit machines I worked with had very low noise.
Just like Fairlight CMI only had 8 bit sampling, yet matched/beat the sound of 12 bit samplers like Akai because of the build quality of the convertors, the pro level 16 bit recorders of the 80-90s had very low noise compared to the far cheaper ADAT machines.
My guess is the ‘support’ electronics have as much or more affect on things like noise floor than the actual recording/storage format.
Meaning, yes you cannot have MORE than 96 dB dynamic range with a 16 bit recording, but you CAN have far less than 96 dB DR with a 16 bit recording. It’s not the bit depth alone that determines noise levels.
Bottom line - you must compare ALL specs, not just one (bit depth in this case) to know what to expect from any audio system. And even then, there are qualities that may cause you to prefer the end effect of a ‘lesser’ quality device, such as microphones with more noise but better overall sonics than others.

The AudioGram sounded pretty good. It was mainly just the relatively low but quite audible noise when listening to headphones. I didn't really hear it much when listening via the MacBook Air speakers. I'm thinking it was mainly just a couple of things:

1. I recorded her voice with no compression and probably closer to the -20 db full scale point (peaking), so her RMS level was most likely alot lower. I rarely recorded any vocal like that with Adats that quiet. Back then I pretty much recorded stuff almost near clipping and with compression (definitely for vocals).

2. Back when I used Adats I rarely monitored with studio headphones so any introduced noise, I probably never heard. And coming from Tascam Portastudio cassette 4 tracks which had an even much higher noise level, I probably would have still been amazed at the quality difference which was a huge jump for me so that level of noise I probably would have ignored.

[avasopht]

16 bit recording: @ nearly twice the dynamic range of 24 track analog machines I never once heard noise coming from the recording. That said, I did little ADAT recording (less than with DA88 format), but the ‘big’ 16 bit machines I worked with had very low noise.
Just like Fairlight CMI only had 8 bit sampling, yet matched/beat the sound of 12 bit samplers like Akai because of the build quality of the convertors, the pro level 16 bit recorders of the 80-90s had very low noise compared to the far cheaper ADAT machines.
My guess is the ‘support’ electronics have as much or more affect on things like noise floor than the actual recording/storage format.
Meaning, yes you cannot have MORE than 96 dB dynamic range with a 16 bit recording, but you CAN have far less than 96 dB DR with a 16 bit recording. It’s not the bit depth alone that determines noise levels.
Bottom line - you must compare ALL specs, not just one (bit depth in this case) to know what to expect from any audio system. And even then, there are qualities that may cause you to prefer the end effect of a ‘lesser’ quality device, such as microphones with more noise but better overall sonics than others.

Bear in mind that this 32bit floating point recording does not have (or need) gain control.

That's its strength and weakness (and probably why it ranks low on THD measurements).

With gain control, sure, 16-bit (or even 15-bit) is more than enough. The 8-bit and 11-bit formats were usually non-linear, so they won't produce as much noise as 8-bit or 11-bit PCM.

The AudioGram sounded pretty good. It was mainly just the relatively low but quite audible noise when listening to headphones. I didn't really hear it much when listening via the MacBook Air speakers. I'm thinking it was mainly just a couple of things:

The Yamaha Audiogram 3 had terrible noise and I had to send it back immediately. Maybe I had a bad batch, but I couldn't believe Yamaha could produce something so shoddy.

Plus its drivers were basic (IIRC it didn't even have its own ASIO drivers and had to rely on things like ASIO4All).

[joeyluck]

Zoom announced the UAC-832

https://m.facebook.com/story.php?story_ ... tid=YvzCfH

FB_IMG_1682354634548.jpg (56.29 KiB) Viewed 1352 times

[stillifegaijin]

Curious if anybody has tried recording 32-bit float into Reason, since that is the topic of this thread.

We have other threads discussing 32-bit float and the benefits. The OP here is asking if it's possible to record 32-bit float into Reason. Reason can load 32-bit float files, but can it record in 32-bit float?

Thank you!! Yes, the topic is can it be done? I'm not particularly interested in the debates over it's merits or more explanation of what it is, I understand that. But has anyone tried recording 32bit float into Reason?

[jlgrimes]

The AudioGram sounded pretty good. It was mainly just the relatively low but quite audible noise when listening to headphones. I didn't really hear it much when listening via the MacBook Air speakers. I'm thinking it was mainly just a couple of things:

The Yamaha Audiogram 3 had terrible noise and I had to send it back immediately. Maybe I had a bad batch, but I couldn't believe Yamaha could produce something so shoddy.

Plus its drivers were basic (IIRC it didn't even have its own ASIO drivers and had to rely on things like ASIO4All).
[/quote]


The noise was pretty high for a "modern" audio interface. I ended up reading the specs one day and the thing that caught me was that it was only 16 bit. That said for playback of audio already in the PC it sounded good. But what I remember was when monitoring the signal before recording I didn't hear any noise but after hearing the recorded audio on playback an audible noise floor was present mainly when listening on headphones where I can hear those type of issues more easily than speakers. I just wrote it off as being only a 16 bit device and recording at low levels. It probably would have sounded ok if I knew better beforehand and used higher gain on the mic preamp. But nowadays something like a Scarlett 3rd gen can do these remote recordings easily without worry much about setting preamp gain.