Benefit of recording 32-bit audio?

If it was just 32 then yes, but when it’s float? More headroom, less errors?

Doesn’t the plugin have to covert the file to 32 then back to 24 bits? I’m a little confused about this.

Same rounding errors, which are principally controlled by the bit depth, which is 24.

Float offers less chance of catastrophic errors, that is overload, due to the 8 bit signed exponent handling the scaling.

This is what they say on the Pro Tools website (don’t shoot me lol).

"So having your 24 bit recordings in 32 bit floating point format will not change the quality of the initial recordings themselves, but creating audio files in this format before they are processed by plug-ins will help you avoid the following:

Clipping during AudioSuite rendering
Unnecessary noise introduced by AudioSuite dithering
Rounding errors during signal processing

These issues then are mostly caused by the fact that with either 16 or 24 bit audio the data requires conversion at the point of processing."

http://www.askaudiomag.com/articles/pro-tools-10-the-benefits-of-32-bit-floating-point-audio

So 24 bit files are up sampled to 32 and then back to 24 in a 32 bit plugin? That is where rounding errors etc are caused? Doesn’t it effect the file when you do an audio mixdown?

As already discussed.

Possibly because the dithering will be avoided.

If signal processing is any more complicated than simple addition or subtraction, there is no simple generic explanation. Signal processing may involve conversion between time and frequency domains and sample rate changes, which are non-trivial in execution and explanation.

That is, conversion from input ADCs and to output DACs.

Upsampling refers to sample rate changes, which may be done in a plugin (most UAD plugins internally upsample to 192k for their calculations, and downsample at their output).

Remembering that 32 bit float and 24 bit integer have the same value bit depth, then the more the 32 bit values are above 0dB, the more least significant bits are lost when converting to 24 bit.

Does the Cubase manual also say, or otherwise Steinberg endorse, turning HT off ?

I don’t want to derail the OP concern, but just a quick comment on HT- this is probably one of the most unclear areas in Cubase’s design and operation for the general public, despite the minimal incremental feedback provided on ASIO Guard when implemented in C7. IMO, the way you presented your argument on HT above is not technically accurate, as there is no rearrangement of events between cores. What there may be, is an interruption of the RT flow due to other threads competing for execution, which can translate in drop-outs if the ASIO buffer is not filled timely as a result. But anyway, carry on.

IIRC an old Cubase manual, 32 bit recording was enabled to allow for Broadcast Wave format. But otherwise makes no difference to the sound of the file.

In fact, it’s a common myth that 24 bit “sounds better” than 16 bit. 24 bit simply allows for more headroom so engineers can record incoming signals at lower levels and still have a nice signal-to-noise ratio. Previously, with older 16-bit methods, you’d have to ride the signal pretty high and risk “digital clipping.” I point this out, as this prevailing myth is likely what lead the OP to ask his question in the first place.

Cheers.

Please don’t consider it derailing my OP. The expansion of the conversation is quite informative.

This is the absolutely most common myth when it comes to Digital Audio. Equally frustrating every time this topic shows up.

My background on the topic is among things like, sitting in with Tomas Lund and TC. Electronic in seminar about Digital Audio, Converters, Bit Rates and Sample Rates, Limiting and the Loudness War etc. In addition, I have a formal education in sound technical engineering (mid to late 90’s) and thousands of pages read on the topic (Lavry, Aldrich +++).

The hardest part is to tell this as simple as possible, so I will try a go on the Bit Rate and Dithering part first:

Higher bit depth gives greater precision in the amplitude domain = WRONG. 1st big misunderstanding across the internet.
As easy explained as possible:

0 dBFS in 24 bit = 0dBFS in 16 bit, No difference there.

1 bit = 6dB of Dynamic range.

16 bit x 6 = 96 dB of dynamic range (from 0 dBFS to -96 dBFS)

24 bit x 6 = 144 dB of Dynamic range (from 0 dBFS to -144 dBFS)

Try to think from the top and downwards. You have to think from the top and downwards.

The signal recorded in the top 96 dB range is exactly the same.

At -96 dBFS is were the difference comes in to play:

When your signal drops below -96 dB, the 16 bit will chop off the signal (the last bit will shut down).
The 24 bit signal has 48 dB’s of dynamic range left (for reverb-tail and other low level signals).

Here is also when Dithering comes into play:

When reducing the bitdepth to 16 bit from 24 bit, the dithering process will add some low level noise in the range that our hearing is least sensitive.
This is so to be able to keep the 16th bit (from top down) “open” as long as possible, to keep as much of reverb-tails and other low level details as possible (explained in the esiest way i could).

Conclusion:
If your signal has a Dynamic range of anything less than 96 dB’s and is located in the top 96 dB range, the recorded signal will be exactly the same.

The biggest misunderstanding is that the 24 bit scale has lesser “space” between measuring points (i.e.better). This is simply not true.
1 bit has only two values, on or off (1 or 0). Do the math and learn to think from the top and downwards.

That was the BIT part.
2x2x2x2x2… should be such an easy math. Don’t make it harder than it is.

Back to the 32-bit part: As all audio and real time processing (calculations - it’s just math) is executed in the 32-bit floating point realm, it doesn’t matter what Bit Size your source files are.

The only advantage of recording 32- bit files comes to play if you are doing a LOT of offline processing (post recording). That will make sure that rounding errors, after the offline processing/calculations, are “added” to the files at a level way way below human hearing.

My challenge to people who are gonna discuss Digital Audio (without wasting time, beleiving in Internet Myths), is to read all you can from reliable sources (google Nika Aldrich and Dan Lavry first). Listen to developers at big name Audio equipment manufacturers, not their marketing departement.

I will leave you with the best (and simplest) Digital Audio Myth breaking video there is (from bigger nerds than me):

Happy myth busting.

CappacinoKid wrote:
I read that 32 bit float reduces rounding errors caused by plugins and reduces noise from dithering?

:mrgreen: In my opinion the rounding errros are minimalised when upping the samplerate, 32 bit has more volume but the resolution is samplerate, so when working in 44.1K the rounding errors are the same, you just have more headroom.

Correct me if I’m wrong :mrgreen:

See the video in the link in my post above. Happy myth busting :wink:

Regarding the oft-related issue of sample rates, are you able to explain why Universal Audio upscales most of their UAD plugins to 192k for internal processing, then downscales back to the project rate.

There is no mention of it in marketing materials, and it is only mentioned in the Operation Manual in regard to latency considerations. Why would they do it if it was not necessary, especially as the extra processing undermines their UAD cards’ plugin number capacity and latency?

And what implications does it have for DAW users, who do a lot of complex processing to get their end result which may be just 44.1k, or even an MP3?

And just to be clear, this is a separate issue from what people can distinguish between as far as the end product goes, where the testing has involved only a single sample rate change and has shown people generally cannot distinguish.

Larger bit rates and floating point are only to help us DAW operators to be able to worry less about optimising recording levels or gain staging.

They are excess to end-user listening requirements.

192k can be beneficial for many kinds of non-linear processing - distortion modelling, compression etc. Having this up/down conversion builtin guarantees the quality of the conversion processes (to UADs standards), while leaving the engineer free to use whatever rate they want in the project.

http://www.youtube.com/watch?v=cIQ9IXSU > … e=youtu.be

Gone through the whole video and I almost bought it.

Fact is that more then 20 years ago the CD format was very often discussed and it became very clear that all CD players incorporated rounding error algoritmics to make playback of CD’s sound good (this is in fact told by software engineers who build these cd players), so in my book there is Always rounding algorithmics involved, the steps that have to be rounded become smaller as the samplerate increases.

The Sample Rate and Bit size is independent on each other.

The rounding errors are on the Bit side of things, not the Sample Rate of things. Just saying.

What iBM said… sample rate is a totally different topic. (but not without it’s own myths, upon which mentioning would likely throw this whole thread off track.)

We’re “stepping” into the sample rate myth I alluded to above. :confused:

Here’s the text version for the video (link below). Skip down to where the heading reads “192khz considered harmful.”

http://people.xiph.org/~xiphmont/demo/neil-young.html#toc_1ch

And another read:

http://www.trustmeimascientist.com/2013/02/04/the-science-of-sample-rates-when-higher-is-better-and-when-it-isnt/

If your A/D convertor in your interface is 24bit and as such can only deliver 24bit digital data, what is the use of padding the signal with nothing when storing it in your computer? Cubase already works internally on 32bit float, regardless of what the bit depth of your recordings is.
Do any of you know if there are interfaces available that convert at 32bit?
Jan

Stop distortion when mixing perhaps?