Quantcast
24 or 32 bit export? - Gearspace.com
The No.1 Website for Pro Audio
24 or 32 bit export?
Old 11th March 2014
  #1
Gear Nut
 
🎧 5 years
24 or 32 bit export?

Hi, I'm going to be submitting my EP tracks soon for professional mastering. I've mixed them in Cubase, all my projects are 44.1/24bit. When I go to export out my .WAV for mastering, Cubase defaults to an option called
32bit (float), should I render at this or change it to 24bit?

thx
Old 11th March 2014
  #2
Lives for gear
 
🎧 5 years
Hi!

I believe you should consult the ME that will take up the job. Intuitively, 32F is a more accurate format so I'd go for that, but basically both 32 and 24 bit resolutions are plenty redundant to the point of overkill, so it is a workflow issue really.
Old 11th March 2014
  #3
Gear Addict
 
brethes's Avatar
 
Verified Member
🎧 10 years
Quote:
Originally Posted by Jex2000 ➑️
Hi, I'm going to be submitting my EP tracks soon for professional mastering. I've mixed them in Cubase, all my projects are 44.1/24bit. When I go to export out my .WAV for mastering, Cubase defaults to an option called
32bit (float), should I render at this or change it to 24bit?

thx
24 bit is usually preferred (and more standard) but 32bit shouldn't be a problem for most mastering engineers, so just ask your ME first.
Old 11th March 2014
  #4
Lives for gear
 
karumba's Avatar
 
🎧 10 years
both is fine (but 24bit more typical). i case of 24bit just make sure, that your output level is not too low & you don't have any clipping.
Old 11th March 2014
  #5
Lives for gear
 
Red Mastering's Avatar
 
Verified Member
🎧 10 years
Quote:
Originally Posted by Jex2000 ➑️
Hi, I'm going to be submitting my EP tracks soon for professional mastering. I've mixed them in Cubase, all my projects are 44.1/24bit. When I go to export out my .WAV for mastering, Cubase defaults to an option called
32bit (float), should I render at this or change it to 24bit?

thx
if your project is 24/44.1, keep it as it is,
don't use dither, or/and upsample, downsample,
as Karumba mentioned, keep 24 bit files far from 0dB FS and you will be fine
I personally prefer 24 bit, as 32 bit are really large files, and they aren't superior to 24bit
Old 11th March 2014
  #6
Lives for gear
 
🎧 5 years
Like everyone said, its better to retain the same format as the project was generated.. And leave the encoding to be done at the mastering stage..

If you encode a 24bit file to 32.. it will just be a copy of the 24 bit file.. even if there is some benefits from the format, the audio will be the same..
Old 11th March 2014 | Show parent
  #7
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Umulamahri ➑️
Like everyone said, its better to retain the same format as the project was generated.. And leave the encoding to be done at the mastering stage..
Why?

Quote:
Originally Posted by Umulamahri ➑️
If you encode a 24bit file to 32.. it will just be a copy of the 24 bit file.. even if there is some benefits from the format, the audio will be the same..
Not if the original files were manipulated even in the slightest. As soon as you as much as touch a volume fader in your DAW, your output becomes 32 bit float.
Old 11th March 2014 | Show parent
  #8
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by Orson Maxwell ➑️
Not if the original files were manipulated even in the slightest. As soon as you as much as touch a volume fader in your DAW, your output becomes 32 bit float.
actually, you don't even need to touch a volume fader, as the internals of any DAW is 32bit FP (or 64bit FP). A 24/44.1 project only means that the raw recordings are made with these parameters. The mix engine of every (respectable) DAW is still 32bit FP even if you select a project with 16/44.1 ...

So, if you want an exact copy of what you hear during mixing, export with 32bit FP. 24bit is still very good, with very slightly lower resolution (it's questionable if the difference is audible) but headroom only to full scale - 32bit FP has a practical not reachable full scale.
Old 11th March 2014
  #9
Lives for gear
 
🎧 5 years
You are right, but there are some logical optimisations in the DAW code that can lead to some obscure irregularities in this axiom.

For example, watch this video starting from 2:55
https://www.youtube.com/watch?v=T05eCpobt4k#t=174

I know the guy speaks Russian, but you can just watch the bitscope on the bottom of the screen. The output switches to 32 bit only when there is at least something happening (a fader is moved or a fade out is used).
Old 12th March 2014
  #10
Lives for gear
 
Kindred's Avatar
 
🎧 10 years
Possibly a related question would be - considering that when it goes on sale (assuming digi only, no vinyl) it will need to be uploaded as 16/44, at which stage should a track be downsampled?

Recently I have switched my Cubase exports to 16/44 (used to be 32bit) as I was seeing no differences in the final mastered 16/44 wav. I am sure there are a range of arguments as to why this may be technically inferior, but I am seeing no benefits to my final wav (let alone the mp3 when it goes on sale on Beatport)
Old 12th March 2014
  #11
Lives for gear
 
🎧 5 years
The ten million dollar question is, can you hear the difference between a 24bit and 32fp.. I believe only in some playback systems and a trained ear when observing a rare track with unnatural amounts of dynamics.. So basically no.. I personally think the quality of a world clock is much easier to hear..

There is a big resolution difference between the formats, But as long as the workstation software is using modern encoding there really is little benefit .. If there is a striking difference in the sound stepping down to a 24bit from 32, most likely the conversion is the problem..

The difference in the size of the files is huge.. But please everyone should make there own judgment on this topic.. output bolth formats and blind test your self.. It has to be a true blind test or it will be a worthless test..

64fp software is becoming more popular..
Old 12th March 2014 | Show parent
  #12
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Kindred ➑️
Possibly a related question would be - considering that when it goes on sale (assuming digi only, no vinyl) it will need to be uploaded as 16/44, at which stage should a track be downsampled?

Recently I have switched my Cubase exports to 16/44 (used to be 32bit) as I was seeing no differences in the final mastered 16/44 wav. I am sure there are a range of arguments as to why this may be technically inferior, but I am seeing no benefits to my final wav (let alone the mp3 when it goes on sale on Beatport)

I can hear the difference between 24 and 16.. But some forms of music is harder to tell. and more natural sounding tracks its easier to detect..

But there is a reason cd format has been around for so long. It sounds great!! the modern loudness abuse of the format is a shame..
Old 12th March 2014 | Show parent
  #13
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by Orson Maxwell ➑️
You are right, but there are some logical optimisations in the DAW code that can lead to some obscure irregularities in this axiom.

For example, watch this video starting from 2:55
https://www.youtube.com/watch?v=T05eCpobt4k#t=174

I know the guy speaks Russian, but you can just watch the bitscope on the bottom of the screen. The output switches to 32 bit only when there is at least something happening (a fader is moved or a fade out is used).
well, some kind of processing will take place, otherwise what sense would it have to load a file into a DAW and export it with no processing at all. It's easier to copy the file ... :-)
The russian guy documents the fact that the conversion from 24bit to 32bit FP and then (with no processing) back to 24bit is bit transparent, meaning that no bits gets changed by (or lost in) the conversion.

So, any processing results into 32bit FP. If you export to 24bit and send the file to your mastering engineer, he will import it into his DAW, that also works with (at least) 32bit FP. Exporting to 32bit FP will preserve all the information exactly as you have mixed it, up to the DAW of your ME.

The argument of file size is also small: 1 minute 24bit takes 15MB while in 32bit FP it will take 20MB. I would not call that difference "huge" ...
I, for one, export to 32bit FP!
Old 12th March 2014
  #14
Lives for gear
 
🎧 5 years
Right..

The size adds up quickly when working on large projects. doesn't seem like much when working on small bit of audio..
Old 12th March 2014 | Show parent
  #15
Lives for gear
 
huejahfink's Avatar
 
Verified Member
🎧 10 years
Quote:
Originally Posted by gigi ➑️
The argument of file size is also small: 1 minute 24bit takes 15MB while in 32bit FP it will take 20MB. I would not call that difference "huge" ...
Yeah but that's cos you're talking about a few megabytes.

That's actually a 33.3333333333333333333333% increase. I think many people would consider that a significant increase in anything.
Old 12th March 2014
  #16
Gear Nut
 
🎧 10 years
what large projects are you talking about?
I'm talking about exporting the final mixes for the ME.
A record with 10 songs, each 3 minutes long, takes 450MB in 24bit and 600MB in 32bit FP = for me still not "huge" or "significant" in the days of multi terra byte hard drives and fast internet connections.
Old 12th March 2014 | Show parent
  #17
Lives for gear
 
FabienTDR's Avatar
 
Verified Member
🎧 5 years
Quote:
Originally Posted by gigi ➑️
It's easier to copy the file ... :-) The russian guy documents the fact that the conversion from 24bit to 32bit FP and then (with no processing) back to 24bit is bit transparent, meaning that no bits gets changed by (or lost in) the conversion
Maybe just a details, but it seems to me that the software is smart enough to notice that nothing needs to be processed and thus directly outputs the input file ("copy").

Technically, fixed to floating point conversion in IEEE 754 (the numeric standard for most software) can be lossy and will be for the case of audio signals. The DAW in use takes this into account and obviously doesn't convert formats for "fun", which is exactly what I would recommend to the OP.
Old 12th March 2014 | Show parent
  #18
Gear Guru
 
UnderTow's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by FabienTDR ➑️
Maybe just a details, but it seems to me that the software is smart enough to notice that nothing needs to be processed and thus directly outputs the input file ("copy").

Technically, fixed to floating point conversion in IEEE 754 (the numeric standard for most software) can be lossy and will be for the case of audio signals. The DAW in use takes this into account and obviously doesn't convert formats for "fun", which is exactly what I would recommend to the OP.
It has nothing to do with being smart. When you convert 24 bit to 32 bit float, the exponent is just padded with zeros. That is why a bitscope won't show any activity on those extra 8 bits. It is just a constant stream of zeros until some processing is applied to the signal.

Alistair
Old 12th March 2014 | Show parent
  #19
Gear Guru
 
UnderTow's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by gigi ➑️
So, if you want an exact copy of what you hear during mixing, export with 32bit FP.
Actually the DAW converts to 24 bit before sending the signal to your interface so you are always hearing 24 bits.

Quote:
24bit is still very good, with very slightly lower resolution
It isn't lower resolution. Both formats have 24 bits of precision. 32 bit float just has more dynamic range due to the 8 bit exponent.

Anyway, 24 bit is fine for the final mix delivery as long as you don't clip the signal by going above 0 dB FS or deliver ridiculously low levels. (If your mix peaks anywhere between say -18 and 0 dB FS you will be fine).

Alistair
Old 12th March 2014
  #20
Gear Nut
 
🎧 10 years
I fully agree with your last sentence!

But no DAW is sending 24 bit to the interface. At least on Windows with ASIO it sends 32bit FP - that is what the driver expects. It's the driver that converts the stream to whatever the D/A can output (today usually 24 bit).

Regarding the lower resolution: think about a 24bit stream that peaks at -40dB - it is using only 16bit of resolution. A 32bit FP that peaks at -40dB still has all the 24 bit of resolution, as the exponent takes care of the -40dB of level.
Old 12th March 2014
  #21
Lives for gear
 
🎧 5 years
Well, we are not talking about listening to the difference - of course it is inaudible, but all the extra resolution is a safety measure when large amounts of processing is due downstream. But that is (hopefully) not the case when sending files to an ME.

I agree with Fabien - there are still 24 significant bits (its just the gaps between possible values that are rearranged in FP) and every single conversion is a loss of data. Surely I'm talking minuscule inaudible difference, but hey - we're engineers - our brains must fill in places where our senses fail

Anyway, I store intermediate files (which will be processed later) in 32bit just to ease my conscience and a sense of logic.
Old 12th March 2014
  #22
Lives for gear
 
🎧 5 years
Of course the best thing to do is go digital, to preserve the recording in it's truest form. A/D convertors are really good in 2014!
Old 12th March 2014 | Show parent
  #23
Gear Guru
 
UnderTow's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by gigi ➑️
But no DAW is sending 24 bit to the interface. At least on Windows with ASIO it sends 32bit FP - that is what the driver expects. It's the driver that converts the stream to whatever the D/A can output (today usually 24 bit).
That is an academic difference wouldn't you say? You are still listening to 24 bits.

Quote:
Regarding the lower resolution: think about a 24bit stream that peaks at -40dB - it is using only 16bit of resolution. A 32bit FP that peaks at -40dB still has all the 24 bit of resolution, as the exponent takes care of the -40dB of level.
Yes in floating point the exponent scales the level to preserve the 24 bit precision of the mantissa as the the signal goes up or down in level. Absolutely correct. Still, in practical terms there is no real difference with real world mix exports. Anyone with a final mix peaking at -40 dB FS has bigger issues than the word length of the exported file.

As a general reminder and a reality check, PT TDM has 24 bit interconnects between all TDM plugins. That means that in a real world mix the signals are being truncated to 24 bit (hopefully after proper dithering by the plugins if they run at 48 bit) at dozens if not hundreds of places. This never stopped people making great sounding mixes on PT TDM rigs!

Alistair
Old 12th March 2014 | Show parent
  #24
Here for the gear
 
🎧 10 years
Quote:
Originally Posted by Orson Maxwell ➑️
You are right, but there are some logical optimisations in the DAW code that can lead to some obscure irregularities in this axiom.

For example, watch this video starting from 2:55
https://www.youtube.com/watch?v=T05eCpobt4k#t=174

I know the guy speaks Russian, .
turn on English subtitles
Old 12th March 2014 | Show parent
  #25
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by OldSonic ➑️
turn on English subtitles
Hey thanks! I know Russian - I just missed the subtitles.
Old 14th March 2014 | Show parent
  #26
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by gigi ➑️
what large projects are you talking about?
I'm talking about exporting the final mixes for the ME.
A record with 10 songs, each 3 minutes long, takes 450MB in 24bit and 600MB in 32bit FP = for me still not "huge" or "significant" in the days of multi terra byte hard drives and fast internet connections.
True, But by the end of the year it dose add up to something. I view it as wasted space because lack of benefits.. Hard drive space is cheap, But its not free..

Its great the daw is using this level of quality internally but there is no clear reason to output the format.. especially working with a modern loud recordings that have reduced dynamics..


There is more incentive to use higher sample rates to gain ultrasonics for play back systems that go beyond 20khz, But that is a different debate..
πŸ“ Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 4616 views: 612554
Avatar for smoke
smoke 7th May 2021
replies: 98 views: 37774
Avatar for dfghdhr
dfghdhr 2 weeks ago
Topic:
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump