The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
16bit or 24bit WAV Masters for streaming in 2018 and onward?
Old 4 weeks ago
  #1
Here for the gear
 
Dr.Bulbus's Avatar
16bit or 24bit WAV Masters for streaming in 2018 and onward?

I'm releasing an album soon through the digital distributor "Distrokid" and have the option to submit 24 bit lossless files at any sample rate.

Are we trusting the multiple streaming services bit depth conversion? Spotify, Apple etc. Or should I just stick to 16bit 44.1khz with dither and call it a day? (I'm recording at 24/48). Not interested in releasing multiple masters for different services, Distrokid doesn't allow that anyway. Looking for opinions...
Old 4 weeks ago
  #2
Lives for gear
 
SmoothTone's Avatar
 

Most streaming services will convert to a lossy format and the codec will handle any wordlength reduction internally. AAC and Ogg Vorbis are most common. 24bit is arguably a better source for lossy conversion, but the advantage over 16bit would be small, if not inaudible, in most listening scenarios.

It's the sample rate you need to be more concerned with. Some may stream at 48kHz, others at 44.1. So your safest bet is to have your ME do the SRC for best quality.

I would go with 24bit 44.1kHz if DistroKid accepts it. Some distributors will only accept 16/44.1.
Old 4 weeks ago
  #3
Lives for gear
 
Hermetech Mastering's Avatar
 

Verified Member
24 in this day and age
Old 4 weeks ago
  #4
Gear Maniac
 

24bit can easily be 'downgraded' to 16bit

16bit cannot be 'upgraded' to 24bit

So it makes sense to go with 24bit.
Old 4 weeks ago
  #5
Gear Head
 
a.l.e.'s Avatar
 

Start with the highest quality that you can have on entry and you will have the highest quality output.
Old 4 weeks ago
  #6
Here for the gear
 
Dr.Bulbus's Avatar
Quote:
Originally Posted by gearhead1 View Post
24bit can easily be 'downgraded' to 16bit

16bit cannot be 'upgraded' to 24bit

So it makes sense to go with 24bit.
I understand that. What I'm really asking is do we trust their conversion process?Would you say the same thing for sample rate?

Ive recorded at 48. Would you send 24bit 48?

My first thought would be to send them the best quality I can.
Old 4 weeks ago
  #7
Lives for gear
 
teebaum's Avatar
most aggrigatiors only accept 16bit/44.1kHz wav

if your 24bit allows, you take of course 24bit
Old 4 weeks ago
  #8
Lives for gear
 
SmoothTone's Avatar
 

Quote:
Originally Posted by Dr.Bulbus View Post
I understand that. What I'm really asking is do we trust their conversion process?Would you say the same thing for sample rate?

Ive recorded at 48. Would you send 24bit 48?

My first thought would be to send them the best quality I can.
As indicated above, the bit depth will be handled in the conversion to Ogg or AAC. I would NOT leave sample rate to chance. Sending 48kHz, if accepted, won't get you better quality if there's crappy sample rate conversion used on the streaming end.
Old 4 weeks ago
  #9
Here for the gear
 
Dr.Bulbus's Avatar
Quote:
Originally Posted by SmoothTone View Post
As indicated above, the bit depth will be handled in the conversion to Ogg or AAC. I would NOT leave sample rate to chance. Sending 48kHz, if accepted, won't get you better quality if there's crappy sample rate conversion used on the streaming end.
Thanks for the advice.
Old 4 weeks ago
  #10
Gear Addict
 

Does it mean every streaming platform will concert 48 kHz to 44 kHz ? If thats the case, yes its better to go for 44 kHz , 24 bits
Old 4 weeks ago
  #11
Here for the gear
 
Dr.Bulbus's Avatar
Now the next question is... If I stay in 24bit, do I need to dither?
Old 4 weeks ago
  #12
Gear Addict
 

Answer is no
Old 4 weeks ago
  #13
Lives for gear
 
SmoothTone's Avatar
 

Actually, the answer is more than likely yes. If you've done any processing then most DAWs will expand the wordlength to 32bit floating point (or greater) so the output will need to be dithered down to 24bit. You can test this yourself with the free Bitter plugin from Stillwell. Put it at the end of your processing chain and it will show you your bit depth. Just check that your DAW doesn't automatically dither on export to avoid double dithering. And avoid noise shaping when further processing is likely (such as lossy conversion or volume normalisation; in other words: it's likely). Flat TPDF is best.
Old 3 weeks ago
  #14
Gear Addict
 

If the session is 32 bits i understand he should dither to go to 24 bits. But i don't understand why he should use a dither if the session is already 24 bits ?
Old 3 weeks ago
  #15
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
Only a flat, totally unprocessed transfer would be 24 bits. Virtually all signal processing expnds the bit depth to 32 or 64 float.
Old 3 weeks ago
  #16
Gear Maniac
 
parisminzer's Avatar
 

This seems like an appropriate time to ask this question out loud since we’re on the 32bit-24 bit topic.

Let’s say I need to deliver 24 bit MFiT-HD Tracks Masters as well as individual CD WAV (which will also get used on DDP).

This is a standard production parts order for one of the labels I work with (they also require other things but I’ll keep this concise)

Mixes always come at 48/24.

(I have a standard practice of exporting the Masters from the Processing stage at 48/32 bit—> to SRC and dither to 44/16 in separate program. My thinking was always that the SRC’d file was 32 bit so why not export that way in the first place...never thought about the final clippers/limiters plugins changing the bit depth: further making this a sound decision.
Even if it’s a 44.1 file and I only need to dither, I still do the same thing, for consistency’s sake...but back to 48k...)

Here’s the point that relates to your response: To get things to Mastered for Itunes, and use that level also for the CD WAV and CD, I might have to tuck the volumes of my 32 bit export tracks here and there down some (as we do) to not run afoul of the AAC codec. Maybe I need to do a couple fade outs also
(Main point...I’m past the plugins part of things and doing simple object gain adjustments)

I do this in a separate ‘album arrangement’ session, and have been exporting these ‘MFiT ready’ tracks from here at 48-24 with no dither applied. The CD WAV’s go out at 48-32 as normal, and go to Izotope RX to do the rest.

Although I’m not using plugins, if I’m changing the object volume, this is causing calculations to be made, so I’m gathering here that I’m not currently doing this 24bit export step right. If I adjust anything in the session, then 24 bit dither needs to be applied. Is that correct in your estimation?

I’ve never received any complaints about anything I’ve sent out this way, but if I’m technically screwing up, I will amend this process.
Thx

Quote:
Originally Posted by SmoothTone View Post
Actually, the answer is more than likely yes. If you've done any processing then most DAWs will expand the wordlength to 32bit floating point (or greater) so the output will need to be dithered down to 24bit. You can test this yourself with the free Bitter plugin from Stillwell. Put it at the end of your processing chain and it will show you your bit depth. Just check that your DAW doesn't automatically dither on export to avoid double dithering. And avoid noise shaping when further processing is likely (such as lossy conversion or volume normalisation; in other words: it's likely). Flat TPDF is best.
Old 3 weeks ago
  #17
Lives for gear
 
SmoothTone's Avatar
 

Quote:
Originally Posted by parisminzer View Post
If I adjust anything in the session, then 24 bit dither needs to be applied. Is that correct in your estimation?
Yes.

Quote:
I’ve never received any complaints about anything I’ve sent out this way, but if I’m technically screwing up, I will amend this process.
Thx
It's not a huge crime. The difference, as I understand it, would be inaudible and below the noise floor of most listening scenarios, but can accumulate. I think there was a trend in the early 2000s to send stuff out undithered because some people preferred the sound of truncation distortion. I'm not aware of any lives ever having been lost because of bad dither practices.
Old 3 weeks ago
  #18
Lives for gear
 
blaugruen7's Avatar
being a pro saxophonist and hobby mixer i send my music in 24bit 48khz to distrokid. all is fine
Old 3 weeks ago
  #19
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
The thing to understand about dithering is that you are preventing chattering distortion that will start to sound crunchy after additional processing such as lossy compression or digital volume and tone controls. Dithering doesn't simply cover it up.
Old 3 weeks ago
  #20
Lives for gear
 

I never apply dither when going from 32 bit (floating point) to 24 bit (fixed point) because their resolution is the same (23 bits + sign). 32 bit is basically only needed for internal CPU processing.

It's not the same thing as going from 24 bit to 16 bit (both fixed point) where data is being truncated.
Old 3 weeks ago
  #21
Gear Addict
 

It seems like opinions can vary a lot from each other ! x)
Old 3 weeks ago
  #22
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
Data IS being truncated when converting from 32 float to 24 fixed. This is a common misunderstanding.
Old 3 weeks ago
  #23
Lives for gear
 
Hermetech Mastering's Avatar
 

Verified Member
Yes, I always dither going from 32float to 24 fixed. It sounds better.
Old 3 weeks ago
  #24
Lives for gear
 
SmoothTone's Avatar
 

Opinions vary but the facts don't:

Truncation occurs whenever wordlength is reduced.

Truncation creates distortion - it sounds nasty.

Dither prevents the distortion (it doesn't just mask it).

At 24bit, truncation distortion is inaudible. So is dither noise. This is where opinions differ: if you can't hear it, whether you dither or not makes no audible difference...

AS LONG AS NO FURTHER PROCESSING IS APPLIED!

Distortion accumulates with multiple truncations.

Cumulative truncation distortion has a higher amplitude than cumulative dither.

Dither completely removes truncation distortion and results in a nice even noise floor. It also preserves a little bit of dynamic range that otherwise gets lost in truncation.

In short: dither leaves the audio in better shape for further processing and has a less detrimental cumulative effect than truncation distortion. (There's always further processing in the digital delivery age).

So, while WE can't hear truncation distortion or dither at 24bit level, any downstream processors CAN. What do we think THEY handle better: low level hiss or low level distortion?
Old 3 weeks ago
  #25
Lives for gear
 

Quote:
Originally Posted by Bob Olhsson View Post
Data IS being truncated when converting from 32 float to 24 fixed. This is a common misunderstanding.
Yes, I see there is a common misunderstanding - let me try to clear it up.

"Floating point" numbers are like using a 12 inch ruler to measure something that's bigger than 12 inches. You can slide that ruler around to measure something that's 10 feet long but you only have precision in the local range of that 12 inch ruler. Anyone who's used a ruler to measure something bigger than 12 inches knows what I mean.

In our case that "12 inch ruler" is 24 bits of audio resolution. A 32 bit floating point number only has 24 bits of resolution - the remaining bits are exponent (that can "slide that ruler around"). When you convert a 32 bit float to a 24 bit fixed data type the decimal point (the "ruler") gets shifted over so you get those 24 bits of precision. There is no loss of resolution. It's 24 bits in either data type.


Quote:
Originally Posted by Hermetech Mastering View Post
Yes, I always dither going from 32float to 24 fixed. It sounds better.
You are being fooled by some placebo or processing side effect. The only thing you could possibly be hearing is the dither itself and that is unlikely. There is not a D/A on Earth that can reproduce the lowest bits of 24 bit audio. Physics (thermal molecular noise) prevents it.
Old 3 weeks ago
  #26
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
That floating point "ruler" moves with every single sample. The 24 bits is precision and not the same as fixed-point bit-depth. All I know is that the folks at Bell Labs have said that dither is necessary for floating-point to fixed conversions or the math is wrong and I'll believe them before I'll believe anybody else. I also hear the difference in some cases provided the speakers are really really flat.
Old 3 weeks ago
  #27
Lives for gear
 
SmoothTone's Avatar
 

From Bruno Putzeys:

Quote:
What floating point does is give you 24 bits of resolution regardless of the signal level. As soon as you do any processing, all samples below -6dBfs carry useful information below the 24-bit mark.

That's 25 bits, actually. The MSB in floating point is implicit so the 24 bit mantissa effectively gives you 25-bit resolution below the signal level. A 32 bit floating point signal always carries more information than a 24-bit signal (unless it is an unprocessed transfer from 24 bits of course). Dither is not wasted.
Old 3 weeks ago
  #28
Lives for gear
 
Hermetech Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by Nonlinear View Post
You are being fooled by some placebo or processing side effect.
Feel free to keep arguing against what my ears tell me in my room with my system, backed up by science in blind ABX tests. To me it’s most apparent as a loss of the sense of depth so try listening for that.

The fact is, not dithering from 32 float to 24 fixed is bad practice, sounds worse, and the error can accumulate down the line.

Last edited by Hermetech Mastering; 3 weeks ago at 10:19 AM..
Old 3 weeks ago
  #29
Lives for gear
it's silly to argue this when it's so easy to just do it the correct way.
Topic:
Post Reply

Welcome to the Gearslutz 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…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump