R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1] 2  All   Go Down

Author Topic: Pro Tools HD10 and Mastering at 32 Bit Floating Point  (Read 8085 times)

Alécio Costa - Brazil

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 89
  • Real Full Name: Alécio T. Costa
Pro Tools HD10 and Mastering at 32 Bit Floating Point
« on: December 03, 2011, 09:22:07 am »

Hi,

I use Pro Tools HD for mastering. I know there are several softwares on the market that can handle 32 bit floating point word length.

1) Are there real advantages mastering at 32 bit floating point?

2) Also, at 32 , doing SRC, are we free from conversion errors like the one where ceilings at -0.2 may go up to -0.1 or 0 db Clipped?

I always open a blank session and do the 44k 24 to 44k 16 with master fader at -0.1, to guarantee no peaks at all will pass to the final product.
However, Soon I am changing to Sample Manager.


BTW... I recommed to my clients mixes with Peaks < -3dB and RMS < -19 for mastering.

Nice Week-end!
Logged
Alécio Costa Studio
www.aleciocosta.com

ggidluck

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 70
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #1 on: December 06, 2011, 05:48:59 pm »


Hi Alecio,

1) Are there real advantages mastering at 32 bit floating point?

I have heard that PT10 does not dither the output of each plugin when using floating point. Overall, floating point may have less cumulative rounding error than the integer math.

2) Also, at 32 , doing SRC, are we free from conversion errors like the one where ceilings at -0.2 may go up to -0.1 or 0 db Clipped?

Yes, but you have to consider what will happen when going out to a DAC or in the final rendered file. But inside the DAW you are safe.

Logged

Alécio Costa - Brazil

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 89
  • Real Full Name: Alécio T. Costa
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #2 on: December 17, 2011, 10:01:03 pm »

Thanks for yours reply!
Anyone else has comments and additional info?
Logged
Alécio Costa Studio
www.aleciocosta.com

Jerry Tubb

  • Full Member
  • ***
  • Offline Offline
  • Posts: 172
  • Real Full Name: Jerry Tubb
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #3 on: December 18, 2011, 11:47:53 pm »

I'm staying with PT9 for awhile.

When PT 10 is able to handle 64-bit processing for the few plug-ins I occasionally use, I'll consider the leap.

Cheers, JT
Logged
Terra Nova Mastering
Celebrating 25 years of Mastering!

Matt_G

  • Newbie
  • *
  • Offline Offline
  • Posts: 39
  • Real Full Name: Matthew Gray
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #4 on: December 19, 2011, 08:59:06 am »

I'm staying with PT9 for awhile.

When PT 10 is able to handle 64-bit processing for the few plug-ins I occasionally use, I'll consider the leap.

Cheers, JT

Hey Jerry, can you define '64-bit processing' in your above statement? If you're talking about waiting until PT's is a 64bit application then ok but If you're talking about 64bit plug-in processing we've had that for many years. There is a lot of double precision 64bit float plug-ins that work fine in PT's 6 thu PT's 10. The plug-in interconnects still come back to 32bit float but I think you'll be hard pressed to hear any rounding errors going from 64bit float down to 32bit float. However the summing mixer is 64bit float when running PT 10 HD in 'Native' mode which is how I'm using it.
     
Logged

Jerry Tubb

  • Full Member
  • ***
  • Offline Offline
  • Posts: 172
  • Real Full Name: Jerry Tubb
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #5 on: December 19, 2011, 11:32:09 am »

Hey Jerry, can you define '64-bit processing' in your above statement? If you're talking about waiting until PT's is a 64bit application then ok but If you're talking about 64bit plug-in processing we've had that for many years. There is a lot of double precision 64bit float plug-ins that work fine in PT's 6 thu PT's 10. The plug-in interconnects still come back to 32bit float but I think you'll be hard pressed to hear any rounding errors going from 64bit float down to 32bit float. However the summing mixer is 64bit float when running PT 10 HD in 'Native' mode which is how I'm using it.
     

OK Mr. Cutting Edge,

 I'll take your word on that, Matt.

The way I understand it there are several levels involved.

OSX running in 64-bit mode.

Pro Tools running in 64-bit mode.

Plug-Ins running with 64-bit processing capability.

Summing Buss at 64-bit mode.

Isn't there supposed to be a " leap" in quality when all of this occurs?

Or is that all just Marketing Bling?

Or do we have to wait on Quantum Computing to catch up? : - )

The improved summing mixer sounds like a good thing, can you hear a difference?

At the moment you're the only ME I know that's using PT 10.

Also, rather than just criticize my statement, can you address Alecio's original questions? ;-)

Cheers, JT
Logged
Terra Nova Mastering
Celebrating 25 years of Mastering!

Alécio Costa - Brazil

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 89
  • Real Full Name: Alécio T. Costa
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #6 on: December 22, 2011, 04:19:14 pm »

 :'(
Logged
Alécio Costa Studio
www.aleciocosta.com

Jerry Tubb

  • Full Member
  • ***
  • Offline Offline
  • Posts: 172
  • Real Full Name: Jerry Tubb
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #7 on: December 22, 2011, 08:43:42 pm »

:'(

Fear not Alecio!

The big leap with Pro Tools for me was version 9,

when it was liberated from Avid/Digi hardware.

I'm running PT 9.0.5 on a Lynx AES-16e card.

On my Mac Pro with OSX 10.6.8. Works fine.

While I'm looking forward to moving to PT 10,

I'm gonna wait for awhile.

Merry Christmas!

JT
Logged
Terra Nova Mastering
Celebrating 25 years of Mastering!

Alécio Costa - Brazil

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 89
  • Real Full Name: Alécio T. Costa
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #8 on: December 22, 2011, 09:03:57 pm »

Thanks , my friend!
BTW.. a piece of cake to you... I have just turned 42 today! LOL
And here reading forums!

Tomorrow I will celebrate ( Friday)

:)
Logged
Alécio Costa Studio
www.aleciocosta.com

Cass Anawaty

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 61
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #9 on: December 22, 2011, 09:38:03 pm »

I would think the main advantage would be compliance with the other DAWs out there and their file formats.   I wouldn't expect a sonic improvement.
Logged
Cass Anawaty, Mastering Engineer
www.sunbreakmusic.com

Jerry Tubb

  • Full Member
  • ***
  • Offline Offline
  • Posts: 172
  • Real Full Name: Jerry Tubb
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #10 on: December 23, 2011, 05:25:59 am »

Thanks , my friend!
BTW.. a piece of cake to you... I have just turned 42 today! LOL
And here reading forums!

Tomorrow I will celebrate ( Friday)

:)

Thanks, that's some good cake bro'!

Have a great birthday, take the day off and enjoy this summer weather.

Iirc 42 is some sorta mystical number.

Best, JT
Logged
Terra Nova Mastering
Celebrating 25 years of Mastering!

Matt_G

  • Newbie
  • *
  • Offline Offline
  • Posts: 39
  • Real Full Name: Matthew Gray
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #11 on: December 23, 2011, 11:49:04 pm »

OK Mr. Cutting Edge,

 I'll take your word on that, Matt.

The way I understand it there are several levels involved.

OSX running in 64-bit mode.

Yes Snow Leopard & Lion are both capable of running in 32bit mode or 64bit mode. SL's default state is 32bit, Lion's default state is 64bit. 64bit mode allows the OS to access more than 4GB of RAM along with optimised 64bit extensions = a faster system.

Quote
Pro Tools running in 64-bit mode.

At the moment Pro Tools is not a 64bit application so it can't run in 64bit mode. So if you're running Lion or SL in 64bit mode Pro Tools will still only be running as a 32bit application. I'm sure Pro Tools 11 will be a 64bit application but this will break the support for current PT's HD hardware. Rather than dump heaps of money to go to the new HDX systems I decided to go completely PT's HD 'Native'. There is no Avid hardware in my new computer at all. Just the iLok2 with a license to use PT's 10 HD Native with the Lynx AES16e I/O. The Lynx cards have 64bit drivers which allows them to be use in Lion X64.

Quote
Plug-Ins running with 64-bit processing capability.

Summing Buss at 64-bit mode.

Yes the Plug-In buss in PT's 10 Native has 32bit interconnects but the internal processing of the plug-ins can operate up to 64bit floating point but on the output of the plug-in it must come back to 32bit float to hand off to either the next plug-in insert or into the 64bit summing mixer.   

Quote
Isn't there supposed to be a " leap" in quality when all of this occurs?

I wouldn't say a "leap" as such but it's simplified & streamlined things. Because it's all floating point there is no dither required until the audio needs to exit the digital domain into a D/A converter at which point it is truncated to 24bit integer. So any master outputs, aux channels going to physical outputs or hardware inserts need 24bit dither theoretically. However there is now less points that need dithering because internally it's all being processed at 32bits or higher and because the summing mixer has such large headroom/precision with it's 64bit processing it makes for a slight sonic improvement over an older HD system imo. It also means there is less latency as the audio no longer needs to be processed between the DSP cards & the CPU. So everything just feels faster, more fluid and with the disk caching my current native system feels significantly faster than my previous HD system & I would say it sounds a little better too.           

As for the OP's questions, converting the wordlength or recording something to 32bit floating point isn't going to improve the 'sonic quality' compared to a 24bit integer file. 24bit is still the maximum wordlength that is supported by the AES standard. So you're always going to be listening to a 24bit representation of a 32bit file anyway. Programs like Pro Tools native, Logic, Cubase etc. all use at least 32bit or higher for internal processing so working in a 24bit session is the same as working in a 32bit session as the minute you alter anything such as the gain, EQ, compression etc. inside ProTools the wordlength is going to increase beyond 24bits anyway. On the master output it still needs to come back to 24bit anyway. So 32bit & 64bit don't really mean much in terms of quality as such but it means there is greater internal headroom to prevent clipping & to prevent the need for internal processing dither until the final output.

Regarding question 2. Yes it is possible to store all of the intersample peaks from an SRC process at 32bit float & it won't be clipped. However you will still need some way of bringing these overs back below 0.0dBFS when converting down to 16bit (or 24bit). When I was working at 96kHz for final limiting I used to re-limit the ISP's with a -0.3dBFS threshold & a -0.3dBFS ceiling @ 44.1kHz along with 16bit dither when converting from 32bit float to 16bit integer.

Hope this helps.. Merry Christmas too all :)     
Logged

Alécio Costa - Brazil

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 89
  • Real Full Name: Alécio T. Costa
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #12 on: December 24, 2011, 12:05:25 am »

Ok, friends!
Thanks!
Nice Xmas and 2012!
Logged
Alécio Costa Studio
www.aleciocosta.com

Jerry Tubb

  • Full Member
  • ***
  • Offline Offline
  • Posts: 172
  • Real Full Name: Jerry Tubb
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #13 on: December 24, 2011, 05:07:08 am »

Matthew,

Nice job taking the time to research, clarify and summarize!

There's nothing like actually using the new version, to give one the needed impetus to investigate it.

I know that the info is on Avid's site and on their DUC, but with the plethora of info and mis-info on the net, distilling the truth can be a task.

I like the idea of going HD Native, which (I assume) eliminates the need for all the expensive Avid DSP boards, & is just smart IMO, considering the on-board power of today's CPUs. Especially in a mastering context, where in my case, would only be using the Pro Tools machine to Pitch to soundBlade HD through my analog EQ chain.

So hat's off to you, for being on the cutting edge, and thanks for the update.

Cheers, JT


Yes Snow Leopard & Lion are both capable of running in 32bit mode or 64bit mode. SL's default state is 32bit, Lion's default state is 64bit. 64bit mode allows the OS to access more than 4GB of RAM along with optimised 64bit extensions = a faster system.

At the moment Pro Tools is not a 64bit application so it can't run in 64bit mode. So if you're running Lion or SL in 64bit mode Pro Tools will still only be running as a 32bit application. I'm sure Pro Tools 11 will be a 64bit application but this will break the support for current PT's HD hardware. Rather than dump heaps of money to go to the new HDX systems I decided to go completely PT's HD 'Native'. There is no Avid hardware in my new computer at all. Just the iLok2 with a license to use PT's 10 HD Native with the Lynx AES16e I/O. The Lynx cards have 64bit drivers which allows them to be use in Lion X64.

Yes the Plug-In buss in PT's 10 Native has 32bit interconnects but the internal processing of the plug-ins can operate up to 64bit floating point but on the output of the plug-in it must come back to 32bit float to hand off to either the next plug-in insert or into the 64bit summing mixer.   

I wouldn't say a "leap" as such but it's simplified & streamlined things. Because it's all floating point there is no dither required until the audio needs to exit the digital domain into a D/A converter at which point it is truncated to 24bit integer. So any master outputs, aux channels going to physical outputs or hardware inserts need 24bit dither theoretically. However there is now less points that need dithering because internally it's all being processed at 32bits or higher and because the summing mixer has such large headroom/precision with it's 64bit processing it makes for a slight sonic improvement over an older HD system imo. It also means there is less latency as the audio no longer needs to be processed between the DSP cards & the CPU. So everything just feels faster, more fluid and with the disk caching my current native system feels significantly faster than my previous HD system & I would say it sounds a little better too.           

As for the OP's questions, converting the wordlength or recording something to 32bit floating point isn't going to improve the 'sonic quality' compared to a 24bit integer file. 24bit is still the maximum wordlength that is supported by the AES standard. So you're always going to be listening to a 24bit representation of a 32bit file anyway. Programs like Pro Tools native, Logic, Cubase etc. all use at least 32bit or higher for internal processing so working in a 24bit session is the same as working in a 32bit session as the minute you alter anything such as the gain, EQ, compression etc. inside ProTools the wordlength is going to increase beyond 24bits anyway. On the master output it still needs to come back to 24bit anyway. So 32bit & 64bit don't really mean much in terms of quality as such but it means there is greater internal headroom to prevent clipping & to prevent the need for internal processing dither until the final output.

Regarding question 2. Yes it is possible to store all of the intersample peaks from an SRC process at 32bit float & it won't be clipped. However you will still need some way of bringing these overs back below 0.0dBFS when converting down to 16bit (or 24bit). When I was working at 96kHz for final limiting I used to re-limit the ISP's with a -0.3dBFS threshold & a -0.3dBFS ceiling @ 44.1kHz along with 16bit dither when converting from 32bit float to 16bit integer.

Hope this helps.. Merry Christmas too all :)     
Logged
Terra Nova Mastering
Celebrating 25 years of Mastering!

Cass Anawaty

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 61
Re: Pro Tools HD10 and Mastering at 32 Bit Floating Point
« Reply #14 on: December 24, 2011, 12:32:45 pm »

There are compelling arguments against 32 bit float for audio.  Moorer's article is one--don't have a link, but I'm sure it's "floating" around out there.  Worth a look.
Logged
Cass Anawaty, Mastering Engineer
www.sunbreakmusic.com
Pages: [1] 2  All   Go Up