Jump to content

Recommended Posts

2 hours ago, ray-dude said:

Getting access to the 32 bit masters is a huge value add.

I guess I know too well how digital audio works to see any value in that 🙂

 

2 hours ago, tailspn said:

BWF (Broadcast WAV) is simply a WAV file with the ability to support metadata:

Looks to me that if the data is larger than 4 GB (which would be likely in case of surround version), then the player needs to support either the "link" chunk or the RF64 extension of the WAVE format: https://en.wikipedia.org/wiki/RF64. Anyone knows how widespread the support is?

Link to post
Share on other sites
4 minutes ago, danadam said:

I guess I know too well how digital audio works to see any value in that 🙂

 

An acoustic recording has virtually no content above 22 bits, The value of 32 bits is in the post processing digital manipulation to support math processes without truncation. Since I did not master this RR Beethoven, I've very limited knowledge of the post processing production processes involved. But I doubt they extend much beyond 24 bits. 

Link to post
Share on other sites
1 hour ago, tailspn said:

An acoustic recording has virtually no content above 22 bits,

Hm... considering noise floor in the studio or in the concert hall, I would expect much less.

1 hour ago, tailspn said:

The value of 32 bits is in the post processing digital manipulation to support math processes without truncation.

True. My comment was only about the value (or lack thereof) of a 32-bit file for the end user.

Link to post
Share on other sites

These days, I'm listening to a lot of 16fs 32 bit content (705/768kHz) as wav files.  When files get bigger than 4GB, I'm splitting them into multiple parts.  With players that support gapless playback (HQP, Roon, etc), this is no issue at all.  The parts play as intended, even with the 4GB limitation

 

With all this preprocessing, I appreciate being able to get content in as close to the mastering format as possible (less about information content, more about information quality after digital processing....basically the same reason the mastering system does processing in 32 bits)

ATT Fiber -> EdgeRouter X SFP -> Sonore opticalModule -> Taiko Audio Extreme -> Chord DAVE -> Voxativ 9.87 speakers w/ 4D drivers

Link to post
Share on other sites
On 2/16/2021 at 8:01 PM, tailspn said:

NativeDSD has decided to also offer the 32 bit BroadcastWAV version of the DXD, which I produced from the Soundmirror supplied 32 bit source, and have uploaded it to native's AWS delivery server. It will take a few days to appear on the site for selection, as the site is not currently programmed to offer multiple DXD deliverables.

Two-channel only, so far.

Kal Rubinson

Senior Contributing Editor, Stereophile

 

Link to post
Share on other sites
12 minutes ago, tailspn said:

It's produced, and on the AWS S3 delivery server. I'll check to see if NativeDSD intends to offer it for sale. 

 

Thanks

They may not want to set a precedent. 🙉

Kal Rubinson

Senior Contributing Editor, Stereophile

 

Link to post
Share on other sites
  • 2 weeks later...
On 2/21/2021 at 8:50 PM, Kal Rubinson said:

They may not want to set a precedent. 🙉

Well, as it turns out, there's no valid supported WAV file format structure (BWF or WAV) supporting 5 or 6 channel 32 bit 352.8KHz PCM. There is of course support for 24 bit FLAC of the same sample rate in either 5 or 6 (5.1) channel. It's no problem to make, or play (in JRiver at least) this 32 bit surround file, it just won't load into our AWS delivery servers. The format is unsupported, returning an error stating unsupported file header.

 

For those set upon playing this 32 bit 5 channel WAV file, I'll send you a direct Dropbox link if you purchase the same FR741 5 channel FLAC from NativeDSD.com. Just send me a mail at [email protected] Fair enough?

 

Thanks,

Tom 

Link to post
Share on other sites

The short answer to your first question of a 32 bit PCM being better than a 24 bit is no as a delivery product. As explained in this Sound Devices article:

 

https://www.sounddevices.com/32-bit-float-files-explained/

 

 

24 bit PCM fixed point supports a dynamic range of 144dB, where the best of acoustic recordings rarely exceed less than half that dynamic range. 32 bits float (actually 64 bit fixed point in the best Digital Audio Workstations (DAW)) is supported for post processing digital signal processing (DSP) math bit growth during processing. Upon post processing (channel mixing and balancing, and/or sweetening) completion, files are truncated down to 24 bit fixed point for delivery, or 32 bit floating point for archival and further processing.

 

32 bit content files are typically floating point, but can also be fixed point. Both 24 bit and 32 bit deliverable files are always fixed point, to be operable within a DAC.

 

Tom

 

Link to post
Share on other sites
On 3/4/2021 at 1:03 PM, tailspn said:

The short answer to your first question of a 32 bit PCM being better than a 24 bit is no as a delivery product. As explained in this Sound Devices article:

 

https://www.sounddevices.com/32-bit-float-files-explained/

 

 

24 bit PCM fixed point supports a dynamic range of 144dB, where the best of acoustic recordings rarely exceed less than half that dynamic range. 32 bits float (actually 64 bit fixed point in the best Digital Audio Workstations (DAW)) is supported for post processing digital signal processing (DSP) math bit growth during processing. Upon post processing (channel mixing and balancing, and/or sweetening) completion, files are truncated down to 24 bit fixed point for delivery, or 32 bit floating point for archival and further processing.

 

32 bit content files are typically floating point, but can also be fixed point. Both 24 bit and 32 bit deliverable files are always fixed point, to be operable within a DAC.

 

Tom

 

I read the article -- and as a user/developer of software that uses both RF64/BEXT and FP .wav files, the nuance of the convenience factor about FP files might be missed:  you don't have to worry about clipping.

 

The clipping issue is somewhat implied in the article, but my software that does *very* massive processing to undo the last-step compression used on consumer recordings, using default gains, it is almost impossible to guarantee no clipping on a signed integer file.  Decreasing the liklihood of clipping requires telling the program to use at least -10dB or more gain on output.   Doing things like that just seems ugly -- because clipping is still possible even with a 10dB output attenuation.  Using FP files eliminates the trouble, then DAW or whatever software can be used to bring the signal back down to +- 1 (0dB peak) again.   Warning:  SoX internal representation is 32bit integer, so cannot deal with larger signal FP files, but I'd expect modern DAWs to work okay.

 

Bottom line:  FP can solve a lot of potential headaches and even potential time wasters during production.   For distribution, FP isn't really necessary.   Sometimes, during production, when really busy, not needing to worry about signal peak limits might help decrease some small amount of stress.   Distributed recordings are already massaged to be at correct signal levels.

 

 

Link to post
Share on other sites
  • 4 weeks later...

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×
×
  • Create New...