Topic: FLAC 1.14 - "WARNING: Legacy WAVE file has format type 1 but bits-per-sample=24"

[Brian Skalinder]

Dunno if anyone else has encountered the same message, but I noticed a new warning when compressing my 24-bit files in FLAC v1.1.4:

WARNING: Legacy WAVE file has format type 1 but bits-per-sample=24

From my brief poking around, it seems this is simply a warning indicating that the WAVE file uses the legacy WAVE format WAVEFORMATEX that did not explicitly allow for 24-bit audio (even though most apps/devices wrote - and still write - 24-bit WAVE files using this format), intead of the new WAVEFORMATEXTENSIBLE.  AFAIK, one may ignore this warning with no consequences.

[JasonSobel]

Brian -
that came around with FLAC v1.1.3 (and continues with v1.1.4, as you've noted)

it's discussed a bit in this thread:
http://taperssection.com/index.php/topic,76692.0.html

basically, it's has to do with the two different file formats used for 24-bit audio.  the "old" standard (the one that just about every program uses), and the newer "Microsoft" standard. 
in CD Wave, if you check the "alternate format" box, you get the older standard.  if you leave it unchecked, you get the "microsoft" standard

FLAC (starting with v1.1.3) will compress either file format, and will result in the exact same FLAC file.
if the 24 bit file is in the "microsoft" format to start, you won't see any warning message.
but if the 24 bit file is in the "old" format (much more common), you'll get that warning.

either WAV format that you use to start will result in the same FLAC file, and FLAC will decode to the files back to the "old" standard, no matter which WAV format was used to begin with.

I verified all this stuff with FLAC Fingerprints and wav md5's and stuff.  it's all in that other thread that is linked to above.

am I rambling, or am I clear enough?

[Brian Skalinder]

am I rambling, or am I clear enough?

Very clear, thanks.  I must've skipped 1.13 and gone from 1.1.2 to 1.1.4 or something, because I don't recall seeing the msg before.  Didn't even think to search for it since I didn't recall seeing a recent thread about it.  Thanks for the heads up.

[terrapinj]

thanks for the info Jason

i just switched to 1.14 this weekend and tested it on a recent 24bit show - i just realized i didn't click on the "alternate 24bit format" box yet FLAC had no issues encoding, good to know the reason now 

[spreadheadtom]

- i just realized i didn't click on the "alternate 24bit format" box yet FLAC had no issues encoding, good to know the reason now 

I never had problems ENcoding, just DEcoding.  might want to check it before you delete them of the HD.

 Tommy

[it-goes-to-eleven]

I never had problems ENcoding, just DEcoding.  might want to check it before you delete them of the HD.

FWIW, the *ONLY* way to really test a flac file is to completely extract the WAV and then compare that wav against the original wav.  I only use the command line version of flac and the 'flac -t' can definitely screw up and incorrectly tell you everything is okay.

The biggest problems are when the wav header has issues like not correctly reflecting the actual wav size.

[terrapinj]

I never had problems ENcoding, just DEcoding.  might want to check it before you delete them of the HD.

FWIW, the *ONLY* way to really test a flac file is to completely extract the WAV and then compare that wav against the original wav.  I only use the command line version of flac and the 'flac -t' can definitely screw up and incorrectly tell you everything is okay.

The biggest problems are when the wav header has issues like not correctly reflecting the actual wav size.

so each time you encode to FLAC you decode every file back to WAV and compare checksums with each original WAV file?

[it-goes-to-eleven]

so each time you encode to FLAC you decode every file back to WAV and compare checksums with each original WAV file?

As a vendor once repeatedly told me in regard to their product features, anything can be scripted

I always retain my original wav masters and don't consider flac a substitute for archival purposes.  So that makes most of my use of flac not critical enough to warrant the compares. Though at 24/96, my masters rarely fit on a single dvd without flac.

When creating backup DVDs (not true archive), I make some checks. I don't do the exact compare because I was concerned about cases where the flac extracted wave header might not be exactly the same (some headers have serial numbers, etc).  That would fail a simple checksum comparison test.  It would be pretty straightforward to compare just the wav data separately from the header but I was in too much of a hurry to start burning discs and start catching up with off-siting.

Currently, I use the -t test option with flac and check the return code. Then the flac metadata is read to get the archived wav size. That size is compared to the actual original wav data size (minus the header).  If any of those tests fail, the flac is discarded and the original wav is used instead.  Once burned, the original files are compared via rsync checksum to everything on the dvd.

[Brennan]

Brian -
that came around with FLAC v1.1.3 (and continues with v1.1.4, as you've noted)

it's discussed a bit in this thread:
http://taperssection.com/index.php/topic,76692.0.html

basically, it's has to do with the two different file formats used for 24-bit audio.  the "old" standard (the one that just about every program uses), and the newer "Microsoft" standard. 
in CD Wave, if you check the "alternate format" box, you get the older standard.  if you leave it unchecked, you get the "microsoft" standard

FLAC (starting with v1.1.3) will compress either file format, and will result in the exact same FLAC file.
if the 24 bit file is in the "microsoft" format to start, you won't see any warning message.
but if the 24 bit file is in the "old" format (much more common), you'll get that warning.

either WAV format that you use to start will result in the same FLAC file, and FLAC will decode to the files back to the "old" standard, no matter which WAV format was used to begin with.

I verified all this stuff with FLAC Fingerprints and wav md5's and stuff.  it's all in that other thread that is linked to above.

am I rambling, or am I clear enough?

+T good explanation, I understood it (well, except for the checksum parts, I've never really read up on using the fingerprints all that much..)

Just to be sure, DVDs use 24bit audio, correct?