Corrupt File Question

[Darryl R]

Spoon,

I've been converting FLACs to AIFF and I encountered a corrupted track. This is no big surprise, as these particular tracks were from a guy who "ripped" a few SACDs for me with workstations that capture and encode the DSD stream (as I understood his explanation). My question is this: how bad can a reported corruption? Is there a wide variation? I can listen to the source and destination files and they sound fine. Here's the error:

Error converting to Aiff, 'C:\Users\Darryl\Music\Bruckner, Anton\Symphony No. 8; Günter Wand, NDR Sinfonieorchester\02 Symphony No. 8 in C minor - II. Scherzo. Allegro moderato.flac' to 'D:\Aiff\Bruckner, Anton\Symphony No. 8; Günter Wand, NDR Sinfonieorchester\02 Symphony No. 8 in C minor - II. Scherzo. Allegro moderato.aif'
md5 did not match decoded data, file is corrupt.

[Darryl R]

The above should read "how bad can a reported corruption be" meaning, do they vary in severity?

Here's another:

Error converting to Aiff, 'C:\Users\Darryl\Music\Mahler, Gustav\Symphony No. 5; Georg Solti, Chicago Symphony\03 Symphony No. 5 in C-sharp minor - 3. Scherzo. Kräftig, nicht zu schnell.flac' to 'D:\Aiff\Mahler, Gustav\Symphony No. 5; Georg Solti, Chicago Symphony\03 Symphony No. 5 in C-sharp minor - 3. Scherzo. Kräftig, nicht zu schnell.aif'
Encountered 'FLAC__STREAM_DECODER_READ_FRAME' in FLAC File 'C:\Users\Darryl\Music\Mahler, Gustav\Symphony No. 5; Georg Solti, Chicago Symphony\03 Symphony No. 5 in C-sharp minor - 3. Scherzo. Kräftig, nicht zu schnell.flac'

[Spoon]

There is no way of knowing, it could be 1 sample or 10000 samples.

[Darryl R]

Spoon,

Much of this may be imbedded in third-party libraries, but can you elaborate at all on what is being summed and checked during a conversion (in my case FLAC to AIFF)?

Also, any idea what the 'FLAC__STREAM_DECODER_READ_FRAME' is about?

Thanks again.

[garym]

https://en.wikipedia.org/wiki/Md5sum
https://en.wikipedia.org/wiki/Cyclic_redundancy_check

[Darryl R]

Gary, thanks for the links, but my question is not as much the concept as what entities are summed, since the resulting files are very different in length by virtue of the unpacking of the FLAC.

Also, the 'FLAC__STREAM_DECODER_READ_FRAME' comes up several places when Google'd, but nothing I saw defined what it is. It is this error that aborts the conversion. The checksum errors go thru with no audible difference (so far).

[garym]

got it. (but don't know the answer). Spoon will have to chime in.

[Spoon]

The whole decoded audio is checked against a single stored md5 at the end. However on really corrupted files it would not be possible to decode the whole file.

[Darryl R]

Spoon, is the stored md5 you mention above provided by the source file? And if no conversion errors are reported by dBpoweramp, could I expect a consistent md5 to be carried forward in each converted output file? The sources of the comparison were not entirely clear. Thanks.

[Spoon]

The md5 is calculated from the source audio.

[Darryl R]

Do you know what's going on with this error?

...
Encountered 'FLAC__STREAM_DECODER_READ_FRAME' in FLAC File ...

[Spoon]

It means the flac file is so corrupted it cannot be decoded.

[Darryl R]

Spoon,

I'm doing SACD conversion with a very smart guy, and we're running into these MD5 issue about every third disc. Can you please email me and tell me more detail about your rationale for concluding this is from a corrupted source file?

[Spoon]

It is not rational, it is a fact, when compressing a FLAC file the input data is hashed, when reading the FLAC file it can be computed again if the md5 does not match the file is corrupted (or the program used to create the FLAC file wrote the wrong md5).

I have 1000's of FLAC files, I can convert them all to 'Test Conversion' and not a single one fails the MD5 test, because none are corrupted.

[Darryl R]

"Rationale" is simply a synonym for logic. We're in the middle of very time-consuming project, all I've requested is you take the time to give a thoughtful, complete answer to my questions. I don't know the format of the files, and I had asked about the origin of the source MD5 above, and you implied you computed it. Little by little I'm putting the picture together from your terse answers. Thank you.