Topic: wav compare utility

[Lil Kim Jong-Il]

Honestly I searched here and all over google and got nothing that was useful but I am sure this has to have come up before:\

I need a utility that will compare wav files and report the sequence of matching and non-matching segments, somthing like the functionality in EAC to determin write offset but but I need it for 24bit.

Anyone have something like this, preferably cmd line tool for batch processing? 

 

[admkrk]

wouldn't flac do that?

[scb]

shntool?

[Lil Kim Jong-Il]

Thanks for the suggestions.

To my knowledge, flac does not do what I need.  I'm not comparing an original to an extracted version which is what flac will report against.

The shntool cmp mode is not much different from the standard cmp/fc command line utility.  What I'm looking for is something that will show duplicate sample ranges, missing sample ranges, difference ranges, etc and the offset of the sync points where the differences are detected.

I need to diagnose errors in decoded DTS-MA tracks.  I can digest the results to see if there is a difference in the LPCM produced but it would be useful to have the more detailed information so that I know if we are dropping frames, duplicating frames, putting frames out of order, etc.  I have hex comparison tools but the files are so large that the tools break when loading the files and I don't want to split the files.  Even using the graphical tools, visual inspection to locate duplicate, out-of-order, etc conditions is to laborious for the test set in use.

I'm really trying to not write something.

[it-goes-to-eleven]

I've done a fair bit of that while developing signal processing software, noise filters, interpolation, etc.  Though your case is different.   There are libraries that help, but sometimes it is quicker to write your own code.

An easy quick check is to invert one source and mix it with the test case.   A null result confirms that they are identical.   If you find they are not identical, analyzing/visualizing the resulting absolute values will give some idea how close they were.

Another lazy option would be to create a two channel mix where the left channel is your original, and the right channel is the left channel of your decoded track.  Then, a tool like baudline under linux would allow you to viz the FFTs over time and give some insight into how they compare.