Hi,
I'm wondering what the best techniques are to test how different a recording is. I have an album that claims to be a remaster, but to my ears it sounds no different to the original. What software can I use to compare the two audio files and measure their percentage difference, much like the diff command on linux.
Thanks guys,
sabrehagen
Hi,
I'm wondering what the best techniques are to test how different a recording is. I have an album that claims to be a remaster, but to my ears it sounds no different to the original. What software can I use to compare the two audio files and measure their percentage difference, much like the diff command on linux.
Thanks guys,
sabrehagen
Audio Diffmaker might help , if they aren't *too* different.
http://www.libinst.com/Audio%20DiffMaker.htm (http://www.libinst.com/Audio%20DiffMaker.htm)
Hi,
I'm wondering what the best techniques are to test how different a recording is. I have an album that claims to be a remaster, but to my ears it sounds no different to the original. What software can I use to compare the two audio files and measure their percentage difference, much like the diff command on linux.
Thanks guys,
sabrehagen
Audio Diffmaker might help , if they aren't *too* different.
http://www.libinst.com/Audio%20DiffMaker.htm (http://www.libinst.com/Audio%20DiffMaker.htm)
Okay thanks, I've used that one before, just wondering if there were any alternatives.
Ugh, the golden test to determine (audible) differences here, is ABX testing.
Yeah, percentages aren’t going to have any useful correlation with quality. Even after you account for differing offsets or whatever and are lucky enough not to be completely confounded by samping drift in the case of transfers from analogue, etc.
If you can’t hear a difference, accept that as your conclusion.
But if you insist upon going beyond audibility, there are a number of far, far more useful metrics that can describe physical differences between two streams.
Edit: I just noticed where this thread was posted. If you decide to conduct some more listening tests, feel free to request it to be moved back, but until then, a thread about using measurements instead clearly does not belong in Listening Tests.
Okay thanks, I've used that one before, just wondering if there were any alternatives.
I do like Masvis (http://www.lts.a.se/lts/masvis) : this soft does a very good analysis of masterings and can be used to compare tracks : levels, bits used, clippings, dynamics,...
Ugh, the golden test to determine (audible) differences here, is ABX testing.
Sure, I agree with that. The reason I was looking for a software comparison was to see if I had exactly the same file, except one was called 'original' and one was called 'remaster'. It was more to test a false labeling of the track information.
while masvis focusses on crest factor,
wavdiff (http://asp.lionhost.ru/en/tools_wavdiff.html) will show the exact differences between 2 wav files.
while masvis focusses on crest factor, wavdiff will show the exact differences between 2 wav files.
As an example, here are 3 versions of PinkFloyd, DSOTM, Money analysed wtith Masvis and wavdiff (last picture).
For me the analysis with Masvis is easier (wavdiff is only with commandline), more convenient to read and gives more information on the differences. When I compare masterings, I want to know if compression has been used, if there are overFS, if some EQ has been done, etc....those things I can read from Masvis analysis, not easily from wavdiff (or maybe I missed something).
Note that I had to convert my FLACs to wav so wavdiff can read it.
(http://i74.servimg.com/u/f74/11/21/77/49/dsotm110.png)
(http://i74.servimg.com/u/f74/11/21/77/49/dsotm210.png)
(http://i74.servimg.com/u/f74/11/21/77/49/dsotm310.png)
(http://i74.servimg.com/u/f74/11/21/77/49/dsotm410.png)
(http://i74.servimg.com/u/f74/11/21/77/49/dsotm510.png)
Concerning AudioDiffmaker and wavdiff, computing the difference between two files and then listening to it is generally not a good idea for our purpose : some many differences just come from diference of timings or phase differences and are difficult to avoid. An audibility of the difference file is no proof of audibility of difference between the source files (only an ABX will help, as skamp told).