Software tool for comparing wav files

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Is there a program out there that will compare two wav files and show me a sample by sample difference ?
Not just an all or nothing ('match' or 'no match') for the whole file, but if there are differences then it should show the specific samples that are different and their values.
 
Last edited:
Audactiy is a open source sound editor - you can certainly add, subtract, scale, zoom .wav

Audacity: Free Audio Editor and Recorder


Audio DiffMaker may be useful too - suposedly will time align, compensate for sample rate drift, linear frequency response

Audio DiffMaker

"Audio DiffMaker is a freeware tool set intended to help determine the absolute difference between two audio recordings, while neglecting differences due to level difference, time synchronization, or simple linear frequency responses.


for only few 10s of seconds files some math software lets you import/export .wav, create any comparision equation you can code: SciLab, Octave are free MatLab workalikes
 
Last edited:
At first Audio Diff sounded like the exact kind of thing I was looking for but it didn't work out very well. It actually tries to "normalize" both tracks first so to speak by normalizing the gain and time alignment and a couple of other things. That defeats the purpose of "numerical" comparison of samples.

I did try unchecking/disabling those adjustments but the program then just freezes up. Perhaps a bug ? I have to kill it from the task manager at that point.

The only closest thing I found was EAC (exact audio copy) which shows individual sample values that I could visually compare but how can anyone do that for 789383 samples ?

Here's what I am trying to do -
I have 2 wav files - one is a ripped cd track and the other is recorded copy of the same track play'd back on a cd player. Its recorded on the pc via spdif. Now I am trying to compare the tracks. Basic wav compare utilities indicate that they are not "identical" copies of each other hence not bit perfect. However, on visual inspection of several samples, and the fft spectrum as well, it appears that the recorded file is - for the lack of a better term - is "near bit perfect". What I mean by that is the spectrum of the recorded file is pretty much the same as the original, the values of randomly picked but still sequential samples were identical and whatever few samples I found that were not identical were differing in value by exactly 1 - meaning a sample value of 58 in the original file was 59 in the recorded file - so not by a whole lot. So I'd say this is as good as bit perfect but I want to confirm this for the whole file. Perhaps there is a pattern or it might lead to something that can explain why its not completely but only almost bit perfect.

I'd just like the program to tell me which samples are different and what are their values. Haven't found one yet so I guess I'll just have to write my own.
 
Last edited:
Hi percy, I have used AudioDiffmaker and Audacity for this over the years. As you have come across there are several issues. Diffmaker will work if you use a shorter length file, maybe up to 1 minute to compare - then it won't crash.

Using Audacity is a real trick as it takes significant effort to line up the tracks to start on the same sample and both end on the same sample. It is easy to be 1 sample off and not know it. I wrote a few procedures on that (and Diffmaker) at computer audiophile under the user name mitchco. Clock drift is also an issue and not easily navigated...

The only other software I know is DeltaWave Audio Null Comparator | DeltaWave documentation in which you may want to check out. Btw, the author of the software is user pkane2001 over at Audiophile Style:
DeltaWave null-testing audio comparator (beta) - Software - Audiophile Style
Paul is a good guy and if you reach out to him with your specific requirements, it may be easy to add the feature you are looking for if it does not already exist.
Cheers,
Mitch

Edit: Oops just saw the thread date...
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.