wavegain seems to chop up the audio when normalizing
Closed this issue · 4 comments
I have an audio file of the following specs:
Input File : 'outsox.wav'
Channels : 2
Sample Rate : 44100
Precision : 24-bit
Duration : 00:01:00.00 = 2646000 samples = 4500 CDDA sectors
File Size : 15.9M
Bit Rate : 2.12M
Sample Encoding: 24-bit Signed Integer PCM
When I try to normalize it with wavegain -y I get a file with the audio being chopped up. It also seems to have a pitch an octave higher than the original, which would point to metadata being wrongly interpreted...
The original file: http://bernhardwagner.net/outsox.wav
The file generated by "wavegain -y": http://bernhardwagner.net/outsox_wavegain.wav
This is wavegain v1.3.1 on Mac OSX 10.8.4, compiled from source.
Thank you
Bernhard
Thanks for the report!
Can you test it against the upstream, windows version of wavegain v1.3.1? My changes to upstream consisted mostly in packaging, compiling and security issues, to make it compatible with Linux (and, thanks to you, Mac OSX). I didn't change the normalization algorithm at all. So a bug like this (audio chopped and higher pitch) is mostly likely from original code.
Ok, I tested it with the windows version and the error did not occur.
Yes, I did the same test and noticed this too. In windows, using the pre-compiled binary, your song is fine, but in Linux the audio is chopped and pitch is higher. Maybe the code is not 64-bits compile-friendly (the windows build is 32 bits), other than that I have no idea. Question: does this happen for all your songs or just for this particular one?
It happens with all files that I record using the Zoom H4 (http://www.zoom.co.jp/english/products/h4/)