For all those people interested in the application of replaygain, here is WaveGain!!

Basically, it applies the same principles to original wave files as are applied using replaygain on ogg/MP3 files, etc. However, rather than tagging the files, the gain is applied directly to the file, so, if you need the original wave in its untouched state, you will need to back it up.

Speek has produced a frontend for this(thanks! ), and the download from Speek's site includes both the frontend and the WaveGain.exe. You will find it useful to read the txt file that Speek has included. In addition to applying either the Album or Title gain, you can also select a different output format - 8bit, 16bit, 24bit or 32bit integer, or 32bit float. Input must be 16bit integer, ie, normal waves.

This is a first public release, so any constructive criticism will be welcome regarding changes/improvements, etc.

john33

Speek's site:
http://home.wanadoo.nl/~w.speek/speek.htm

1) How does this handle "album" gain: I mean - I like to prevent clipping but don't want a quiet part to sound as loud as the rest!

2) This might be a "newbie" question... reducing the level of the wav file won't alter the quality of the encoded file? - psychoacoustic analysis has something to do with loudness, right?

Rather than waste a lot of words, I suggest that you try it and see whether you like the result. Briefly, it works exactly the same way as replaygain for ogg except that instead of writing tags for subsequent use by a decoder, it applies the gain (+ or -) directly to the wave file. You can apply Album gain by specifying a list of input files or by inputting the whole directory as '*.wav', for example. It's all in the 'help'.

john23

Hehe, like that. Will do, thanks!

This is great! Now I can burn custom audiocds where all tracks are the same volume

I'm curious though... How are the gain calculated? Is it just a rms value over the whole file, or is it more advanced?

I was thinking of something like this: First divide the file into a couple of frequency bands, then measure the energy of each band based on some ath value (like mp3...). Finally sum all the energy values togheter. This way you will have the real percieved loudness of the audio (I think). But this might be awful slow compared to doing just a rms...

The guy that is responsible for the philosophy is David Robinson(2BDecided, on here). There is full info on his site: http://www.replaygain.org/ .

I claim no real originality other than deciding to implement it this way.

Glad you see some benefit in it!!

john33

.../replaygain.exe -a ABBA*Best*Album*By*Best*Band*Ever*.wav

It's the -a (audiophile/album) switch that does the trick. Calculates the overall replaygain value for all the specified files and applies the same numeric gain

Maybe. I don't know for sure, but I think it might have a small significance. If you use wavegain on wave-files and then encode to some lossy format the reduced volume of the sourcefiles might have some importance. But if you are going to encode to one of the three "popular" lossy formats (OGG, MPC, or MP3) I think you'd be better off using the replaygain-tools for those instead of wavegain

Just as a point of interest, the replaygain programs for these "popular" lossy formats, as you put it, all use exactly the same backend program (well, almost) that does the basic work.

john33

I know. It's just that if you use wavegain, encode and share these files, other users won't know what the "real" volume should be. Of course you could argue that Mp3gain have the same disadvantage.

I once had the fun experince of listening to a MP3 downloaded by swarm via the Gnutella network which had audible different volumes in segments of the song. If I hadn't found it so amusing I would have found it quite annoying

Well, life's all about options isn't it? And this is another one. Maybe it will be useful to some people and not others.

john33

Yup. And I appreciate having gotten the ability to use wavegain. It's perfect for making audio-CD's with songs from mixed sources. Thanks