Well, the title and description pretty much say everything, but I'll explain it with more detail anyway.

I haven't really tested it much to really see if it causes problems, but just thinking of it suggest me that it must not be very good to feed vorbis (or any lossy codec) with badly mastered stuff like clipped or dynamic range squished cds due to loudness wars without normalizing them so the encoder has something easier to swallow that may end up with a better encode.

My point is that, in the end, I would still play them with replaygain and the result would be almost identical, and the only reason I can think of to not use a destructive process such as normalizing would be if encoding to a lossless format.

So, my question is if, given sub-optimal source material (clipped or almost), it could be advisable to normalize/replaygain lossless material prior to encode to vorbis and other lossy formats. Would there be a benefit as I suspect or am I just talking nonsense?

Wavegain > lossless > vorbis . That way replaygain will work on any player.

Welll, thanks, but that's nothing to do with what I'm asking.

If you use oggdropXPd on FLAC files that have been replaygained, you can specify that the input data should be scaled according to the RG tags prior to encoding.

Edit: And, yes, it would probably be of benefit.

1) Dynamic range compression has no effect on lossy compression quality, other than to possibly affect the final bitrate.

2) If the material is clipped then adjusting gain before lossy compression will make no difference. It will sound just as bad either way.

3) Lossy compression can increase the probability of clipping on playback, but if you are playing them with replaygain then that should prevent this from happening, whether or not you adjust the gain before encoding.

The bottom line is - just use replaygain and don't adjust the gain before encoding. Or, as shadowking suggested, apply wavegain and then you don't need replaygain.

You all are telling me methods of doing what I proposed and not telling me if (and why, if possible) there is a benefit to it.

If you are curious, I'm actually using foobar2000 as a means for converting some lossless material I have to vorbis, so applying replaygain to the PCM data before encoding is as trivial as checking the "ReplayGain Processing" checkbox in the converter dialog.
But I dare to guess all other methods described would yell very similar if not identical results.

Now.
john33 actually thinks it could be of benefit -as I was thinking-, but pdg says it isn't... at least I think he's saying that.

Any other opinions?

The obvious benefit of applying gain adjustment right before encoding (via encoder's scale parameter preferably) is that it will not sound too loud or cause additional clipping on devices without RG capability.

Replaygaining before lossy is ok, and shouldn't cause problems since the quality that you could loss is already below what a lossy encoder would keep.
Normalizing wouldn't be as good as replaygaining, since the advantage of replaygaining is that you don't apply too much nor too little.

If you want more information, you always can go to this thread:

http://www.hydrogenaudio.org/forums/index....showtopic=10637