I've written a little script that walks through my Flac library and uses oggenc to transcode to Ogg Vorbis, creating a 'mirror' library in the latter format.

What's the recommended way of dealing with the ReplayGain tags in the Flac files when transcoding? Can they be transferred as-is to the Ogg Vorbis files, or should they be dropped and then run VorbisGain if you want gain tagging?

The gain value will differ slightly, potentially significantly at much lower bitrates.

The peak value should definitely be recalculated if you have your player configured to adjust gain to prevent clipping. Transform codecs usually go over 1.0 on transients so this could be a factor.

If you don't mind a little bit of (likely inaudible) clipping and are encoding at a high bitrate, you probably don't need to re-calculate replay gain. If you need the RG values to be exact, you'll have to re-calculate them.

Thanks.

How exactly does the --FLAC-scale option work?



I take it that it actually alters the music instead of adding gain tags. Does it apply the gain to the WAV samples before encoding in Ogg Vorbis? It's not crystal clear, but that's how I read the phrase 'scale input data'.

You're right. The decoded FLAC samples are scaled according to the RG tags prior to passing to the vorbis encoder.