Flac to mp3 foobar/converter/replaygain questions
Reply #3 – 2008-06-15 20:02:40
Will it calculate it's own replaygain parameters or must the original flac file have replaygain already in it's parms? Is apply gain or apply gain and prevent clipping according to peak recommended? Anyone know what prevent clipping according to peak does? Pretty certain it won't scan for RG when converting. You still need to scan for replaygain or it'll just apply the default pre-amp for files without RG if you don't have RG on a particular file. Simply select all (Ctrl-A) for the whole playlist you wish to convert or for your whole collection, then right-click and scan them from the Replaygain sub-menu. It'll use the tags to work out which albums belong together. It usually won't scan files that already have full RG info in tags so won't waste time on those. For some double-albums, I select the tracks of both CD1 and CD2 and "Scan selection as single album", though it's usually close enough not to bother. I usually apply Album Gain instead of Track Gain in case I listen to a whole album, unless I know I'm only shuffling (and even when shuffling Album Gain is very good). As for clipping, FLAC files can't clip when decoded unless a positive gain is applied, so as most RG values are negative, you'll be fine. A few very quiet tracks on modern (loud) albums might clip when track-gained but not when album-gained. I have a few classical albums which clip because a positive gain is applied but usually only momentarily during a loud percussive crescendo such that I can't ABX the difference. I'd be tempted not to prevent clipping, but perhaps to employ the Advanced Limiter at the end of the converter's DSP chain. This look-ahead DSP won't modify anything unless clipping would actually occur then tries to minimise the distortion by applying soft peak limiting at and around the time of the excessive peak. You then cannot pass any hard-clipped data to the MP3 encoder even with a positive gain applied. As a side note, it's possible for an MP3 to contain higher peaks than the WAV/PCM data fed into the encoder, but any of these that clip, esp the few that you'll have at an 89 dB RG target, are highly likely to be inaudible because they'll be so brief (often only one sample amid a cacophony of sound). I'd recommend letting foobar2000 send 24-bit data (24 = max bits per sample) to LAME (assuming that's your encoder of choice) so that the replaygain adjustment plus output dither can have no adverse effect on the relative noise floor of your audio. (edit: it's probably doing this by default, anyway, unless you've set up a custom commandline encoder preset)