On Mon, 18 Jul 2005, Joerg Lehmann wrote:
On 18.07.05, Dag Wieers wrote:
I would already be glad if we had support for the ogg replay-gain support (most of my soundfiles are ogg files anyway) and creating tags is best left to specialized tools anyway, so using the correct (available) tags is the only thing we need to care about.
And looking what tags have been added by the authoritative (available) tools for Linux will further trim the possibilities I guess.
That's what I was trying to do for MP3 files. However, I failed because I couldn't manage to bring the supposedly authorative tool mp3gain to insert tags (and not to adjust the whole file) ;-) And if somebody provide me with a pointer to a reference where I can find all this, it would really help.
How did you verify that not only the tags were changed ? I'm now running vorbisgain through my whole collection of music and I don't know if it is doing more than just adding the tags :)
Did you dd parts of the file and checksummed them ?
As far as I understood, the difference between track gain and album gain is a technical one in the sense that you may want all tracks from the same album being 'leveled' based on all tracks (and not on a per track basis). For a classical CD or a rock concert, this is pretty much required.
I don't think these are different tags, just the value is based on different sources. So for playback (pytone) it doesn't really matter.
I think it matters. The idea is proably that when you play the whole album, you should use the album gain. Unfortunately that's not so easy to achieve. Suppose the following playlist
1. Album1/Track 1 2. Album2/Track 1 3. Album2/Track 2
....
If we only look one song in advance, we will find that the first playlist item uses the track gain, the second one as well because it's from a different album, and when playing the third one, we realize that it's from the same album and that we had better taken the album gain for the previous song. But then it's too late...
And even, if we recognize albums correctly, we will have a different loudness of track 1 and 2.
So, it's probably impossible to do the correct thing when mixing album and single song playback and maybe the only solution is to ignore the album gain altogether.
Well, it would be already nice if we had 'radio' gain support. But an improvement would be if the algorithm could scan the remaining queue (if not empty) and verify if the next song(s) are from the same album when it starts playing a song.
In the unlikely event that this is the case, than it could switch to 'audiophile' gain support. Doing it like that makes sense to me.
In the cases where you want 'radio' gain support you're not planning to play a song from the same album in a row, so that's ok. And in the unlikely case that the random playback chooses a song from the same album (when you are in 'radio' mode anyway for a party or at home) it will not use 'audiophile' gain since the queue is empty at all times.
Does that make sense to you too ? :)
PS My previous description was wrong, sorry for the confusion. Like Stefan I pretty much agree that replaygain support is a requirement these days. It's annoying to have to adapt the volume from time to time.
Kind regards, -- dag wieers, dag@wieers.com, http://dag.wieers.com/ -- [all I want is a warm bed and a kind word and unlimited power]