Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: ReplayGain Implementation / metaflac vs fb2k vs bs1770 (Read 3268 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

ReplayGain Implementation / metaflac vs fb2k vs bs1770

I'm getting inconsistent results.
Code: [Select]
foobar2000 replaygain scanner
-2.59dB

bs1770gain --replaygain
integrated:  -15.42 LUFS / -2.58 LU

metaflac --scan-replay-gain
-0.760002 0.825562 -0.760002 0.825562
I see finally the same RG values in fb2k & bs1770
but metaflac still produces a few dB louder than that.

Why is that?

Re: ReplayGain Implementation / metaflac vs fb2k vs bs1770

Reply #1 –
Both foobar2000 and bs1770gain use loudness estimation defined in ITU-R BS. 1770. It can have some differences with the original ReplayGain method but on average they are pretty close.

I'm curious, what do your ears say about this track? Which loudness value gets the track closer in amplitude to all your other tracks?

Re

Reply #2 –
Quote from: Case on
Which loudness value gets the track closer in amplitude to all your other tracks?
Not sure. I guess it depends on how
other tracks have been replaygained.
Or do you mean can i hear the 2db diff?

I'm looking for consistency.

I've noticed this phenomena on my Rockboxed device
when even replaygained tracks had definitely var loudness.
So i've been researching the rootcause of this discrepancy.

That was when i noticed this issue.


All of those replaygainers have
the same 89db ref level it seems
whatever that means.

The only other clue
maybe i've heard there's
some kind of calibration
to that reference freq is done.
Sine/square wave or something.
But i've not found enough inf on that yet,

Re: ReplayGain Implementation / metaflac vs fb2k vs bs1770

Reply #3 –
Hmm wtf
i just tried
another track of
different music genre.

All three RG values
are almost the same now.

So does it depend on the track itself?
But var players are gonna apply
just the written RG value.
So it could cause
volume inconsistency
is that right?

Re: ReplayGain Implementation / metaflac vs fb2k vs bs1770

Reply #4 –
The difference between the original ReplayGain algorithm and the newly adopted ITU-based one indeed differs per track. The differences shouldn't be huge in general. The estimations can never be perfect as playback loudness matters, equipment have different frequency responses and everyone hears differently.

But if you want to be sure all your files at least agree on the computation method I recommend scanning your collection with foobar2000. Its scanner is heavily optimized and multithreaded so the task finishes pretty much as fast as is possible.

Also check the ReplayGain settings you use in Rockbox. If you play individual tracks you should use track based ReplayGain. Clipping prevention especially combined with positive preamp could easily cause unnecessary volume variation.

Remember that you can also always manually adjust the RG values for any track. Though this is most likely not doable on Rockbox.

Re: ReplayGain Implementation / metaflac vs fb2k vs bs1770

Reply #5 –
Quote from: Case on
The estimations can never be perfect as playback loudness matters,
 equipment have different frequency responses and everyone hears differently.
Thanks for your response..

Re: ReplayGain Implementation / metaflac vs fb2k vs bs1770

Reply #6 –
I recently rescanned my collection and it was a mistake.

The new little darling 1770 punishes heavily compressed tracks too much, at least in the hard rock/metal genres.

Re: ReplayGain Implementation / metaflac vs fb2k vs bs1770

Reply #7 –
Quote
but metaflac still produces a few dB louder than that.
I'm not familiar with metaflac, but those numbers look like gain factors (not dB).    A factor of 0.76 (76%) is -2.38dB.

Re: ReplayGain Implementation / metaflac vs fb2k vs bs1770

Reply #8 –
There's a negative sign in there.

As such, it should be fairly obvious the -0.760002 is the correction in dB and the 0.825562 is the peak value.

My recommendation is to spend 5 minutes and familiarize yourself with metaflac. ;)