Author Topic: Issues when analysing / writing replay gain tags of my FLAC archieve  (Read 1536 times)

tubescreamer

  • Jr. Member
  • **
  • Posts: 83
Hi,

I am facing some problems when performing volume analysis to write replay gain tags.
I have a big music archive, volume normalization takes quite some time.

After the replay gain analysis I feel like alone with a huge list in a tiny window
where it's nearly impossible to browse through all titles to get an overview which adjustment I have to perform.

Why is it not possible that MusicBee finds a proper value for adjusting the replay gains up or down on it's own ?

I am aware of that we are not peak level optimizing, that we talk about something like subjective perceived loudness.
But still there is the question, when seeing all the different adjustment values, what final adjustment do I need to perform ??

Proposal: additional sortable colums for title and album replay gain tag adjustment would support the review process.

I see that foobar2000 for example 1st analyses and then writes the replay gain tags.

Wouldn't it be better for MusicBee to 1st gather the max level and loudness of all titles and then coming to conclusions to which common loudness all titles could be led ? At the end we write replay gain tags per title.

If the music material is very different I could think of it's tricky to get both .. a common loudness for all titles on one hand and on the other hand not come over 0dB for each of the titles.

But somehow I have the feeling that this should be solveable by algorithms and not letting the user alone to adjust something where he can not really oversee what to do .. because nobody tells him exactly which value to take or what makes sense.

Foobar2000 also claims to be more precise when using oversampling (to increase sample rate by x2, x4, ...).
Would maybe be also a nice idea to make this analysis more thorough.

Any thoughts or recommendations to this to make this process better ?

Many thanks
Last Edit: April 27, 2019, 09:45:59 AM by tubescreamer

captain_paranoia

  • Full Member
  • ***
  • Posts: 207
What tool are you using to do volume analysis?

Tools/Analyse Volume

That does exactly what it sounds like you are trying to do. You tell it what target volume to level to.

If you're playing files to a UPnP/DLNA renderer, the volume levelling tags may not be used.

BTW, volume analysis and volume normalisation are generally two different things; volume analysis creates a replay gain tag. Volume normalisation modifies all the samples with a volume scaling.

tubescreamer

  • Jr. Member
  • **
  • Posts: 83
Thanks for your reply, I use MusicBee.

But sorry, you didn't answer my question, see in bold.

For me is all clear except the point what value shall I dial-in after the warning that for some titles distortion might happen.


hiccup

  • Sr. Member
  • ****
  • Posts: 7873
Why is it not possible that MusicBee finds a proper value for adjusting the replay gains up or down on it's own ?

Because it is not implemented.

There is a wishlist topic for requests.


captain_paranoia

  • Full Member
  • ***
  • Posts: 207
Thanks for your reply, I use MusicBee

You must asked my point: what tool, within MusicBee, are you using to do volume analysis?


redwing

  • Guest
Why is it not possible that MusicBee finds a proper value for adjusting the replay gains up or down on it's own ?

That means such action is not expected normally and they are provided only as reference since it will defeat the purpose of volume normalizing of the entire library.
But I saw some people manually re-analyzing clipping tracks/albums in accordance with the suggested values, which amounts to a manual implementation of "prevent clipping" option that some other players (foobar, winamp, etc.) offer.

But still there is the question, when seeing all the different adjustment values, what final adjustment do I need to perform ??

There's no definitive answer. It's up to how you're gonna deal with the clipping issue from ReplayGain since everyone has a different music collection and different opinion about the issue.

Proposal: additional sortable colums for title and album replay gain tag adjustment would support the review process.

It's already possible if you use virtual tags. Also you can re-adjust RG tags without having to re-run volume analysis with a different target loudness, if you want.

See:
https://getmusicbee.com/forum/index.php?topic=9317.msg140321#msg140321
https://getmusicbee.com/forum/index.php?topic=9317.msg140323#msg140323

I see that foobar2000 for example 1st analyses and then writes the replay gain tags.

Wouldn't it be better for MusicBee to 1st gather the max level and loudness of all titles and then coming to conclusions to which common loudness all titles could be led ? At the end we write replay gain tags per title.

Not sure what you mean, but MB also writes RG tags right before it pops up the clipping warning dialog.

If the music material is very different I could think of it's tricky to get both .. a common loudness for all titles on one hand and on the other hand not come over 0dB for each of the titles.

But somehow I have the feeling that this should be solveable by algorithms and not letting the user alone to adjust something where he can not really oversee what to do .. because nobody tells him exactly which value to take or what makes sense.

There are various ways programs/people are using to prevent clipping, but each way has its own pros and cons. You might want to google "prevent clipping replaygain" to get more information. Keep in mind that if you're not using RG tags, there will be no clipping issue. RG tags are only useful if you usually listen to tracks across albums like with shuffle on. If you usually listen through an entire album, you don't need RG tags.
Also you need to consider how many tracks have really audible clipping in your collection, and whether it's really worth it if you want to use a certain way to prevent the issue since every way has some downsides.