MPC-HC plays music without ID3 tag more loudly

With normalization enabled for built-in audio switcher the same mp3 file played more loudly, when I removed the ID3 tag (and it returns to play this file less loudly when I added an ID3 tag again).

It is true for both 32-bit and 64-bit versions, but only if normalization is enabled.

(Windows 7 Ulitmate 64-bit, both 32-bit and 64-bit MPC-HC 1.7.9.165)

The ID3 tag has no effect at all on normalization. The algorithm reset each time you seek to a different position. So that is probably why you heard a difference in volume.

Thank you for your explanation but theory is one thing and the reality is sometimes other one.

I tried it again and again (e.g. with “Tchaikovsky - Piano Concerto No.1 in B flat minor, op.23“) from beginning without seeking to a different position, both with the reusing the same instance of MPC-HC in RAM and with fresh instance each time, and it behaves as I wrote. The difference is significant.

Did you try it by yourself?

The normalization algorithm operates on the decoded audio. It has no knowledge of the audio format that you are playing, nor about metadata such as ID3 tags. That are pure facts. If something different is happening, it must be done by the decoder. Does it have a ReplayGain tag?

Can you upload the two files somewhere?

Normalization is mainly a feature for movie viewing, where there can be big volume differences (voices vs explosions). It is not something I would recommend for audio playback, certainly not classical music.

Here are these 2 files:

http://letitbit.net/download/52898.5eeb83429f7a865c764bda46e87d/Tchaikovsky.mp3.html

http://letitbit.net/download/18146.147941c9c7065ce58e3c749c9956/Tchaikovsky_with_ID3.mp3.html

For adding / removing ID3 tag I used freeware program TagScanner 5.1.668 at http://www.xdlab.ru/en/

I am having trouble downloading from that site. Could you use sendspace.com or datafilehost.com?

OK, they are now here:

Tchaikovsky.mp3
Tchaikovsky_with_ID3.mp3

And thank you for your explanation, I didn’t know neither that the flaw may be also in the decoder nor the true purpose of the normalization.

I honestly don't hear any difference.

You could try with a different filter:
MPC-HC options > External filters > Add "ffdshow audio processor" and set it as preferred.
Double-click on the added filter to access its settings. Enable "Volume".
Disable mixing in LAV Audio Decoder.

Thank you for your help and here is my result of using it:

I set all as you wrote (first time except your last sentence) - the difference in loudness is smaller but still very, very apparent.

Your last sentence was a bit confusing for me, so then I added the LAV Audio Decoder, too (but not as preferred), and disabled mixing in it. Without any new effect.

But then I examined it more closely and I discovered that loudness of these two files reached the same level after some time of playing:

- without external filters about 35 seconds after the start,
- with your recommendations as early as about 7 seconds after the start.

Is this delay caused by the normalization algorithm?

You can see which filters are used by looking at:
MPC-HC menu > Play > Filters

The algorithms of MPC-HC and ffdshow are slightly different, so it is not surprising that they behave different as well. But I still don't have any explanation why the behavior differs for you depending on the presence of an ID3 tag.

You can click on ffdshow systemtray icon during playback. It will show the current amplification factor on the volume page.

Thank you for your advice, so this time I watched optically (and more precisely) the amplification factor:

• WITH ID3 tag it began at 100% and increased progressively to about 260% at 27th second (so NOT at 7th second). Then it oscillated around this value.
  
  
• WITHOUT it it began at about 210% and increased progressively to about 240% at 10th second and then it began to oscillate.

My guess is that the normalization algorithm computes some crucial value a lot of seconds – concurrently with playing music – and that ID3 tag makes it longer. However, it is in contradiction with your statement that this algorithm operates on the decoded audio.

What else may be different between your settings and my ones? I have no idea, but I am not very experienced.

Options > Internal Filters > MPEG Audio

With that option enabled (which it should be by default in the pack) I get consistent behavior. In this case LAV Splitter is used.
With that option disabled, I can reproduce your numbers. In this cause the Microsoft MPEG-I Stream Splitter gets used. I get an audible glitch at the start with the ID3 file. That probably confuses the normalization algorithm.

Thank you very much!

I do apologize - this whole problem was caused by disabled “MPEG Audio” internal filter.

I noticed it by other awful effects as bad indication of total time, huge difference between loudness of individual MP3 files and strong snap in the begin of the playing of files.

It is some mystery for me how it may occurred as I am not tending to change things which I don’t understand.

So now ID3 tag has no side effect.

Thank you again for your great patience with me.

No problem at all. Analyzing this kind of issues is a good way to gain new knowledge.