Doom9's Forum - View Single Post

nurbs · 10th July 2009, 08:57

As I said on paper the 24bit files are definitly better, but the difference will not be audible in most cases. For a discussion of 16-bit vs. 24-bit you can take a look on this thread at Hydrogenaudio.
My point in my first post was that while the 24-bit file is preferable you shouldn't dismiss a tool just because it will decode the file only to 16-bit.

About Dialog Normalization:
Yes, it helps to get the dialog of different files to the same average level (not exactly the same!) if the value is set correctly, but that is different from the level the master of the file had. If you want to reencode the file I would remove the Dialog Normalization information so that the data you encode from is as close to the original as possible. Note that AC3 and E-AC3 tracks are the only ones which have dialog normalization by default. If you have multiple versions of the same track (e.g. blu-ray: an ac3 track and one lossless version) only the ac3 track will have the normalization. The lossless track will be exactly like the master and no normalization is applied.
Then there is the fact that, as tebasuna51 said, the dialog normalization is often left at the default value by the studios which makes it useless anyway.

About Stereo:
All my stereo tracks that originally came from 5.1 files are downmixed with DPLII and I haven't noticed a problem yet. In fact that method has become the default in many applications.
I like eac3to to do that, because it defaults to high quality. For instance when you downmix to stereo the volume of a track is increased and eac3to will automatically detect clipping in the output and correct it with a second pass. By the way, the higher volume after the downmix may be a reason to normalize the audio before encoding (has nothing to do with dialog normalization of course).

10th July 2009, 08:57	#5 \| Link
nurbs Registered User Join Date: Dec 2005 Posts: 1,460	As I said on paper the 24bit files are definitly better, but the difference will not be audible in most cases. For a discussion of 16-bit vs. 24-bit you can take a look on this thread at Hydrogenaudio. My point in my first post was that while the 24-bit file is preferable you shouldn't dismiss a tool just because it will decode the file only to 16-bit. About Dialog Normalization: Yes, it helps to get the dialog of different files to the same average level (not exactly the same!) if the value is set correctly, but that is different from the level the master of the file had. If you want to reencode the file I would remove the Dialog Normalization information so that the data you encode from is as close to the original as possible. Note that AC3 and E-AC3 tracks are the only ones which have dialog normalization by default. If you have multiple versions of the same track (e.g. blu-ray: an ac3 track and one lossless version) only the ac3 track will have the normalization. The lossless track will be exactly like the master and no normalization is applied. Then there is the fact that, as tebasuna51 said, the dialog normalization is often left at the default value by the studios which makes it useless anyway. About Stereo: All my stereo tracks that originally came from 5.1 files are downmixed with DPLII and I haven't noticed a problem yet. In fact that method has become the default in many applications. I like eac3to to do that, because it defaults to high quality. For instance when you downmix to stereo the volume of a track is increased and eac3to will automatically detect clipping in the output and correct it with a second pass. By the way, the higher volume after the downmix may be a reason to normalize the audio before encoding (has nothing to do with dialog normalization of course).