ac3 encoding - error in reasoning with parameters?

[kotyczka]

For a DVD I encoded a wav file to ac3 by Audio Transcoder 3.0.1.
In order to get it loud (my notebook has only very small
speakers) I set the dialog level to -31 , but the real
sound RMS is larger. If using Dynamic compression now
spoken word would have been compressed. Therefore I choosed
'no compression' as well as 'no RF overmodulation protection'.
At silent places the background noise grows, until some
starts speaking, the I reduces immediately. I had thought
that such an effect will only happen if I use dynamic range
compression _and_ set a wrong dialog level.
Is there still an error in reasoning?


I must tell, that ac3 is rather new to me.
I would be very glad if someone could enlighten me
what#s going on.

[Skelsgard]

The parameters are correct (but not for the right reasons) but for a "loud" that can be defined as to play the sound in its full volume without dynamic compression.
If you need to sound to be louder as in to have higher RMS, you need to give some extra gain to it when transcoding.

Cheers.

[kotyczka]

The parameters are correct (but not for the right reasons) but for a "loud" that can be defined as to play the sound in its full volume without dynamic compression.
If you need to sound to be louder as in to have higher RMS, you need to give some extra gain to it when transcoding.

Cheers.

Oops, Didn't understand that exactly. How do I "give some extra gain"?
In the wav file the peak value is already
at -0.05 db.
Or is there some "extra gain" option in the encoder software?

[Skelsgard]

To get a higher RMS you can use software like BeHappy or BeSweet but with that peak level any more gain will produce digital clipping so it´s not recommended.
Such option is not available in Audio Transcoder (the Scenarist one, right?).
The Dialog Normalization is something like this (not exactly):
the dialog is at about -31dB from the loudest sound. So if you select -31 dB it will take that the dialog is 0dB away from the "normal". But if you select -27dB then the dialog is +4dB above the "normal", so when reproduced the sound should be lowered 4dB. this explains why AC3 streams play at a lower volume when DNR is active during decoding.
Dynamic Compression on the other hand, depending on the profile selected, basically (not technically) compresses the sound, lowering or raising the volume to achieve a given level throughout the entire sound. This explains the fluctuations of volume for a given sound (i.e. speech) when a louder sound appears, when that particular sound actually had no changes in volume in reality.
This parameters can be ignored by the decoder in good quality decoders (example: AC3filter, PowerDVD) but not in others.
The difference is that AC3filter is meant to be used by any Directshow based player, whereas PowerDVD´s is meant to be used with PowerDVD (some features might not be available for other Directshow players).

Cheers.