Sound 5.1 to 2.0 (voice loudness enhancing)

[alexVS]

I don't have audio receiver. My Samsung DVD player is connected to TV Sony Bravia via SCART. The 5.1 AC3 from DVDs sound bad. Sometimes too loud, or too quiet. Speech sometimes is hard to understand. When the same movie goes on TV program, sound is much better.

I tried to convert it to 2.0 in advance with eac3to, foobar2000 etc, but the loudness is the same. With the same dynamic compression, and hard-to-listen speech level.

Some software players (and almost all receivers) allow to use "night mode", audyssey Dynamic Volume, etc that make speech clear and other sounds more quiet.

Is there an algorithm of this conversion?
I just want to take 5.1 AC3 track, convert it to 2.0 using "night mode", write it back to DVD-RW.

Take some time, but I want to try.

[Emulgator]

I would go the manual route and make my own envelopes on top of all the forced thunder.
But if you prefer an automated way you may want to use Audacity and plugin Leveler.

[pandy]

Check TV settings - sometimes additional audio processing is lost (by used algorithm) for complex Dolby (other multichannel?) downmixed sound personally i have this impression with my TV set when AVL (automatic voice level) function is ON - some downmixed 2.0 (from 5.1) are OK sometimes they are horrible - voice is barely audible - sound like removed from overall audio by some "karaoke" filter - this is source dependent - plain stereo settings without additional audio processing are OK.

PS
AC3 use dynamic compression, RF mode (or Line) and one of preselected render modes (Room, Theater etc) - this all affect way how loudness and overall dynamics is perceived - some of those things are controlled by metadata, some "hard"coded in decoder... - i hate Dolby for this...

[hello_hello]

If you can decode via DirectShow while re-encoding, maybe use ffdshow for the decoding and it's mixer filter to increase the volume of the centre channel. With the mixer filter enabled and set to downmix to stereo, you can use a slider to adjust the volume of the centre channel (up to 300%) which should increase the volume of most speech, given it's generally in the centre channel. You can also use a slider to adjust the level of the surround and LFE channels in the "mix".
There's a checkbox to select a "normalize matrix" for downmixing, which I'd probably use in this case to ensure there's no clipping. If you re-encode using MeGUI's audio encoder you can choose to "normalize" (increase the volume so the peaks are at 0db) which should make up for the drop in volume resulting from enabling ffdshow's "normalize matrix".

MeGUI's audio encoder lets you choose a preferred decoder, which should make it fairly easy to force DirectShow decoding. Alternatively there's an AVS input plugin (http://www.foobar2000.org/components/view/foo_input_avs) for foobar2000 which lets you open video/audio files via a DirectShowSource Avisynth script.

ffdshow also has a Winamp filter which lets ffdshow use winamp plugins. Here's a couple of compressor plugins you might also like to try using while re-encoding.
http://www.winamp.com/plugin/rocksteady-2-1/1099
http://www.winamp.com/plugin/loudmax-1-13/221811
I use my PC as a media player so rather than re-encode with compression applied, I use the RockSteady compressor on playback.