1. Ac3 to Wave (azid-1.7.1.zip)
This is just some info on why I've moved on to this method.  I've switched from px3conv to azid because it allows DRC(Dynamic Range Compression) using the same method as your PC DVD player. Azid is regarded as the best ac3 to wav converter by many, but since it doesn't have a 2 pass normalization feature(yet), I had to re-introduce myself to normalize.exe. But I think it's worth it. "azid -cnormal -L -3db input.ac3 output.wav" is the recommended command to use.  I found that using -clight allows LAME to create a lower bit-rate mp3, so I use that.  My first method of converting to mp3 didn't involve any DRC, so I'm happy using just light DRC as it will not alter the original dynamics as much as -cnormal would.  If you find that you can't hear some of the quieter scenes(could occur with lower end sound cards/speakers), by all means, put -cnormal instead of -clight. I have recently found that -cnormal can create lower bitrate mp3's on some movies than -clight so it's really a matter of personal preference.  -L -3db is what Doom9 recommends for correct downmixing.

2. Sample Rate Conversion (ssrc-1.28.zip)
This is not a new step to my guide.  This used to be done by LAME, but higher quality conversion can be done separately by SSRC.  Sample rate conversion seems to be a pretty controvercial issue in the DVD-ripping world, so if you feel you don't need it, go ahead and skip it.  I use it because I cannot detect any degradation in quality and it helps acheive a lower bitrate mp3 and with a batch file or BeSweet, the process is just as simple as before.

3. Normalizing (normalize0241.zip)
Next you want to normalize so that the audio spans the complete volume spectrum.  I think this can be done by simply dragging the wav on to normalize.exe, but I play it safe and use "normalize -m 100 audio.wav" which means search for peak value and normalize the entire wave accordingly to 100% of that peak value.  This is the same thing that PX3Conv does after it creates the wav.

4. Encoding the audio (Download the latest alpha Mirror 1 Mirror 2)

There are many different opinions on LAME options, r3mix.net is another great place to find out more about lame.

Here is my newest settings for DVD's using LAME: --nspsytune -V9 -mj -h --athtype 3 --resample 44.1 --ns-sfb21 2 -Z --nohist
My second setting that I used: -b 32 -m j -h -V 9 -B 192 --resample 44.1 --vbr-new
Here's what I used to do: -b 32 -m j -h -V 9 -B 192 --resample 48

I suppose I will attempt to explain the switches, but if you really want to know what's going on, visit r3mix.net.
--nspsytune: advanced psycho acoustic model enhancements
-V9 (VBR's lowest bitrate setting), sometimes I change it to -V8 if my video quality won't suffer. VBR defaults to --vbr-old so it is not needed
-mj: joint stereo (saves on bits, LAME is very good at it, unlike other encoders that attempt JS)
-h: high quality, same as -q2, it was found that using a smaller -q setting could result in lower quality at such small bitrates
--athtype 3: something about an athcurve, I have no idea, r3mix uses it and it works, that's all I know(athtype 4 is also a good idea as it does a smart switch between type 0 and 2 based on the -V value)
--resample 44.1: I've gone back and forth on this switch, some say not to resample, but I've found it can save 5-15kbps, which is a big deal with such long mp3's
--ns-sfb21 2:  I dunno, go to r3mix for this one.  It works and lowers bitrates and goes with the athcurve, that's all I know
-Z: disables noise shaping that is automatically enabled by nspsytune
--nohist: disables the histogram from being displayed. Has no effect on the mp3 file, but might speed up encoding slightly
I also got rid of the max bitrate 256 after finding that it saved less than 0.3MB on a 100MB file.

From my testing and with my speakers, this gives acceptable results.  My main focus is to get high video quality, and in order to do that, I need the smallest audio file possible. Bottom line is that I'm going for smallest filesize with acceptable quality, and this is what I've found to work the best for that purpose. It is certainly better than any CBR file of equal size.

Sample 1-CD Batch File
---------------------------------------------------------------------
@echo off
set ac3=%1%
set wav=%ac3%.wav
azid -clight -L -3db %ac3% %wav%
del %ac3%
ssrc --rate 44100 %wav% %ac3%1.wav
del %wav%
normalize -m 100 %ac3%1.wav
lame --nspsytune -V9 -mj -h --athtype 3 --resample 44.1 --ns-sfb21 2 -Z --nohist %ac3%1.wav %ac3%.mp3
del %ac3%1.wav
pause
--------------------------------------------------------------------

If you plan on making a DivX file that is bigger than 1-CD, you might want to consider the following:

Disable resampling to 44.1kHz

Increase the audio quality by changing some lame options:
lame --nspsytune -V1 -mj -h -b96 --lowpass 19.5 --athtype 3 --ns-sfb21 2 -Z --nohist %wav% %ac3%.mp3

This will result in a larger mp3 file, but with 2 CD's or more, your video quality won't suffer.  If you want a slightly smaller mp3, you can increase the -V setting, decrease the --lowpass setting, and/or remove the -b96 tag.  Something like this would still be excellent quality:

lame --nspsytune -V3 -mj -h --lowpass 18 --athtype 3 --ns-sfb21 2 -Z --nohist %wav% %ac3%.mp3

Sample 2-CD Batch File
---------------------------------------------------------------------
@echo off
set ac3=%1%
set wav=%ac3%.wav
azid -cnormal -L -3db %ac3% %wav%
del %ac3%
normalize -m 100 %wav%
lame --nspsytune -V3 -mj -h --lowpass 18 --athtype 3 --resample 48 --ns-sfb21 2 -Z --nohist %wav% %ac3%.mp3
del %wav%
pause
---------------------------------------------------------------------
As in all things, try it yourself and see what you think.  Good luck.

Settings:
Since the mp3 encoding of HeadAC3he is pretty limited, I like to stick with the 24-bit Wav out using float mode to normalize to 99%.  Then I use a batch file to accomplish Sample Rate Conversion (SSRC) and mp3 encoding (LAME)

Settings: BeSweet -core( -input audio.ac3 -output audio.mp3 ) -azid ( -c light -L -3db -g max ) -ssrc( --rate 44100 ) -lame( -v -V 9 -m j )

Lowering the -V value will increase the quality and filesize of the audio.

back