5.1 Ac3 -> 2.0 Ac3

[DPR]

Hi,

are there any recommendations for the 5.1 AC3 to 2.0 AC3 conversion ? I'm searching for the "optimal" BeSweet/Azid settings.

I've tried AC3Machine but the results don't satisfy me. The speech-volume (i.e. the mixed in center channel) is for example much too low.

Many thanks in advance for any recommendation !

[KpeX]

AC3 Machine is not software, it's just a GUI for BeSweet. BeSweet is the only free software that I know of that can encode AC3. Too-soft voices could be helped by gaining or using some DRC (dynamic range compression). Search, read the BeSweet help files, post your logfile if you need more help.

[DPR]

So what settings should I use if I want to stay as close as possible to the original AC3 track ? Or let me express it a bit differently: which settings can be used without having to think further about 'em ? Aren't there some settings which are generally "safe" (good output quality) for a 5.1 to 2.0 conversion ?

I've already read the BeSweet help but it couldn't really help me.

E.g. what does the "-d" option of the Azid plugin exactly do ? Does it effect the output data or is it just a kind of flag for the decoder ? And is it generally recommended to use this and the "-f" (rear-channel filtering) option together ?
And what's the difference between "surround" and "surround 2" ? (the "ProLogic" thread didn't clarify this point)
And last but not least: which dynamic compression (if any) is best for AC3 movie tracks ?

[KpeX]

1. Looking at your problem again, I think you're making it more complicated than it needs to be. Things like DRC and the such are not that important, since your destination format is also AC3, you can use DRC and other features when playing back the audio. Any decent filter such as AC3 filter or a good amp can do it on playback. The only thing you need to worry about is the 5.1 > 2.0 downmix, personally the only azid option I would use in this case would be -L -3db, but that's up to you. And remember not to ask what's best, it's too subjective. Do some testing and do what sounds good to your ears.

2. There's a -d azid option?

3. Have never needed -f (rear channel filtering) personally.

4. Again there is no best, when converting to MP3 or some other 2 channel format I prefer light or normal DRC.

5. Simple, 'surround' is for Dolby Pro Logic, 'surround 2' is for Dolby Pro Logic 2. They are different matrixed sound formats. Use whichever you prefer or is supported by your amp.

hth,

KpeX

[DPR]

Originally posted by KpeX
[...]And remember not to ask what's best, it's too subjective. Do some testing and do what sounds good to your ears.
"Testing" is exactly what I don't want to do - it takes too much time. At least I don't want to do a lot of encoding cycles until I gain the result which satisfies me.
Your argumentation about the DRC option persuaded me not to use it.

2. There's a -d azid option?
Yes. The option is indirectly described under the "-C" option. It sets the "output decode mode". But I don't know what this exactly means ...

3. Have never needed -f (rear channel filtering) personally.
Regarding the "-f" option the help file sounds as if this option should always be activated - especially for a 2.0 conversion.

4. Again there is no best, when converting to MP3 or some other 2 channel format I prefer light or normal DRC.
What a pity ...

5. Simple, 'surround' is for Dolby Pro Logic, 'surround 2' is for Dolby Pro Logic 2. They are different matrixed sound formats. Use whichever you prefer or is supported by your amp.

Many thanks for your help !

I've now tested almost all options but I still have the problem that the overall volume ist too low (nut just the speech loudness). Do you have any idea how I could fix this ?

The log-file looks like this:

BeSweet.exe -core( -input e:\source.ac3 -output e:\dest.ac3 -logfilea BeSweet.log ) -azid( -z1 -b1 -s surround2 -d 2/0 -g max -L -3db -C BSI -S BSI ) -ac3enc( -b 224 ) -iobuffer( -in 10240 -out 10240 )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : e:\source.ac3
[00:00:00:000] | Output: e:\dest.ac3
[00:00:00:000] | Floating-Point Process: No
[00:00:00:000] +---- I/O Buffer -----
[00:00:00:000] | Input Buffer : not supported, yet.
[00:00:00:000] | Output Buffer Size : 10240 bytes
[00:00:00:000] +-------- AZID -------
[00:00:00:000] | Input Channels Mode: 3/2, Bitrate: 448kbps
[00:00:00:000] | Output Stereo mode: Dolby Surround 2 compatible
[00:00:00:000] | Total Gain: 7.240dB, Compression: None
[00:00:00:000] | LFE levels: To LR -3.0dB, To LFE 0.0dB
[00:00:00:000] | Center mix level: 51.5db
[00:00:00:000] | Surround mix level: 51.5db
[00:00:00:000] | Dialog normalization: No
[00:00:00:000] | Rear channels filtering: No
[00:00:00:000] | Source Sample-Rate: 48.0KHz
[00:00:00:000] +------- AC3ENC ------
[00:00:00:000] | Bitrate method : CBR
[00:00:00:000] | AC3 bitrate : 224
[00:00:00:000] | Channels Mode : 2.0
[00:00:00:000] | Error Protection: Yes
[00:00:00:000] +---------------------
[01:24:31:968] Conversion Completed !
[01:24:31:968] Actual Avg. Bitrate : 224kbps
[00:15:17:000] <-- Transcoding Duration

[bleo]

The readme file in http://www.doom9.org/Soft21/Audio/azid-1.8.zip explains Azid's switches very well :)

From your logfile, the center and surround mix levels are a bit strange... Mine normally says BSI which should be -3.0dB. Try not setting -C and -S in your command line or set them to -3db...

[nuked]

Unless something cahnged in the last few days, DRC in ac3filter is not as simple as just turning it on. Ac3filter uses metadata for DRC. DVD ac3s have this metadata. In other words it expects the raw audio stream to be uncompressed and then reads the compression metadata from the ac3 to figure out how to compress it. You cannot simply encode a wave into ac3 and have ac3filter apply DRC on playback because ac3filter does not have it's own DRC algorithms/curves. You must encode the DRC metadata into the ac3. ac3enc for besweet cannot do this. softencode does quite easily(don't forget to check dialog norm in azid when making the waves though if you want the levels right for encoding the DRC), but that may not be the solution for someone who wants a one touch no hassle transcode.
This DRC option is the one and essentially only reason I am also considering doing ac35.1 to ac3 2.0 transcodes (that and I wonder if using essentially the same parameterization to reincode may reduce loss some but it's not clear to me how big a deal that is). There are other filters though that can do DRC on playback for mp3 stuff (not compatible with ac3 filter) I'm just not sure what I think of them yet. I know for instance that TFM doesn't have a documented required dialogue normalization level so that bugs me a little even if it is working off a little diferent philosophy.

[nuked]

-ac3enc has had a well known bug regarding low volume. Supposedly it's bit rate dependant even if I remember right.

[DPR]

Originally posted by bleo
[...]From your logfile, the center and surround mix levels are a bit strange... Mine normally says BSI which should be -3.0dB. Try not setting -C and -S in your command line or set them to -3db...
Well "-C BSI" and "-S BSI" should be correct since it should be the default switches. Actually - I don't know it for sure :(

Regarding the ac3enc low volume bug, I've already seen this in the BeSweet FAQ. But I thought this bug would have already been fixed because it's well known for more than one year :confused:
However, I'll give "softencode" a try.

BTW it seems that the source code for BeSweet/ac3enc/... on http://dspguru.notrace.dk are not up to date, right ???