BeSweet 5.1 AC3 -> 6 Mono Wav -> 5.1 AC3

[jweathers7]

I am in the processing of re-editing Attack of the Clones so that I can reinsert the deleted scenes into the film. I have had great success at actually editing the MPEG2 video and creating smooth transitions using Womble MPEG Video Wizard. However, because MVW doesn't support creating AC3 transitions I'm doing the audio part by hand.

I have everything working except for the following: when I use BeSweet to convert from 5.1 AC3 to 6 Mono WAVs, do my editing work, and then convert those 6 Mono WAVs back into 5.1 AC3, the file is noticeably different in volume.

From what I've read on these forums, I'm guessing that it has to do with normalization.

Is there a way to use BeSweet so that it doesn't change anything about the sound via remixing, normalization, dynamic compression, etc. in either direction so that if I were to take a given 5.1 AC3 turn it into 6 Mono WAVs and then create a 5.1 AC3 from them, the volume and sound would be the same except for any loss in quality from re-encoding?

I have pasted the text from my log file below:

BeSweet v1.5b31 by DSPguru.
--------------------------
Using azid.dll v1.9 (b922) by Midas (midas@egon.gyaloglo.hu).

Logging start : 12/06/05 , 11:17:01.

BeSweet -core( -input 1stClips04.ac3 -output 1stClips04 -6ch -logfile test.log )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : 1stClips04.ac3
[00:00:00:000] | Output: FL, FR, SL, SR, C, LFE
[00:00:00:000] | Floating-Point Process: No
[00:00:00:000] +-------- AZID -------
[00:00:00:000] | Input Channels Mode: 3/2, Bitrate: 448kbps
[00:00:00:000] | Total Gain: 0.000dB, Compression: None
[00:00:00:000] | LFE levels: To LR -INF, To LFE 0.0dB
[00:00:00:000] | Center mix level: BSI
[00:00:00:000] | Surround mix level: BSI
[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] +---------------------
[00:00:03:008] Conversion Completed !
[00:00:00:000] <-- Transcoding Duration

Logging ends : 12/06/05 , 11:17:01.


BeSweet v1.5b31 by DSPguru.
--------------------------
Using AC3enc.dll v1.20 (Feb 18 2004) by Fabrice Bellard (http://ffmpeg.org).

Logging start : 12/06/05 , 11:17:01.

BeSweet -core( -input 1stClips.mux -output 1stClips.ac3 -logfilea test.log ) -ac3enc( -b 448 -6ch )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : 1stClips.mux
[00:00:00:000] | Output: 1stClips.ac3
[00:00:00:000] | Floating-Point Process: 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 : 448
[00:00:00:000] | Channels Mode : 5.1
[00:00:00:000] | Error Protection: Yes
[00:00:00:000] +---------------------
[00:00:03:008] Conversion Completed !
[00:00:03:008] Actual Avg. Bitrate : 448kbps
[00:00:01:000] <-- Transcoding Duration

Logging ends : 12/06/05 , 11:17:02.

[tebasuna51]

This is a know problem.
To encode ac3 use a commercial encoder, the old Sonic Foundry Soft Encode, or if you need a free encoder (all based in the same library) use HeadAC3he v0.24-a13 by Dark Avenger 15.12.2004, and apply a Global Gain of 6 db.

[jweathers7]

Thanks for the reply.

I did notice something that seemed similar to my problem in another thread, but I wanted to make sure as my situation seemed a little different than the norm in that the other thread was talking about re-encoding directly at different bitrates rather than going through an intermediate 6 WAV phase and making some edits.

I have tried Soft Encode which seems to work great if I use a Dialog Normalization of -31dB, however, I had nearly given up on it last night after it suddenly began crashing everytime I used it. I found out through a post on this forum that clearing its registry appears to give life to the program when it gets into this state.

Thus, I could use Soft Encode except that I just found another problem with it: I used it to decode my 5.1 AC3 file to a 5.1 WAV which I then split into 6 Mono WAVs with BeSweet. I then took these unaltered WAVs and used Soft Encode to create a new 5.1 AC3. I then decoded this AC3 back into WAVs and found that the result appears to have a tiny bit of silence at the beginning (and a corresponding bit of truncation at the end) resulting in a file that is out of sync with the original which will not be good for a drop in replacement.

Is this a known issue with Soft Encode or just the consequence of some setting that I need to adjust?

What's magical about the Global Gain of 6dB when using HeadAC3he?

[tebasuna51]

Thus, I could use Soft Encode except that I just found another problem with it: I used it to decode my 5.1 AC3 file to a 5.1 WAV ... and found that the result appears to have a tiny bit of silence at the beginning
This is typical in encoders/decoders, at least 1 frame (32 ms for 48 KHz) may be inserted at the beginning. I don't use SoftEncode for decode after the help:

"Dolby Digital (decode) (*.ac3)
By opening a Dolby Digital file decode mode, Soft Encode will decode the file into its composite tracks. This option is mostly for reference as it is not recommended that you reencode a decoded Dolby Digital file."

I use BeSweet-azid directly to 6 mono wav.
What's magical about the Global Gain of 6dB when using HeadAC3he?
After many test the max volume obtained with ac3enc.dll (not only with HeadAC3he) is -6 db (50%), then I use Global Gain 6 db and obtain 0 db output without clipping. But you must use SoftEncode instead ac3enc.dll if possible.

[jweathers7]

This is typical in encoders/decoders, at least 1 frame (32 ms for 48 KHz) may be inserted at the beginning. I don't use SoftEncode for decode after the help

Is there any way to remove that delay? Because the file that I am re-encoding is not the full soundtrack - rather it is a tiny transition clip created from using BeSplit on the full soundtrack AC3 file. My intention is to then join the final resulting transition clip back with the other clips to form the full soundtrack again in AC3 with most of the soundtrack remaining unmodified and unreencoded. If there is a delay, it will create a noticeable gap in the soundtrack near each transition.

After many test the max volume obtained with ac3enc.dll (not only with HeadAC3he) is -6 db (50%), then I use Global Gain 6 db and obtain 0 db output without clipping. But you must use SoftEncode instead ac3enc.dll if possible.


ac3enc also appears to do even more than I thought to the sounds. It seems that in addition to creating an AC3 file that is quieter than the original, it appears to also have a delay (much much smaller than the one in SoftEncode) and strangely enough it also appears to be inverting the sound. I decomposed the resulting AC3 file and then used Audacity to compare the new center channel WAV with the original center channel WAV. After applying a normalization to each file and inverting one of them, the two were virtually identical except for the tiniest of delays.

[tebasuna51]

Well you can see the problems of this kind of edition: delays, different volume, possible click in join, different parameters in BSI section of ac3 header along the whole track, ...

I think is better make the edition in wav files and after encode the whole track.