Transcode an ac3 2.0 dpl

[dodone]

I'm using besweet to transcode an ac3 2.0 surround to another one in order to make a time stretch, but when i play the resulting file i get a "2/0 - stereo" in the ac3 filter panel (on "bitstream information") while the original one plays as "Dolby surround".

This is the command line i'm using:
"C:\BeSweet\BeSweet.exe" -core( -input "i:\eng_23.ac3" -output "c:\eng_25.ac3" ) -azid( -s dpl ) -ota( -r 23976 25000 ) -ac3enc( -b 320 )

[tebasuna51]

I'm using besweet to transcode an ac3 2.0 surround to another one in order to make a time stretch, but when i play the resulting file i get a "2/0 - stereo" in the ac3 filter panel (on "bitstream information") while the original one plays as "Dolby surround".
Ac3enc and other free encoders can't set the flag "Dolby surround" in BSI ("BitStream Information"), only the new Aften can do this.
Use instead your command line:
BeSweet -core( -input i:\eng_23.ac3 -output c:\eng_25.wav -2ch ) -soundtouch( -r 23976 25000 )
Aften -b 192 -dsur 2 c:\eng_25.wav c:\eng_25.ac3
-azid( -s dpl ) not necessary, used to downmix 5.1 to 2.0 dpl.
-ac3enc( -b 320 ) better aften in output volume and options.
-soundtouch( -r 23976 25000 ) better than ota preserving the pitch.
-b 192 (320 is to much for 2.0, 192 Kb/s is enough).
-dsur 2 (set the flag 'Dolby surround encoded').

If you don't want intermediate wav file you can use BeHappy.

Warning, the time stretch process can destroy the dpl info.
I make a simple test with a test-channel and seems work ok, but I can't generalize.

[Skelsgard]

-azid( -s dpl ) means only that when downmixing to 2ch, a DPL matrix will be used. It doesn´t mean that the output file (AC3 in this case) will be flagged as a DPL surround.
If u want to have a "DPL surround" flagged AC3, u can use Aften. The downside is that u will have to do "Besweet AC3 --> stretched WAV" --> "aften for AC3" (an intermediate WAV will be created in the HD).
Cheers.

[3dsnar]

BTW. time stretching may significantly affect phase -> this
will affect DPLII downmix.
Maybe, decoding the signal to 5.0, applying time stretching,
and downmixing it back to 2 channels will be better...

[Rockaria]

Some check points that may result in better satisfaction & understanding :

a. Do we actually need the 'dpl-encoded' flag any longer?
It is used only in the ac3 2.0 format but many existing decoders do not automatically set the decoding mode by the tag or stream inside. We manually set the dpl(II) decoding mode of the decoders on 2ch (decoded) pcm. And does ac3filter use it for decoding or matrix upmixing?

b. What is the besweet internal dsp chain (order) ?
If the dpl(ii) mixing is the last step before the final encoding, like in many approaches, the time stretching won't affect the DPL(II) stream phase.
Is there any way to manually set the dsp chain(order) in besweet, like foobar2k or ffdshow?

c. Which tool has the better & flexible DPL(II) encoding emulation?
Depending on the input stream images(90deg. rear phase shifts, invert, proper gain & coef ratio,,) expected by the DPL(II) decoders, it results in different channel seperations(unbalanced, un-normalized, panned,,). Be cautious of the invert-only & uncertain non-dolby-way 'DPL(II)' mixers.

d. What are the constraints of the existing encoding environments and which is best for you?
We may have to focus on the flexibility(no hidden self-limitations,,), adaptability(i.e. easy to incorporate the cli encoders),, and so on.
In addition to the above mentioned environments, I added foobar2k & avisynth as my first option long ago.

[dodone]

Thank you all for replying :o
I'm using the wav method, now the ac3 works fine :) :)

[Rockaria]

.. now the ac3 works fine :) :)Which means, by the initial inquiry, the ac3Filter correctly DISPLAYS the DS mode BSI info set by Aften?:)