BeHappy - AviSynth based audio transcoding tool (UPD 19-07-2006)

[jones1913]

Originally Posted by jones1913

...
https://www.mediafire.com/file/1m4hb...0.3.1.zip/file

I just noticed that mediafire flagged my upload as "virus".
So I dropped the zip file to virustotal, and some encoders are detected as virus or trojan or whatnot (by max 4 search engines out of 69).
See here:
https://www.virustotal.com/gui/file/...0499/relations

To compare I have also tested with encoders from Mulders site:
https://sourceforge.net/projects/mul.../Opus%20Tools/
and the results are even more crazy:
https://www.virustotal.com/gui/file/...46d9/relations

I am very confident that Mulders compiles contain no viruses and these are all false flags.
So I guess it is the same with the encoders in my upload. They are mostly from rarewares.org .
I think everyone must decide if they trust downloaded programs.

If someone has suggestions for alternative download sources than rarewares.org ...

[LigH]

I am not surprised that some Antivirus engines report false alarms for assembler optimized SSE builds of encoders, as provided by RareWares.

[john33]

I am confident that anything from Rarewares will be a false positive. I am afraid that virustotal is in many cases more of a hindrance than a help, their algorithms are not as smart as they would have people believe.

[tebasuna51]

The Upmix DSP's can't work with Avs+ and SoxFilter.dll, maybe in the future...

Until that the Avs+ users can work with the commented soxfilter usage in d_UpMix.ext:

front = a#.soxfilter("filter 20-20000")
back = a#.soxfilter("filter 100-7000")
lfe = cc.Amplify(0.0)#ConvertToMono(a).soxfilter("lowpass 120","vol -0.5")

The filters are, for me, unnecesaries and the lfe is much better empty.

[tebasuna51]

@jones1913
I can't see the new sources in https://github.com/jones1913/BeHappy

BTW I proppose a change:

Line 338 in old Encoder.cs:
int[] defmask = {0, 4, 3, 7, 51, 55, 63, 319, 1599, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

Num  Mask                                                            Mask
cha  old   old maskchannel defaults      new maskchannel defaults     new   Name
--- ----  ----------------------------  ---------------------------  ----  ------
 1     4  FC                            FC                              4  mono
 2     3  FL FR                         FL FR                           3  stereo
 3     7  FL FR FC                      FL FR    LFE                   11  2.1
 4    51  FL FR        BL BR  (quadro)  FL FR FC           BC         263  4.0 (first surround)
 5    55  FL FR FC     BL BR            FL FR FC              SL SR  1543  5.0
 6    63  FL FR FC LFE BL BR            FL FR FC LFE          SL SR  1551  5.1
 7   319  FL FR FC LFE BL BR BC         FL FR FC LFE       BC SL SR  1807  6.1
 8  1599  FL FR FC LFE BL BR    SL SR   FL FR FC LFE BL BR    SL SR  1599  7.1

Then must be:
int[] defmask = {0, 4, 3, 11, 263, 1543, 1551, 1807, 1599, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

Now the defaults are the same than ffmpeg, and with the pair SL/SR for surround, instead BL/BR like can see here

Question: decoders like BestAudioSource or FFAudioSource export a runtime variable with the current dwChannelMask.
How can BeHappy read it to assign the correct maskchannel without need put it manually?

[jones1913]

I can't see the new sources in https://github.com/jones1913/BeHappy

It's up now.

Question: decoders like BestAudioSource or FFAudioSource export a runtime variable with the current dwChannelMask.
How can BeHappy read it to assign the correct maskchannel without need put it manually?

Good question. I guess this has to be wired up in Avisynthwrapper.dll .
To start small: how would I retrieve this variable in an ordinary script?

[tebasuna51]

Originally Posted by jones1913

It's up now.

Thanks.

how would I retrieve this variable in an ordinary script?

I don't know, I ask also in Avisynth usage thread

There are some news https://forum.doom9.org/showthread.p...68#post1984568

[pinterf]

Originally Posted by tebasuna51

@jones1913
I can't see the new sources in https://github.com/jones1913/BeHappy

BTW I proppose a change:

Line 338 in old Encoder.cs:
int[] defmask = {0, 4, 3, 7, 51, 55, 63, 319, 1599, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

Num  Mask                                                            Mask
cha  old   old maskchannel defaults      new maskchannel defaults     new   Name
--- ----  ----------------------------  ---------------------------  ----  ------
 1     4  FC                            FC                              4  mono
 2     3  FL FR                         FL FR                           3  stereo
 3     7  FL FR FC                      FL FR    LFE                   11  2.1
 4    51  FL FR        BL BR  (quadro)  FL FR FC           BC         263  4.0 (first surround)
 5    55  FL FR FC     BL BR            FL FR FC              SL SR  1543  5.0
 6    63  FL FR FC LFE BL BR            FL FR FC LFE          SL SR  1551  5.1
 7   319  FL FR FC LFE BL BR BC         FL FR FC LFE       BC SL SR  1807  6.1
 8  1599  FL FR FC LFE BL BR    SL SR   FL FR FC LFE BL BR    SL SR  1599  7.1

Then must be:
int[] defmask = {0, 4, 3, 11, 263, 1543, 1551, 1807, 1599, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

Now the defaults are the same than ffmpeg, and with the pair SL/SR for surround, instead BL/BR like can see here

Question: decoders like BestAudioSource or FFAudioSource export a runtime variable with the current dwChannelMask.
How can BeHappy read it to assign the correct maskchannel without need put it manually?

Had a look into av_get_default_channel_layout which iterates over channel_layout_map[] and returns the first entry with matching channel count, I see it differently.

Based on this:
https://ffmpeg.org/doxygen/trunk/cha...8c_source.html
and
https://ffmpeg.org/doxygen/trunk/cha...8h_source.html

The default five and six-channel setting is 5.0 BACK and 5.1 BACK, so they shouldn't be changed.
Changing the 3 and 4-channel defaults to 2.1 and 4.0 is O.K.

[tebasuna51]

Originally Posted by pinterf

https://ffmpeg.org/doxygen/trunk/cha...8c_source.html
and
https://ffmpeg.org/doxygen/trunk/cha...8h_source.html

The default five and six-channel setting is 5.0 BACK and 5.1 BACK, so they shouldn't be changed.
Changing the 3 and 4-channel defaults to 2.1 and 4.0 is O.K.

And 6.1 (7 channels) must be changed also.

Well, and ffmpeg must change 5.0 BACK and 5.1 BACK defaults.

In old times SL/SR (Side) was at 90º and BL/BR at 135º (Back like cuadro) and the AC3/DTS sources 5.1 with Surround channels (Ls/Rs) at 120º was decoded to BL/BR (120º near 135º than 90º)

But now the preference seems assign 110º to Side (SL/SR) and Surround (Ls/Rs)

And ffmpeg is coherent with that:
DTS 5.0 decoded like 5.0(side)
DTS 5.1 decoded like 5.1(side)
AC3 5.0 decoded like 5.0(side)
AC3 5.1 decoded like 5.1(side)
...
What source must be decoded to BL/BR pair?

[pinterf]

Originally Posted by tebasuna51

And 6.1 (7 channels) must be changed also.

Well, and ffmpeg must change 5.0 BACK and 5.1 BACK defaults.

In old times SL/SR (Side) was at 90º and BL/BR at 135º (Back like cuadro) and the AC3/DTS sources 5.1 with Surround channels (Ls/Rs) at 120º was decoded to BL/BR (120º near 135º than 90º)

But now the preference seems assign 110º to Side (SL/SR) and Surround (Ls/Rs)

And ffmpeg is coherent with that:
DTS 5.0 decoded like 5.0(side)
DTS 5.1 decoded like 5.1(side)
AC3 5.0 decoded like 5.0(side)
AC3 5.1 decoded like 5.1(side)
...
What source must be decoded to BL/BR pair?

Then they at ffmpeg project know it well, just keep the above mentioned old defaults for compatibility reasons?

[tebasuna51]

Originally Posted by pinterf

Then they at ffmpeg project know it well, just keep the above mentioned old defaults for compatibility reasons?

I don't think so, but no problem for me (I am one of the last defender of decode Ls/Rs to BL/BR, but with eac3to I changed my mind and recommend the pair SL/SR).

BTW the BeHappy defaults was created by me, and I recommend actualize them because all sources 5.0/5.1 are decoded to SL/SR, and encoders accept better the SL/SR pair to create 5.0/5.1 files.

All AVR's 5.1 play SL/SR or BL/BR with the Ls/Rs speakers without problems.

[jones1913]

Originally Posted by tebasuna51

@jones1913
BTW I proppose a change:

Line 338 in old Encoder.cs:
int[] defmask = {0, 4, 3, 7, 51, 55, 63, 319, 1599, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

Num  Mask                                                            Mask
cha  old   old maskchannel defaults      new maskchannel defaults     new   Name
--- ----  ----------------------------  ---------------------------  ----  ------
 1     4  FC                            FC                              4  mono
 2     3  FL FR                         FL FR                           3  stereo
 3     7  FL FR FC                      FL FR    LFE                   11  2.1
 4    51  FL FR        BL BR  (quadro)  FL FR FC           BC         263  4.0 (first surround)
 5    55  FL FR FC     BL BR            FL FR FC              SL SR  1543  5.0
 6    63  FL FR FC LFE BL BR            FL FR FC LFE          SL SR  1551  5.1
 7   319  FL FR FC LFE BL BR BC         FL FR FC LFE       BC SL SR  1807  6.1
 8  1599  FL FR FC LFE BL BR    SL SR   FL FR FC LFE BL BR    SL SR  1599  7.1

Then must be:
int[] defmask = {0, 4, 3, 11, 263, 1543, 1551, 1807, 1599, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

Now the defaults are the same than ffmpeg, and with the pair SL/SR for surround, instead BL/BR like can see here

Its done.
Compiled bins can be found on github in "bin" folder: https://github.com/jones1913/BeHappy/tree/master/bin

[jones1913]

Originally Posted by tebasuna51

The Upmix DSP's can't work with Avs+ and SoxFilter.dll, maybe in the future...

Until that the Avs+ users can work with the commented soxfilter usage in d_UpMix.ext:

front = a#.soxfilter("filter 20-20000")
back = a#.soxfilter("filter 100-7000")
lfe = cc.Amplify(0.0)#ConvertToMono(a).soxfilter("lowpass 120","vol -0.5")

The filters are, for me, unnecesaries and the lfe is much better empty.

Soxfilter seems to work now with latest avs+ test version, should we still use your fix?

And 2 more questions:
- Audiolimiter.dll (for Down Mix) is 32bit only, so we must disable it in 64bit BeHappy build or is there an alternative?
- Do we still need BassAudioSource and NicAudioSource? Have they any use case that is not covered from our ffmpeg based filters?

[tebasuna51]

Originally Posted by jones1913

Soxfilter seems to work now with latest avs+ test version, should we still use your fix?

It is a test version, we can wait until we know how use it to include the maskchannel. BTW can be a user decision because:

- front = a#.soxfilter("filter 20-20000")

Who need that filter? For what a limit here? The encoders have filters to limit the banwidth based in bitrate requested.

- back = a#.soxfilter("filter 100-7000")

To save bitrate? Maybe this can have some sense.

- lfe = cc.Amplify(0.0)#ConvertToMono(a).soxfilter("lowpass 120","vol -0.5")

Never. Please don't mistake the LowFrequencyEfects channel with the subwoffer in an AVR.
The AVR, and also simple audio equipment 2.1, extract the low frequencies from ALL channels and send them to the subwoofer.
Create a unexistent LFE only can produce problems with differences in phase or cut frequencies betwen that filter and the AVR's filters configured more accurate with the speakers attached to them.
Please leave the lfe empty, or use a 5.0, when upmix 2.0 to 5.1

- Audiolimiter.dll (for Down Mix) is 32bit only, so we must disable it in 64bit BeHappy build or is there an alternative?

There are an alternative using the compand sox filter.
See the ffmpeg downmix, and in MeGui 64 thread

I will test it.

- Do we still need BassAudioSource and NicAudioSource? Have they any use case that is not covered from our ffmpeg based filters?

NicAudioSource is not necesary anymore after the other decoders support the drc=0.

BassAudioSource have some special decoders, for instance .mid, not supported in standard audio decoders. Maybe it is not used frequently but it is a option.