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

[dimzon]

@tebasuna51
Great!
Unfortunally I'm not DSP expert.
Why not to use Normalize() BEFORE Dynamic Range Compression?

[tebasuna51]

Originally Posted by dimzon

Why not to use Normalize() BEFORE Dynamic Range Compression?

The decoder output (NicAudio) have the full dynamic range and can have samples with 0 dB, then the Normalize function do nothing with this signal.

We need, also, apply attenuation over the decoder output to compensate DialNorm (if distinct of -31 dB) for input correctly in DRC curves.

The solution is make alternate curves with normalized output. I'm work about this.

[dimzon]

Originally Posted by tebasuna51

The decoder output (NicAudio) have the full dynamic range and can have samples with 0 dB, then the Normalize function do nothing with this signal.

Do not forget - we can use DRC on NON-AC3 input

Originally Posted by tebasuna51

The solution is make alternate curves with normalized output. I'm work about this.

Yeah, it's the best

[dimzon]

fresh beta is out

[Dayvon]

Originally Posted by dimzon

2 All
Hi! You can help BeHappy project!
I need:
BeHappy logo & icons

I'd like to help with this Dimzon. You have any preferences or specifics in mind?

[NorthPole]

Made some changes to the upmix extension I posted earlier...

Removed the option to use the avisynth equilizer because it is extremely slow in execution.

I added a second upix option that tries to remove most of the dialog from the left, right and surround speakers and sends it just to the center speaker.
Also, inverted the LFE, attenuated and delayed the surround speakers as before.

<?xml version="1.0"?>
<BeHappy.Extension xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://workspaces.gotdotnet.com/behappy">
  <AudioDSP UniqueID="9579E57B-2D27-4583-99A4-921718E25B41">
    <Plugin>
      <MultiOptionDSP Type="BeHappy.Extensions.MultiOptionDSP, BeHappy">
        <TitleFormatString>{0}</TitleFormatString>
        <ScriptPrologue>
	# Store clip in variable
	Stereo_{2} = convertaudiotofloat(last)
        </ScriptPrologue>
        <Option>
          <Name>Upmix with equalizer adjustments</Name>
          <Value>
	# Sox filter using frequency selection, 20ms surround delay and attenuation
	Front_{2} = mixaudio(Stereo_{2}.soxfilter("filter 0-600"),mixaudio(Stereo_{2}.soxfilter("filter 600-1200"),Stereo_{2}.soxfilter("filter 1200-7000"),0.45,0.25),0.50,1)
	Back_{2} = mixaudio(Stereo_{2}.soxfilter("filter 0-600"),mixaudio(Stereo_{2}.soxfilter("filter 600-1200"),Stereo_{2}.soxfilter("filter 1200-7000"),0.35,0.15),0.40,1)
	fl_{2} = GetLeftChannel(Front_{2})
	fr_{2} = GetRightChannel(Front_{2})
	cc_{2} = ConvertToMono(stereo_{2}).SoxFilter("filter 625-24000")
	lfe_{2} = ConvertToMono(stereo_{2}).SoxFilter("lowpass 100","vol -0.5")
	sl_{2} = GetLeftChannel(Back_{2})
	sr_{2} = GetRightChannel(Back_{2})
	sl_{2} = DelayAudio(sl_{2},0.02)
	sr_{2} = DelayAudio(sr_{2},0.02)
          </Value>
        </Option>
        <Option>
          <Name>Upmix with center channel dialog</Name>
          <Value>
	# channel subtraction with sox filter, 20ms surround delay and attenuation
	left_{2} = stereo_{2}.GetLeftChannel()
	right_{2} = stereo_{2}.GetRightChannel()
	fl_{2} = mixaudio(left_{2}.soxfilter("filter 0-24000"),right_{2}.soxfilter("filter 0-24000"),0.6,-0.5)
	fr_{2} = mixaudio(right_{2}.soxfilter("filter 0-24000"),left_{2}.soxfilter("filter 0-24000"),0.6,-0.5)
	cc_{2} = ConvertToMono(stereo_{2}).SoxFilter("filter 625-24000")
	lfe_{2} = ConvertToMono(stereo_{2}).SoxFilter("lowpass 100","vol -0.5")
	sl_{2} = mixaudio(left_{2}.soxfilter("filter 0-24000"),right_{2}.soxfilter("filter 0-24000"),0.5,-0.4)
	sr_{2} = mixaudio(right_{2}.soxfilter("filter 0-24000"),left_{2}.soxfilter("filter 0-24000"),0.5,-0.4)
	sl_{2} = DelayAudio(sl_{2},0.02)
	sr_{2} = DelayAudio(sr_{2},0.02)
          </Value>
        </Option>
        <ScriptEpilogue>
	# Return result
	MergeChannels( fl_{2}, fr_{2}, cc_{2}, lfe_{2}, sl_{2}, sr_{2})
	ConvertAudioTo16Bit()
        </ScriptEpilogue>
      </MultiOptionDSP>
    </Plugin>
  </AudioDSP>
</BeHappy.Extension>

@tebasuna51

Nice dynamic range compression extension. I agree that it would be nice to be able to use negative values in tweak.

@Dimzon

I tried this code but it executed extremely slowly

left_{2} = last.GetLeftChannel()
right_{2} = last.GetRightChannel()
fl_{2} = mixaudio(left_{2},right_{2},0.6,-0.5)
fr_{2} = mixaudio(right_{2},left_{2},0.6,-0.5)
cc_{2} = ConvertToMono(stereo_{2}).SoxFilter("filter 625-24000")
lfe_{2} = ConvertToMono(stereo_{2}).SoxFilter("lowpass 100","vol -0.5")
sl_{2} = mixaudio(left_{2},right_{2},0.5,-0.4)
sr_{2} = mixaudio(right_{2},left_{2},0.5,-0.4)
sl_{2} = DelayAudio(sl_{2},0.02)
sr_{2} = DelayAudio(sr_{2},0.02)

[dimzon]

Originally Posted by Dayvon

I'd like to help with this Dimzon. You have any preferences or specifics in mind?

Just make it eyecandy

[Dayvon]

Originally Posted by dimzon

Just make it eyecandy

Is this creative liberty you've given me?

[dimzon]

Originally Posted by Dayvon

Is this creative liberty you've given me?

Yes.
By the way - since BeHappy is avisynth based software maybe we can use portion of avisynth logo in it (just proposal, feel free not to do it)

[Dayvon]

Originally Posted by dimzon

Yes.
By the way - since BeHappy is avisynth based software maybe we can use portion of avisynth logo in it (just proposal, feel free not to do it)

I'll look into it!

[tebasuna51]

@Dimzon
I get incorrect wav output (BlockAlign incorrect) with your last release (and also incorrect aac encodes).

Maybe there are a bug in Encoder.cs (now there are 2 writeHeader routines)

CORRECT:
private void writeHeader(Stream target )  {
...
	target.Write(BitConverter.GetBytes(m_wavHeader.nBlockAlign),0,2);
	target.Write(BitConverter.GetBytes(m_wavHeader.wBitsPerSample),0,2);
...
}

INCORRECT:
private void writeHeader(Stream target, AviSynthClip x )  {
...
	target.Write(BitConverter.GetBytes(x.BitsPerSample/8),0,2);  // ?????
	target.Write(BitConverter.GetBytes(x.BitsPerSample),0,2);
...
}
BUG: BlockAlign = ChannelsCount * BitsPerSample/8

[TheBashar]

Can someone please give me a sanity check with this scheme? I like to cut out the opening and ending credits of my TV series encodes with something like Trim(90,6000)+Trim(9000,61000) for the video. Cutting, merging, and muxing the audio is ehh okay, but I've always struggled with some synch issues.

I'd like to try avisynth processing via BeHappy to see if I can use the same Trim statements to manipulate the audio as I do with the video so maybe I can avoid or eliminate the synch issues.

I set up BeHappy to do an ac3 -> aac conversion and exported the avs to look at. I was wondering if this would work.

audio = DirectShowSource("audio.ac3")
audio = EnsureVBRMP3Sync(audio)
# Above same as BeHappy Generated:
video = MPEG2Source("video.d2v")
AudioDeubEx(video, audio)
# Use same trim arguments as in video encoding script:
Trim(90,6000)+Trim(9000,61000)
AudioDubEx(Tone(), last)
# Below same as BeHappy Generated:
AudioDubEx(BlankClip(length=Int(1000*AudioLengthF(last)/Audiorate(last)), width=320,height=32,pixel_type="RGB24",fps=1000), last)
Normalize...
Encode...
Kill Video again...

So in essence I'd bring in the audio, apply the vbrsynch black magic, dub it to the actual video clip, trim with the video trim settings, kill the real video, dub it to the 1000 fps blank video, normalize, encode, and kill the blank video.

Is this a sane way to go about this task?

Thanks!

[tebasuna51]

You don't need:

AudioDubEx(BlankClip(length=Int(1000*AudioLengthF(last)/Audiorate(last)), width=320,height=32,pixel_type="RGB24",fps=1000), last)
...
Kill Video again...

Because this AudioDubEx sentence is only to use Trim in ms. (fps=1000), and Trim is used before based in video fps.

If you want maintain the ac3 without re-encode there are other method to trim the audio with DelayCut:

You want a first segment like this: Trim(90,6000)
If fps is 25 (replace 25 with your appropriate fps) you want a segment between 90/25 = 3.6 sec and 6000/25 = 240 sec. Then open DelayCut and Cut your original ac3 between Start = 3600 and End = 240000 ms.

For the second segment Cut between 360000 ms (9000/25 sec) and 2440000 ms. (61000/25 sec).

After join the two segments with:

> copy /B first.ac3 + second.ac3 full.ac3

[dimzon]

Originally Posted by tebasuna51

BUG: BlockAlign = ChannelsCount * BitsPerSample/8

Are you 100% shure? (this will affect MeGUI too)

[tebasuna51]

Originally Posted by dimzon

Are you 100% shure? (this will affect MeGUI too)

With this kind of wavheader (PCM 16 or 32 bits int) always:

BlockAlign = ChannelsCount * BitsPerSample/8

[dimzon]

Originally Posted by tebasuna51

With this kind of wavheader (PCM 16 or 32 bits int) always:

BlockAlign = ChannelsCount * BitsPerSample/8

Thanx a lot, will be fixed soon

[dimzon]

http://forum.doom9.org/showthread.php?t=83752 - additional info about 5.1 upmix

[3dsnar]

Hi Dimzon. I have prepared the pseudosurround DLL for ya

I cannot attach it to this post, cause it is to big.
Please drop me an email: programmers AT aud-x.com
and I will reply and send you the attachement.

(ie. the DLL, and the complete project. Maybe it will be useful)

After downloading, please let me know, and I will remove it.
I will be happy to answer all your questions (if you have any).

[shon3i]

@Dimzon how to boost 6ch audio aac like hibridgain in besweet or autogk with mp3 ,with behappy

[grokwik]

Dimzon, thanks a lot for this awsome tool !
I can now transcode audio from multiple avi quickly. Every thing works fine.
I have noticed that nero aac encode works with or without neroipp.dll without any difference. Is this dll really necessary ?