Encoding AVS Script with FFMPEG

[StainlessS]

Originally Posted by MysteryX

Edit: ah great... it doesn't support MT

What you mean by that ?

It dont work under MT ?

I dont use MT so would not know (life is just too short to be messing with dodgy stuff).

[Motenai Yoda]

Originally Posted by qyot27

The -map parameter is completely unnecessary when there's only one video and one audio input. Just saying.

Yep, but using muxed streams that one of them has coverart or some strange metadata can happens.

@Stains TWriteWAV how work?
ie Soundout write the file when the script is loaded

[StainlessS]

TWriteAVI/TWriteWAV:- http://forum.doom9.org/showthread.php?t=172837

Can write Video and/or Audio, and does not return immediately (as SoundOut does), it will only return when eg WAV is written and exists so that it
can be available in same script for second pass processing.

EDIT: TWriteWAV writes audio as an avs file is played, BUT,
ForceProcessAVI/ForecProcessWAV force reads the video/audio so that it is output without playing file.
Can use same to eg forcibly write some kind of metrics file (deshaker log or whatever), so it is available
for a second pass, within the same avs file.
Also, eg ForceProcessAVI scans an AVI file faster than VDub Video Analysis Pass does, about 70% of VDub taken time.

[MysteryX]

Originally Posted by StainlessS

What you mean by that ?

It dont work under MT ?

I dont use MT so would not know (life is just too short to be messing with dodgy stuff).

Since you don't support scanning back and forth, it makes sense it crashes. MT causes your plugin to process frames in a random order while opening various instances of your script at once.

Which means you're probably trying to open and write to the file 8 times at once while seeking back and forth.

[StainlessS]

Yes, I insisted on strict 1st to last frame processing, god knows how it would perform (old version) in MT, totally messed up AVI file I would guess.
As I said, dodgy stuff.

[MysteryX]

Originally Posted by StainlessS

Yes, I insisted on strict 1st to last frame processing, god knows how it would perform (old version) in MT, totally messed up AVI file I would guess.
As I said, dodgy stuff.

Blue screen followed by laptop batteries melting

[StainlessS]

Arh, not too bad then

[MysteryX]

If I have the WAV file and the encoded AAC file, is there a way to do a "diff" and measure the difference? Something that would easily tell if the quality is high enough to preserve the details?

[smok3]

@MysteryX: no, ABX test is what you want.

[MysteryX]

Originally Posted by smok3

@MysteryX: no, ABX test is what you want.

... and what about a AI software that does the ABX test for the user and then just displays the result?

[smok3]

Digital ear/AI = lossy audio codecs, so you would need something smarter than whats already there, so no.

[MysteryX]

It *was* working, until I started re-tuning the audio with
AssumeFps(FrameRate*.981818, true)

at which point I get my WAV file but Nero says:
ERROR: could not open AAC encoder

[Groucho2004]

Originally Posted by MysteryX

It *was* working, until I started re-tuning the audio with
AssumeFps(FrameRate*.981818, true)

at which point I get my WAV file but Nero says:
ERROR: could not open AAC encoder

It obviously doesn't support non-standard sample rates.

You have to change the pitch (only). If TimeStretch() does not satisfy your needs you might consider writing a plugin based on Subband Sinusoidal Modeling Synthesis (SBSMS).

[MysteryX]

Originally Posted by Groucho2004

You have to change the pitch (only). If TimeStretch() does not satisfy your needs you might consider writing a plugin based on Subband Sinusoidal Modeling Synthesis (SBSMS).

TimeStretch(pitch=100.0 * 0.981818)

This causes bad distortion. Not an option.

R = AudioRate
AssumeFps(FrameRate*.981818, true)
ReSampleAudio(R)

This works. It actually sounds better than if I don't resample(?)

As for SBSMS, I wanted to try the command-line conversion to see how it compares. The default download is a very basic GUI that really doesn't do much, and it has no documentation whatsoever. Calling the exe with "-help" doesn't do anything either expect launching the window.

The other download links with the source code don't contain any compiled library, and I still haven't seen any kind of documentation about how to use it from the command line. Would love to give it a try though.

[MysteryX]

OK I got SBSMS to work through the GUI for a test. Quality sounds good.

However
1. It is extremely slow to initialize and create the .sbsms file
2. The .sbsms file is huge 255MB off of a 55MB WAV file
3. I don't have enough fine-grained control over the pitch to get the desired tone

[MysteryX]

Rubberband might do.
http://rubberbandaudio.com/

It can be integrated into FFMPEG but I don't think it's in the default builds, unless I'm mistaken.

This command works
rubberband.exe -p -0.3176665363342977 in.wav out.wav