Audio Tracks & Apple TV 2

[tannebil]

Most of my TiVo programs have an AC-3 audio track but end up with an AAC (stereo) audio track after I transcode them to H.264 using a modified version of the iPhone profile. That's good for my iPhone/iPad (because they only grok stereo) but bad for my Apple TV 2 (which groks both stereo and AC-3). Apple handles the issue by including both audio tracks in the file but VRD doesn't have the capability to create two audio tracks out of one.

Handbrake can do it and, since my VAP workflow uses Handbrake for 720p files, my 720p files are getting two tracks and play AC-3 on the ATV 2 and stereo on my iPhone/iPad. My 1080i files are done in VRD so they end up with the lowest common denominator (stereo).

The issue is discussed here:

http://www.videoredo.net/msgBoard/sh...ad.php?t=29620

One approach suggested by hicmc is to use VRD to extract the original AC-3 audio and mix and group it back at the end using a 3rd party muxer (mp4box). Could it be done under the control of VAP?

It would help if VAP could at least extract the AC-3 audio track so I could remux it in post processing.

Thanks, Bill

[dlflannery]

Not sure whether the VRD COM object can extract the audio track. Even if it can, It seems pretty messy to iincorporate that into VAP. How about doing it as the first step in your VAPpostProcess script using ffmpeg or perhaps another command line utility? Well, probably not, since I don't believe ffmpeg handles .tivo inputs(?). So maybe your script has to run TiVoDecode to get an MPEG2-PS video that can be input into ffmpeg?

Can't imagine any reason mp4box can't be run by the VAPpostProcess script. I assume you're NOT cutting commercials since the entire scheme would be impossible if the extra audio track isn't cut like the video.

As a minimum you're going to be doing a QSF in VRD/VAP and then muxing in an audio track extracted from the input file (pre-QSF). Not sure but I'd guess there's a potential problem since the QSF'ed video has had some potential timestamp corrections but the audio track hasn't had those corrections.

Looking at the thread you linked left me confused. I don't understand what "grouping" audio tracks is or how you do it. Do mp4box or handbrake support this?

[KryptoNyte]

Can you clarify the details on your source file, and what you want as an output file? Are you trying to start with raw Tivo files (with AC3 audio only), and end with h264 handbrake encoded files with AC3 and AAC tracks?

I really believe that Handbrake is your goto tool here, and you can just have VAP call it up at the end of your processing and have it do whatever you wish (multiple audio tracks, file extension m4v, etc.). This is probably more a matter of opinion, but when it comes to encoding h264, I prefer Handbrake over VRD by a longshot.

[tannebil]

I like to option of being able to control when files are processed by VRD and when they are processed by a VRD/Handbrake combination. I've had files that VRD chokes on but HB handles and the reverse as well. If my process pushes everything through HB, when HB doesn't work, I'm screwed.

My source material is TiVo (mpg in a TiVo container) and Windows Media Center (mpg in a wtv container). I do 15-20 shows per week and want a process that's 100% automated and 99% reliable. Everything is going into h.264 format for viewing on Apple devices.

I was thinking a process (controlled by a setting) that flows something like this:

1. Check if the file has an AC-3 audio track
2. QSF the file into mpg format
2. Extract the audio
3. Transcode (respecting the profile match string)
4. Pass everything to post processing

(2) should take care of an timecode issues in the source file.

If you can't do (2) through VRD, that clearly creates some complications although I'm sure there are other options than VRD that could be used.

My vappostprocessing.cmd file is already pretty hefty and a bit funky so it's the weakest part of my workflow. VAP is much more reliable than anything I write.

[KryptoNyte]

I've had files that VRD chokes on but HB handles and the reverse as well...

...I do 15-20 shows per week and want a process that's 100% automated and 99% reliable...

The combination of those two things present quite a challenge. Over the past two years, I've never had handbrake choke on even one file I've thrown at it. Do you know ahead of time which files HB won't encode, and why?

[dlflannery]

@tannebil,

I can't envision any mod to VAP that would accomodate you and would be acceptable to me. I think KryptoNyte has suggested the right approach.

Can HB take an mpeg2 input file with the 3 audio tracks your have and produce the desired output file in one step? And when VRD (VAP) does the QSF on your Tivo file to the mpeg2 output, can it preserve all three audio tracks?

I'll be glad to help with the VapPostprocess script.

[KryptoNyte]

...And when VRD (VAP) does the QSF on your Tivo file to the mpeg2 output, can it preserve all three audio tracks?

Probably fodder for another thread, but I have a feeling this is going to be a recurring issue in many scenarios, a real thorn in the side. I have one show, Criminal Minds, that continually puts the "closed captioning voiceover for the blind" up as the first audio track, which VRD always selects in QSF. I always end up processing these manually. This can also raise a problem even with the solution I have in mind, if the AC3 track is not the top audio track.

[tannebil]

All my input files (at least those I checked) have a single AC-3 audio track and Handbrake has no problem converting it into a two audio track h.264 output file. One track is ACC (Dolby Pro-Logic II) and the other is AC-3 (passthrough) as discussed here: https://trac.handbrake.fr/wiki/SurroundSoundGuide

I was hoping that you could do something like you did to create the file I feed into Drax. Since it didn't attract your interest, I changed my workflow to make Handbrake the default encoder.

[dlflannery]

Sorry I couldn't accomodate you. It isn't from lack of interest. It looks like using HB for the encoding is a good choice since it gives you exactly the output audio tracks you want, and is considered at least as good an encoder as VRD.

Looking back at your first post, it appears you simply have decided to process your 1080i files the same way as you already do your 720p files, or have I missed something? So the question occurs to me: what was your motivation for processing the 1080i files differently in the first place? Is there some difference (other than resolution) that made encoding with VRD the first choice for the 1080i files?

[KryptoNyte]

Tannebil, I just wanted to mention that DL helped me get (2) different VAP instances running simultaneously. VAP1 does my QSF and commercial detection/cuts, but even with comskip, I still have to check and adjust cuts manually to my liking. VAP2's input directory is actually VAP1's output directory, so once all the QSF and commercial cutting is complete, VAP1 hands the file(s) off to VAP2 which basically only does file-rename and auto post-processing with HandbrakeCLI.

This allowed me to separate the processes, and I only start VAP2 processing when I go to bed, so that the processor intensive Handbrake encoding all happens when the system is otherwise dormant.

I'm not sure if this is something you would be interested in, but it has been working quite well for me.