Encode H264 AAC proxies with discrete clean stereo mix SOLVED
Example of issue:
Lets say you have typical rushes material as a .mov (ingested or from file such as xdcam). When shooting you had 2 mics, an onboard mic for natural sound on Ch1 and a wireless lav mic for interviews on Ch2.
Once you have your files ingested you will notice the audio files are actually 2 separate tracks within quicktime. You will most likely have the video track, 2 separate mono audio tracks and a timecode track.
If you were to encode this file into a proxy h264 with stereo audio through the quicktime engine/Catdv (AAC Channels: Stereo L R), ie one track to the left and one to the right you will simply get a mixdown of both on each left and right channel, ie no more clean separate channels. This is because Quicktime is treating these as 2 completely seperate audio tracks (FCP deals with them slightly differently but that is more to do with is audio channel mapping).
It is also not possible to encode a 2 channel mix through AAC, at least not through Quicktime.. Episode is the only software I have found that can handle this but it is expensive and then requires further work to automate tasks via the worker and offloading to episode etc.
It would be great if we could get this to work through the the native QT engine used by CatDV...
The solution to getting a clean stereo encode is to tell the quicktime engine prior to encoding that one of the channels is the LEFT and the 2nd is RIGHT. You can also set this to be discrete-0 and discrete-1 (ie first speaker, second speaker), it will map the same way and is very useful if you need to encode multiple tracks of audio.
To do this you open the file in Quicktime 7 Pro, go to movie properties, select the first audio channel's 'Sound Track' and under the audio settings tab assign the 1st track to 'Left' and the second track to 'Right'. If you have the QT inspector up you will notice the words left and right appear in inspector next to each corresponding track.
Save the file and then use the Quicktime engine to encode your proxy with AAC Stereo (L,R) and you will now get the correct mix. You can open it in FCP and see that it will treat the audio tracks in basically the same way as the original file with full control over each channel (it's simply flagged as linked stereo). Open the audio mixer, you can now mute each channel and hear them individually. In CatDV it will technically show just 'A1' but pan it fully to the left for Ch1 and fully to the right to hear Ch2 just as if it was a full stereo mix.
Ideally I would like to have a process that could change the properties element for each file. It would be great if you could do this through CatDV as you could then batch change all your files before making proxies or simply have the worker do it on import.
Unfortunately the 'Sound Track' properties do not seem to be an editable metadata tag. The useful app 'Media Info' has this flagged as 'Channel positions : Front: L' (this is only visible once you have selected this in the quicktime properties). If you select 'discrete-0, -1' you will also not see this flagged but the encode will still be correct.
As far as I can see none of these properties are present in any of the CatDV metadata.
Any further info on how to batch change this property would be great. I have seen some apple scripts for doing similar things but it would be useful to know more about these quicktime properties first.
Neo Verite Limited