Determine audio properties from recorded file instead of schedule data

[doug@…]

Hi All,

I found that the audio properties (e.g. Dolby, Stereo, etc) in my schedule data were often different from the actual content of my recorded files. This is a patch to update the audio properties after a recording finishes with the values that are in the file as opposed to the ones passed from the listings source.

The patch includes a new class called AudioPropsGenerator? which works similarly to PreviewGenerator?. It takes a ProgramInfo? and will then determine the AudioProperties? in the current thread or a new thread. I've also updated tv_rec.cpp to call this Generator after a recording ends. Finally, each decoder (avformatdecoder, nuppeldecoder, etc) is responsible for doing the actual work of determining the audio properties. The generator then takes the new value and updates the db.

There are a few open issues (mainly around my ignorance):

1. I don't know too much about audio processing, so I guessed that the speaker layout is a good proxy for determining the audio props. This seems to prove true for my test cases, but I don't know if this applies to everyone.

2. I don't know too much about the hard hearing properties so I guessed on that too.

3. I don't know what the visual impair property is so I ignored it

4. I don't have any recorders that use the nuppledecoder so I wrote the code, but I couldn't test it.

Thanks!

Doug

[danielk]

Doug, this is an interesting enhancement. Can you update it to apply against jya's ​http://svn.mythtv.org/trac/browser/branches/mythtv-hdaudio branch (or trunk if you see this after the merge of that branch to trunk) ?

[doug@…]

Hi,

I'm a little scared of moving to the hdaudio branch, but I've been keeping this patch up to date locally so when the hdaudio branch is merged with trunk I'll be sure to post here with my changes.

Thanks!

D

[Doug Haber <doug@…>]

This patch should apply against 24875 (post hdaudio merge)

[Doug Haber <doug@…>]

Hi,

I attached a new patch that applies against 24875 (post hdaudio merge). There's lots of room for improvement (e.g. I didn't improve on 1-4 above, and the AudioGenerator? is basically a copy of the PreviewGenerator? - they should probably share a common base class). But hopefully this can be a start.

Thanks!

D

[JYA]

I don't think I'm the best one to handle this one..

[robertm]

Hot Potato

[stuartm]

Milestone 0.25 deleted

[danielk]

Doug, while interesting I don't think it makes sense as is. The audio properties of a recording can change during the recording. To do this properly I think we need to note the audio properties in the entire file during something like the commflag run and then determine the static audio properties based on those properties in effect through most of the recording.