I agree with you on the playability. I just had too many re-containerized files that would not play. Just wasn't worth the hassle for the quality improvement (which I could not really see anyway).
Would you elaborate on your audio conversion process?
Sure. I discovered that both Subler and Handbrake aren't doing a great job when it comes to audio conversion. And you have very crude audio settings to chose from.
For example, when having a 5.1 AC3/DTS track, you can let Handbrake downmix it to a stereo track. You can chose normal stereo, or Dolby Pro Logic II, etc. However, the downmix matrices are not documented anywhere. I had a look at a downmixed 2 channel DPLII track, and it had severe clipping. It seemed like the downmix matrix was not normalized, or something else was bad. In either way, I can't take the risk of having Handbrake produce severely clipped stereo tracks that sound horrible to listen to. Also, the volume of the tracks are usually too low. You can control this via the gain function, and also apply some dynamic compression. However, both options are too crude and are a gamble, today. Gain function doesn't seem to take into account clipping, and there is no normalization function.
My problem with Subler is similar. No real documentation, too crude settings, and several failures of channel mappings when going from a 5.1 track to 2 channel track. I simply cannot use software that don't produce reliable results, because it's quite tedious having to doublecheck the results after each conversion.
With that said, both programs are of course excellent. It's just that they lack precision and quality I need when it comes to audio conversion.
So what I do instead, is to use ffmpeg to extract the AC3/DTS track from my MKV, and let it uncompress to a .caf file (which supports larger 2GB file size). Having the .caf file, I downmix it with Sox, using a well working stereo downmix matrix, that is normalized.
After having the downmixed .caf file, I open it up in Audacity, and apply desired dynamic compression, and amplification of the track to reach a good volume. Finally, I produce a playable .aac track via the afconvert utility. Then I mux the videostream (from Handbrake) and the .aac track together with Subler, adding subtitles, chapters, artwork. I also use another program to produce an additional AC3 track from DTS, which I also just mux with Subler.
While the above process gives me what I want, it would of course be more convenient to just use one program like Handbrake to achieve everything. But for audio conversion, it not reliable for me, yet.
