[Matroska-devel] multiple audio tracks with different languages

Matthew Heaney matthewjheaney at hotmail.com
Thu Apr 1 22:41:25 CEST 2010


The most recent release of VOB2MKV (vo2mkv-1.0.4) included a splitter-style
DirectShow filter, MKVSPLIT.  


Previous releases included only a source-style demuxer, MKVSOURCE, but this
doesn't work for network downloads, hence the need for the splitter filter

But downloading a file over a network made me aware about how much data was
actually being downloaded (a problem I didn't worry about when only playing
a local file).

Suppose I mux an MKV file with multiple audio tracks, each having a
different language.  Typically a client will only view a single language
track.  If all of the tracks are interleaved in the file, and the file
downloaded from a server to a client, and the client only selects a single
track, then that would mean lots of data is being downloaded from the server
only to be thrown away by the client.

My question is whether there is a way to mux the MKV file (or files) such
that only a single track of audio need be downloaded from the server.  Is
there an intended mechanism for doing this?

Could each audio track go in its own segment?  (Or perhaps there is a better
way.)  My question about that approach is that the multiple audio tracks,
each in their own segment, all start at the same time.  Are segments
intended to be stitched together sequentially, such that one segment begins
after the previous segment ends?  Or can segments exist "in parallel"?

My understanding is that segments in different files can be linked together,
using segment UIDs.  But what about putting multiple segments in a single
file?  Does this make any sense?  Is there a difference between putting
multiple segment in a single file, versus multiple segments in multiple


More information about the Matroska-devel mailing list