On Mon, Oct 4, 2010 at 9:44 AM, cristiurban <cristianurban86@gmail.com> wrote:

The pipeline I am using is :
gst-launch -v filesrc location=<path to media file> ! nv_omx_mp3dec !
audioconverter ! audioresample ! alsasink (used -v for debug)
This is a command that you should verify that it is working, for these to
work you must have the nvidia sound drivers. Of course, my app is in C
source, and somewhere after I give the playing state I try to extract the
current time and then I get zero.
If this doesn't work try the same pipeline but you must replace the
"nv_omx_decoder" with "mad" and works just fine, but this I know. In time I
had beed looking in OMX API to see if something wrong there, but no result

from filesrc, you will not get timestamps.  So things like seek won't work unless your mp3 decoder is also a parser and generating it's own timestamps and supporting seeking.  My guess is that mad does this, and your OMX component does not.

Please try putting mp3parse between filesrc and nv_omx_mp3dec