GES-Selecting Audio Stream

[Chris Wine]

Here is my select-tracks-for-object handler: https://pastebin.com/jTW1ZCgF

The main idea is that I know the integer index of the audio stream I want
(clip_data->target_stream), so I loop through the audio streams to find the
stream_id based on the index. Once I know that stream_id, the timeline's
track will only be returned if the stream_id of the track element matches.

--Chris

On Thu, Dec 3, 2020 at 11:32 AM Chris Wine <chriswine at gmail.com> wrote:

> Hi Thibault,
>
> I am using 1.18.1 on Windows, the x64 mingw version. That code does look
> like it should be doing the correct thing; however all the audio pads on
> the uridecodebin are still being created in my case. Here's a piece of the
> graph that shows the uridecode bin in the GESAudioURISource connecting to
> the volume adjustment bin: https://imgur.com/a/JpBDBV9
>
> --Chris
>
> On Thu, Dec 3, 2020 at 10:20 AM Thibault Saunier <tsaunier at gnome.org>
> wrote:
>
>> Hello,
>>
>> What you are doing is correct, and it used to be broken but should work
>> just fine in 1.18 thanks to:
>>
>>
>> https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/commit/09cc8dde0036420932704631f59531a1d84de018
>>
>> What version of GStreamer are you using?
>>
>> Regards,
>>
>> - Thibault
>>
>>
>> On Wed, Dec 2, 2020 at 9:40 PM Chris Wine <chriswine at gmail.com> wrote:
>>
>>> Hi, I'm having a problem with selecting the correct audio stream from an
>>> input file that I'm using with GES. The input file has 1 video stream and 4
>>> identical (as far as the CAPS are concerned) audio streams. I am currently
>>> using GStreamer/GES mingw x64 1.18.1 on the Windows platform.
>>>
>>> As the documentation suggested, I wrote a signal handler for the
>>> GESTimeline's "select-tracks-for-object" signal, and I've verified that it
>>> is working as expected with returning a NULL for GESTrackElements that are
>>> not the audio stream I want, while returning a GPtrArray with the
>>> timeline's audio track when it is the audio stream I want.
>>>
>>> The problem though is that only the 1st audio stream of the input file
>>> is ever chosen for the output. I've been looking through the GES source
>>> code, and it appears that when ges_source_create_topbin in GESSource is
>>> called for the audio, it will always just look for the first pad created
>>> when the "pad-added" signal is on uridecodebin is called; this made me
>>> wonder if something in the creation of uridecodebin itself was supposed to
>>> be responsible for selecting the correct audio stream. However, the only
>>> thing I can find there is the "caps" property which does restrict the
>>> streams, but since my 4 audio streams all have identical CAPS, this doesn't
>>> end up being a restriction.
>>>
>>> My question is first, am I missing something obvious that would make
>>> this work as expected? If not, aside from code changes to
>>> GESSource/UriSource/AudioUriSource, is there something tricky I can do like
>>> finding the uridecodebin, intercepting the pad-added signal before it gets
>>> to the GESSource, and only letting it through when getting to the pad that
>>> I want to be connected?
>>>
>>> Thank you, responses are much appreciated.
>>> --Chris
>>> _______________________________________________
>>> gstreamer-devel mailing list
>>> gstreamer-devel at lists.freedesktop.org
>>> https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
>>>
>>
>>
>> --
>> Thibault Saunier, Igalia - www.igalia.com <http://www.centricular.com>
>> _______________________________________________
>> gstreamer-devel mailing list
>> gstreamer-devel at lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/gstreamer-devel/attachments/20201203/1c9e5116/attachment-0001.htm>