Add an output format option in playback settings. The options are
Detect, S16LE, and F32LE. Selecting Detect will use the existing
behavior and detect the native format when the pipeline starts. The
other options will set the format when the pipeline is built.
Modify the signals emitted by the engine when a url is determined to be
valid or invalid to send the entire request. This will allow additional
metadata to be added to the request, providing a mechanism to better
identify the request source.
GstEnginePipeline::InitFromString was attempting to link the decoder and audio
bins twice and returning the result of the failed second attempt as it's own
status. The return value was also being ignored by the caller.
Add a class to wrap the URL in the playback engines. In the future, this will
contain authentication information for the specified URL. It can also include
the start and end time as well as other data that is currently specified along
with the URL.
The fix-up for URLs for files that that begin with // no longer works since the
QUrl class determines that these modifications are invalid, resulting in an
empty string when converted. Instead of attempting to modify the QUrl, add a
utility function that makes the correction on the encoded byte array at time of
usage.
Let's the user see the error message what failed instead of Clementine crashing.
Also don't do gst_object_unref unless bin is set.
This fixes GStreamer-CRITICAL gst_object_unref: assertion 'object != NULL' failed
When the network connection changes while playing an HTTP stream, I always get the "Server does not support seeking." error from GStreamer.
It seems like GStreamer tries to seek on reconnect, which fails, an propagates the error to Clementine which in turn ceases playback with
the error message handed through from GStreamer, even though there is now a perfectly fine network connection again.
As a workaround, try to reload the stream when this error occurs.
fixes#5116
This change provides the ability to set a fixed pipeline sample rate as an alternate to automatically negotiating it.
This can be useful on systems with sound cards that work at a fixed rate, as well as it can triage issues (on Windows)
where changing tracks hangs due to a problem with gstreamer's caps negotiation.
This is a regression in the upgrade to gstreamer1.0.
The gst_buffer_unref() is incorrect as it removes the buffer when it is still needed by the chunker.
Forcing the pointer to be null prevents it from segfaulting but causes it to skip all chunks in the buffer, dropping the framerate and causing a worse case of #4321.
Removing these 2 lines restores original functionality.
- Rewrite gstspectrum (1.0) to use FFTW (2x faster) and emit raw magnitude
values (not log scaled).
- Rewrite the moodbar generation code to be somewhat understandable, and
do it in Clementine instead of gstreamer.
There was a possibility of a dbz when a buffer sent to the
analyzer was shorter than 1ms long, such as what may happen at the end
of a track when stopping. This patch guards against this.
The chunks are now determined by the density of data in the buffer
to the length of audio in the buffer. The chunk length can change
size so that the audio that is analysed is exactly what is being played
at the instant the frame is requested.
The analyzers are sent new buffers of audio data to process each time
they pass through the gst pipeline. Different file formats and bit depths/
sample rates can change the size of these buffers, in some cases making them
large and therefore infrequent. This causes choppiness in the analyzer
as it is not getting new data with every frame. This patch chunks the buffers
coming off the pipeline to correspond with the framerate of the analyzer.
Jim Broadus
Jim Broadus
John Maguire
Jonas Kvinge
Chocobozzz
benpicco
Florian Bigard
Santiago Gil
Chocobozzz
John Maguire
Arnaud Bienner
Mark Furneaux
Eduardo Sánchez Muñoz
Mark Furneaux
Andreas
David Sansome