Now we can configure how much time to wait before deciding that a
discont has happened.
Also, adds getter and setter to allow derived implementations to set
this value upon construction.
Suggestions and several improvements by Havard Graff.
Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
A common problem for audio-playback is that the timestamps might not
be completely linear. This is specially common when doing streaming over
a network, where you can have jittery and/or bursty packettransmission,
which again will often be reflected on the buffertimestamps.
Now, the current implementation have a threshold that says how far the
buffertimestamp is allowed o drift from the ideal aligned time in the
ringbuffer. This was an instant reaction, and ment that if one buffer
arrived with a timestamp that would breach the drift-tolerance, a resync
would take place, and the result would be an audible gap for the
listener.
The annoying thing would be that in the case of a "timestamp-outlier",
you would first resync one way, say +100ms, and then, if the next
timestamp was "back on track", you would end up resyncing the other way
(-100ms) So in fact, when you had only one buffer with slightly off
timestamping, you would end up with *two* audible gaps. This is the
problem this patch addresses.
The way to "fix" this problem with the previous implementation, would
have been to increase the "drift-tolerance" to a value that was greater
than the largest timestamp-outlier one would normally expect. The big
problem with this approach, however, is that it will allow normal
operations with a huge offset timestamp vs running-time, which is
detrimental to lip-sync. If the drift-tolerance is set to 200ms, it
basically means that lip-sync can easily end up being off by that much.
This patch will basically start a timer when the first breach of
drift-tolerance is detected. If any following timestamp for the next n
nanoseconds gets "back on track" within the threshold, it has basically
eliminated the effect of an outlier, and the timer is stopped. If,
however, all timestamps within this time-limit are breaching the
threshold, we are probably facing a more permanent offset in the
timestamps, and a resync is allowed to happen.
So basically this patch offers something as rare as both higher
accuracy, it terms of allowing smaller drift-tolerances, as well as much
smoother, less glitchy playback!
Commit message and improvments by Havard Graff.
Fixes bug #640859.
Otherwise we'll just error out when the first buffer gets pushed.
This is a porting artefact, in 0.10 the infos were allocated on the
heap, now we're doing everything with stack-allocated structs.
Not sure how this one got pulled into a merge. In 0.10, it was moved away to
gst-template a long time ago. gstaudiofilterexample.c got generated from
gstaudiofiltertemplate.c.
The array we're writing to is limited to 64 ... but the amount of
input positions might be lower than 64. Therefore use MIN and not
MAX to know how many values to read from the array.
Add a method to configure the output caps. Subclasses can't use
gst_pad_set_caps() anymore because then we won't see the caps.
Unbreak the padtemplate registration, the GTypeClass that is configured in the
object during _init is not the right one, we need to use the klass passed as the
argument to the init function..
This reverts commit 11e375486e.
GST_BOILERPLATE() can't define an abstract type and
G_DEFINE_ABSTRACT_TYPE() does not pass the class struct to
the instance_init function and there's no way to get the
class struct of the current type in instance_init().
Merge in doc updates for audio enums from 0.10, and get rid
of the #if #else in the enum list, since that confuses gtk-doc.
Conflicts:
gst-libs/gst/audio/audio.c
gst-libs/gst/audio/audio.h
Remove the _ in front of the endianness prefix.
Remove the _3 postfix for the 24 bits formats.
Add a _32 postfix after the formats that occupy extra space beyond their
natural size.
The result is that the GST_AUDIO_NE() macro can simply append the endianness
after all formats and that we only specify a different sample width when it is
different from the natural size of the sample. This makes things more consistent
and follows the pulseaudio conventions instead of the alsa ones.
Leaving the GST_USE_UNSTABLE_API guards in until some of the
ported decoders have been updated and it's clear that I didn't
mess up anywhere porting things to the new audio API.
Adds little beyond baseaudiocodec (seeking, bit of query), and what it adds
is mainly out-of-scope (e.g. decoder seeking, should be done by upstream
demuxer/parser) and/or based on non-prime example (mad).
Moved most of the code to GstBaseAudioCodec, GstBaseAudioDecode is
now really small, maybe we do not really need it (or its encoder
counterpart). Added more API for subclasses and documentation.
Rework the audio caps similar to the video caps. Remove
width/depth/endianness/signed fields and replace with a simple string
format and media type audio/x-raw.
Create a GstAudioInfo and some helper methods to parse caps.
Remove duplicate code from the ringbuffer and replace with audio info.
Use AudioInfo in the base audio filter class.
Port elements to new API.
This can be used by sinks to take compressed formats, correctly payload
these in IEC 61937 frames and feed these to sinks that support
passthrough output over IEC 60958 (S/PDIF) or, in the case of MP3, over
Bluetooth.
Initial implementation includes AC3, E-AC3, MPEG-1, MPEG-2 (non-AAC),
and DTS (type-I/II/II) payloading. More formats can be added as needed.
API: gst_audio_iec61937_frame_size()
API: gst_audio_iec61937_payload()
https://bugzilla.gnome.org/show_bug.cgi?id=642730
This allows subclasses to provide a "payload" function to prepare
buffers for consumption. The immediate use for this is for sinks that
can handle compressed formats - parsers are directly connected to the
sink, and for formats such as AC3, DTS, and MPEG, IEC 61937 patyloading
might be used.
API: GstBaseAudioSinkClass:payload()
https://bugzilla.gnome.org/show_bug.cgi?id=642730
Adds support for pushing E-AC3 buffers and doing bytes-to-ms conversion
correctly. The assumption (as with other formats) is that something like
IEC 61937 payloading will be used. Correspondingly the ringbuffer spec
is populated so that the data rate is 4x normal AC3.
https://bugzilla.gnome.org/show_bug.cgi?id=642730
These are meant to be used for buffers containing AAC data. Nothing uses
this yet, but for now it serves to distinguish from GST_BUFTYPE_MPEG
which represents non-AAC MPEG audio.
API: GST_BUFTYPE_MPEG2_AAC
API: GST_BUFTYPE_MPEG4_AAC
Remove the android/ top dir
Fixe the Makefile.am to be androgenized
To build gstreamer for android we are now using androgenizer which generates the
needed Android.mk files.
Androgenizer can be found here:
http://git.collabora.co.uk/?p=user/derek/androgenizer.git
A race was observed between query() and setcaps() where the latter would
change the ringbuffer spec while the former was performing operations
based this data.
Observed a case where the src went to null-state during the query,
hence the spec pointer was no longer valid, and
gst_util_unit64_scale_int crashed (assertion `denom > 0´failed)
Add locking to make sure the ringbuffer can't disappear.
Given a large enough drift-tolerance, one could end up in a situation
where one would keep aligning the written buffers behind the current
read-segment position. The result for the reader would be complete
silence, possible preceded by very choppy audio.
By checking the available headroom, one can determine if there is
room to do alignment, or if one should resort to a resync instead to get
the pointers back on track.
Also refactor the alignment-logic out of the render function for cleaner
code.
Commit ba2e500bd9 ensured to provide
a running clock when EOS had finished rendering. However,
other measures are needed (and were in place before) to ensure a
running clock when EOS still needs rendering (i.e. waiting).
So, specifically, re-introduce eos_rendering removed in aforementioned commit,
this time as a public variable so subclasses can be aware of the situation.
Fixes (part of) #645961.
API: GstBaseAudioSink:eos_rendering
Fixes introspection failures caused by type assertions/warnings.
Since we now moved from _get_type() functions to external GType
variables in a couple of places, we actually have to call gst_init()
to make sure these are set when we use GST_TYPE_FOO.
Make sure to use the PKG_CONFIG_PATH set at configure time instead of
just relying on an env-var set one. This makes sure both g-ir-compiler
and g-ir-scanner use the same PKG_CONFIG_PATH for determining include
paths etc.
Observed a case where the sink went to null-state during the query,
hence the ringbuffer-pointer was NULL, causing a crash.
Moving the ringbuffer-check code until after the query, and hold the
lock during the check and while using the spec-values. It should not matter
to the query wether the ringbuffer is present or not, and it actually
gets a time bit more time to get the ringbuffer set up in this case!
Fixes#635231
So run-time bindings can introspect the names correctly (we abuse this
field as description field only in elements, not for public API
(where the description belongs into the gtk-doc chunk).
https://bugzilla.gnome.org/show_bug.cgi?id=629746
Don't take an extra ref on the sink and source because that creates a reference
cycle. Instead, use the invalidate method of the clock when the sink and source
are freed. This way, we don't call into the time function anymore after the
objects are disposed.
When calling gobject-introspection scanner, make sure our own
freshly-built libs within the source tree (well, build dir) come
first in the PKG_CONFIG_PATH. May or may not help to make sure
that it doesn't pick up older external plugins-base libs (or
.gir files) from outside the source tree / build directory as
dependencies of the introspected lib instead of using the
stuff we just built in a sibling directory.
https://bugzilla.gnome.org/show_bug.cgi?id=623698
Elements usually use their own instance as instance data but the
clock can have a longer lifetime than their elements and the clock
doesn't own a reference of the element.
Fixes bug #623807.
Check for the state of the ringbuffer before doing the checks of the other
buffer properties, when we're not started, we don't care about those values.
This reverts commit cea2644ed8.
Many audio sink assume that they can create a clock in
the instance init function and it will be there forever
and not be cleared by the state change functions.
Point g-ir-scanner to the .la file of our library, which hopefully
makes it find the right dependencies in all cases (ie. our locally
built libgstreamer and not the system-installed one). This is also
how it's done in Gtk+ and how it's documented in the wiki, see
http://live.gnome.org/GObjectIntrospection/AutotoolsIntegrationFixes#603710.
Use new girdir and typlibdir from core .pc files, so we can figure
out the right includes to pass to the gobject-introspection tools,
whether core is installed in the same prefix as gobject-introspection
or in a different prefix or uninstalled. This also keeps us from adding
bogus paths to the includes that only work if core is uninstalled.
Also add some missing includes/pkgs where needed.
Increase default drift tolerance to 40ms to avoid glitches with decoders
or formats where there's a lot of timestamp jitter for some reason or
another (in this case: asf/wma), at least until we implement timestamp
smoothing.
Our calibration against the pipeline clock is done with the adjusted
ringbuffer time, so take the adjustement into account. Fixes some audio dropouts
when reusing audio sinks after switching clocks and slaving methods in a
pipeline.
When we are calibrating the internal clock against the external clock take into
account the time offset applied to our internal clock because we will subtract
that in the render_function again.
When we start and we need to produce the first sample, go to the next sample
that will be written into the ringbuffer instead of trying to go to sample 0.
We relied on rather small ringbuffer sizes to correctly go to the current
sample, which breaks whith large buffers.
Fixes#600945
Add drift-tolerance property (defaulting to 20ms) to handle resync after clock
drift or timestamp drift instead of relying on the latency-time value for clock
drift and 500ms for timestamp drift.
Remove warning about discont timestamp and simply resync. The warning is in some
cases not correct and is triggered more frequently now that we lower the
tolerance value.
There's not much point in using GST_DEBUG_FUNCPTR with GObject
virtual functions such as get_property, set_propery, finalize and
dispose, since they'll never be used by anyone anyway. Saves a
few bytes and possibly a sixteenth of a polar bear.
I also renamed glib_enum_prefix to glib_gen_prefix as we also use that for the
marshallers. Also rename the rtsp-marshal.list to work with the unified prefix.
Check for pulsesink < 0.10.17 because it includes code that is now included in
baseaudiosink. Disable that code in baseaudiosink to be compatible with the
older version.
Take the time of the clock so that the last_time field is set. This is important
for sinks that restart their internal ringbuffer after a caps change and need to
know the last know position.
When going to NULL, we reset the ringbuffer so that it starts beck from 0. We
also make sure that the clock is updated with the elapsed time so that it
alsways increments even when the ringbuffer goes back to 0. When this happened
we need to adjust the sample position for the reset ringbuffer.
Fixes#594136
The old one did the mistake of not actually advancing the ringbuffer, it just
adjusted the segbase, introducing the whole lenght of the ringbuffer as an
extra delay in the pipeline.
Also make sure that the resync can never go back in time, producing the same
timestamps that has already been produced, as this can cause severe problems
for sinks and other synching mechanisms.
Fixes#594256
Add section docs for multichannel, so that it has a short desc in the toc too.
Move the section docs in adio up, so that the follow the copyright like
elsewhere.
Unparent and free the ringbuffer when going to NULL, like we do with the
audiosrc element. We can do this now because we correctly manage the time
jumping back to 0.
Hack around thread-safety issues in GObject and our racy _get_type()
functions (we could easily fix the _get_type() functions, but we still
need to hack around the GObject class races until we require a newer
GLib version, I think).
Use the unadjusted internal clock times to calculate the internal/external
offset when calibrating the clock.
When going to NULL, unparent and free the ringbuffer, like we do in the source
element.
See #578506
Add two methods for getting the unadjusted time of the clock and one for
adjusting an internal time. We will need these methods for correctly handling
the time after a gst_audio_clock_reset().
Add a debug category and some debug lines to the audio clock.
API: gst_audio_clock_get_time()
API: gst_audio_clock_adjust()
API: GST_AUDIO_CLOCK_CAST()
After we pause the stream and interrupt the writeout to the ringbuffer, also adjust
the amount of output samples we consumed. We can't do this reliably with the
current API when we are doing trick modes but we can do the right thing for
normal playback.
Original commit message from CVS:
Patch by: José Alburquerque <jaalburqu svn gnome org>
* gst-libs/gst/audio/gstaudioclock.c: (gst_audio_clock_new):
* gst-libs/gst/audio/gstaudioclock.h:
Make gst_audio_clock_new use const gchar* to ease the wrapping of
C++ bindings. Fixes#566723.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_change_state):
Avoid holding the OBJECT_LOCK when calling ringbuffer functions that
take the ringbuffer lock because rinbuffer lock > OBJECT_LOCK. We can do
this because the async_play method is deprecated and usually not called
anymore.
Original commit message from CVS:
* gst-libs/gst/audio/Makefile.am:
* gst-libs/gst/audio/audio.c:
* gst-libs/gst/audio/multichannel.h:
* gst-libs/gst/audio/testchannels.c:
* win32/MANIFEST:
* win32/common/audio-enumtypes.c:
(gst_audio_channel_position_get_type),
(gst_ring_buffer_state_get_type),
(gst_ring_buffer_seg_state_get_type),
(gst_buffer_format_type_get_type), (gst_buffer_format_get_type):
* win32/common/audio-enumtypes.h:
* win32/common/multichannel-enumtypes.c:
* win32/common/multichannel-enumtypes.h:
* win32/vs6/grammar.dsp:
* win32/vs6/libgstaudio.dsp:
* win32/vs7/libgstaudio.vcproj:
* win32/vs8/libgstaudio.vcproj:
Switch glib-mkenum for gst-libs/gst/audio from multichannel- to
audio- in order to wrap all enums declarations of that library.
This modification should not matter since that header file is not a
public header (it will be included by public headers).
Modify win32 crap^Wfiles accordingly.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosrc.h:
* gst-libs/gst/audio/gstbaseaudiosink.h:
Complete Sebastien's commit from the 13th by exporting the
_slave_method_get_type() methods.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_setcaps):
Pause the write thread before deactivating and releasing the ringbuffer
to avoid a deadlock when we do gapless playback with different sample
rates in playbin2. Fixes#564929.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosrc.c:
Make GstAudioSrcSlaveMethod get_type() function non-static
as it's public now.
* win32/common/libgstaudio.def:
* win32/common/libgstnetbuffer.def:
Add some missing functions to the list of exported symbols.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_slave_method_get_type),
(gst_base_audio_sink_class_init):
* gst-libs/gst/audio/gstbaseaudiosink.h:
* gst-libs/gst/audio/gstbaseaudiosrc.c:
(gst_base_audio_src_slave_method_get_type),
(gst_base_audio_src_class_init):
* gst-libs/gst/audio/gstbaseaudiosrc.h:
API: Add GST_TYPE_BASE_AUDIO_(SRC|SINK)_SLAVE_METHOD to the
public API. This is needed for the C++ bindings to be able
to use this base classes. Fixes bug #564200, #564206.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosrc.c:
(gst_base_audio_src_get_offset), (gst_base_audio_src_create):
Avoid nasty int overflows after about 12 hours and 25 minutes when these
code paths are triggered.
A free beer to Håvard Graff for finding this!
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_drain), (gst_base_audio_sink_event),
(gst_base_audio_sink_sync_latency), (gst_base_audio_sink_render),
(gst_base_audio_sink_change_state):
Really fix audiosink drain handling by keeping track of the running_time
of the last sample.
