mirrors/gstreamer
mirrors/gstreamer
1
1
Fork 0
mirror of https://gitlab.freedesktop.org/gstreamer/gstreamer.git synced 2024-12-18 14:26:43 +00:00
Code Issues 1 Projects Releases Wiki Activity
gstreamer/ext/webrtcdsp/gstwebrtcechoprobe.cpp

352 lines
10 KiB
C++
Raw Normal View History

webrtcdsp: Add WebRTC Audio Processing support This DSP library can be used to enhance voice signal for real time communication call. In implements multiple filters like noise reduction, high pass filter, echo cancellation, automatic gain control, etc. The webrtcdsp element can be used along, or with the help of the webrtcechoprobe if echo cancellation is enabled. The echo probe should be placed as close as possible to the audio sink, while the DSP is generally place close to the audio capture. For local testing, one can use an echo loop pipeline like the following: autoaudiosrc ! webrtcdsp ! webrtcechoprobe ! autoaudiosink This pipeline should produce a single echo rather then repeated echo. Those elements works if they are placed in the same top level pipeline. https://bugzilla.gnome.org/show_bug.cgi?id=767800
2016-06-06 19:10:05 +00:00
/*
* WebRTC Audio Processing Elements
*
* Copyright 2016 Collabora Ltd
* @author: Nicolas Dufresne <nicolas.dufresne@collabora.com>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*
*/
/**
* SECTION:element-webrtcechoprobe
*
* This echo probe is to be used with the webrtcdsp element. See #gst-plugins-bad-plugins-webrtcdsp
* documentation for more details.
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "gstwebrtcechoprobe.h"
#include <webrtc/modules/interface/module_common_types.h>
#include <gst/audio/audio.h>
GST_DEBUG_CATEGORY_EXTERN (webrtc_dsp_debug);
#define GST_CAT_DEFAULT (webrtc_dsp_debug)
#define MAX_ADAPTER_SIZE (1*1024*1024)
static GstStaticPadTemplate gst_webrtc_echo_probe_sink_template =
GST_STATIC_PAD_TEMPLATE ("sink",
GST_PAD_SINK,
GST_PAD_ALWAYS,
GST_STATIC_CAPS ("audio/x-raw, "
"format = (string) " GST_AUDIO_NE (S16) ", "
"layout = (string) interleaved, "
"rate = (int) { 48000, 32000, 16000, 8000 }, "
"channels = (int) [1, MAX]")
);
static GstStaticPadTemplate gst_webrtc_echo_probe_src_template =
GST_STATIC_PAD_TEMPLATE ("src",
GST_PAD_SRC,
GST_PAD_ALWAYS,
GST_STATIC_CAPS ("audio/x-raw, "
"format = (string) " GST_AUDIO_NE (S16) ", "
"layout = (string) interleaved, "
"rate = (int) { 48000, 32000, 16000, 8000 }, "
"channels = (int) [1, MAX]")
);
G_LOCK_DEFINE_STATIC (gst_aec_probes);
static GList *gst_aec_probes = NULL;
G_DEFINE_TYPE (GstWebrtcEchoProbe, gst_webrtc_echo_probe,
GST_TYPE_AUDIO_FILTER);
static gboolean
gst_webrtc_echo_probe_setup (GstAudioFilter * filter, const GstAudioInfo * info)
{
GstWebrtcEchoProbe *self = GST_WEBRTC_ECHO_PROBE (filter);
GST_LOG_OBJECT (self, "setting format to %s with %i Hz and %i channels",
info->finfo->description, info->rate, info->channels);
GST_WEBRTC_ECHO_PROBE_LOCK (self);
self->info = *info;
/* WebRTC library works with 10ms buffers, compute once this size */
self->period_size = info->bpf * info->rate / 100;
if ((webrtc::AudioFrame::kMaxDataSizeSamples * 2) < self->period_size)
goto period_too_big;
GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
return TRUE;
period_too_big:
GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
GST_WARNING_OBJECT (self, "webrtcdsp format produce too big period "
"(maximum is %" G_GSIZE_FORMAT " samples and we have %u samples), "
"reduce the number of channels or the rate.",
webrtc::AudioFrame::kMaxDataSizeSamples, self->period_size / 2);
return FALSE;
}
static gboolean
gst_webrtc_echo_probe_stop (GstBaseTransform * btrans)
{
GstWebrtcEchoProbe *self = GST_WEBRTC_ECHO_PROBE (btrans);
GST_WEBRTC_ECHO_PROBE_LOCK (self);
gst_adapter_clear (self->adapter);
GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
return TRUE;
}
static gboolean
gst_webrtc_echo_probe_src_event (GstBaseTransform * btrans, GstEvent * event)
{
GstBaseTransformClass *klass;
GstWebrtcEchoProbe *self = GST_WEBRTC_ECHO_PROBE (btrans);
GstClockTime latency;
webrtcdsp: Rewrite echo data synchronization The previous code would run out of sync if there was packet lost or clock skews. When that happened, the echo cancellation feature would completely stop working. As this is crucial for audio calls, this patch re-implement synchronization completely. Instead of letting it drift until next discont, we now synchronize against the record data at every iteration. This way we simply never let the stream drift for longer then 10ms period. We also shorter the delay by using the latency up the probe (basically excluding the sink latency. This is a decent delay to avoid starving in the probe queue. https://bugzilla.gnome.org/show_bug.cgi?id=768009
2016-06-29 19:56:47 +00:00
GstClockTime upstream_latency = 0;
GstQuery *query;
webrtcdsp: Add WebRTC Audio Processing support This DSP library can be used to enhance voice signal for real time communication call. In implements multiple filters like noise reduction, high pass filter, echo cancellation, automatic gain control, etc. The webrtcdsp element can be used along, or with the help of the webrtcechoprobe if echo cancellation is enabled. The echo probe should be placed as close as possible to the audio sink, while the DSP is generally place close to the audio capture. For local testing, one can use an echo loop pipeline like the following: autoaudiosrc ! webrtcdsp ! webrtcechoprobe ! autoaudiosink This pipeline should produce a single echo rather then repeated echo. Those elements works if they are placed in the same top level pipeline. https://bugzilla.gnome.org/show_bug.cgi?id=767800
2016-06-06 19:10:05 +00:00
klass = GST_BASE_TRANSFORM_CLASS (gst_webrtc_echo_probe_parent_class);
switch (GST_EVENT_TYPE (event)) {
case GST_EVENT_LATENCY:
gst_event_parse_latency (event, &latency);
webrtcdsp: Rewrite echo data synchronization The previous code would run out of sync if there was packet lost or clock skews. When that happened, the echo cancellation feature would completely stop working. As this is crucial for audio calls, this patch re-implement synchronization completely. Instead of letting it drift until next discont, we now synchronize against the record data at every iteration. This way we simply never let the stream drift for longer then 10ms period. We also shorter the delay by using the latency up the probe (basically excluding the sink latency. This is a decent delay to avoid starving in the probe queue. https://bugzilla.gnome.org/show_bug.cgi?id=768009
2016-06-29 19:56:47 +00:00
query = gst_query_new_latency ();
if (gst_pad_query (btrans->srcpad, query)) {
gst_query_parse_latency (query, NULL, &upstream_latency, NULL);
if (!GST_CLOCK_TIME_IS_VALID (upstream_latency))
upstream_latency = 0;
}
webrtcdsp: Add WebRTC Audio Processing support This DSP library can be used to enhance voice signal for real time communication call. In implements multiple filters like noise reduction, high pass filter, echo cancellation, automatic gain control, etc. The webrtcdsp element can be used along, or with the help of the webrtcechoprobe if echo cancellation is enabled. The echo probe should be placed as close as possible to the audio sink, while the DSP is generally place close to the audio capture. For local testing, one can use an echo loop pipeline like the following: autoaudiosrc ! webrtcdsp ! webrtcechoprobe ! autoaudiosink This pipeline should produce a single echo rather then repeated echo. Those elements works if they are placed in the same top level pipeline. https://bugzilla.gnome.org/show_bug.cgi?id=767800
2016-06-06 19:10:05 +00:00
GST_WEBRTC_ECHO_PROBE_LOCK (self);
self->latency = latency;
webrtcdsp: Rewrite echo data synchronization The previous code would run out of sync if there was packet lost or clock skews. When that happened, the echo cancellation feature would completely stop working. As this is crucial for audio calls, this patch re-implement synchronization completely. Instead of letting it drift until next discont, we now synchronize against the record data at every iteration. This way we simply never let the stream drift for longer then 10ms period. We also shorter the delay by using the latency up the probe (basically excluding the sink latency. This is a decent delay to avoid starving in the probe queue. https://bugzilla.gnome.org/show_bug.cgi?id=768009
2016-06-29 19:56:47 +00:00
self->delay = upstream_latency / GST_MSECOND;
webrtcdsp: Add WebRTC Audio Processing support This DSP library can be used to enhance voice signal for real time communication call. In implements multiple filters like noise reduction, high pass filter, echo cancellation, automatic gain control, etc. The webrtcdsp element can be used along, or with the help of the webrtcechoprobe if echo cancellation is enabled. The echo probe should be placed as close as possible to the audio sink, while the DSP is generally place close to the audio capture. For local testing, one can use an echo loop pipeline like the following: autoaudiosrc ! webrtcdsp ! webrtcechoprobe ! autoaudiosink This pipeline should produce a single echo rather then repeated echo. Those elements works if they are placed in the same top level pipeline. https://bugzilla.gnome.org/show_bug.cgi?id=767800
2016-06-06 19:10:05 +00:00
GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
webrtcdsp: Rewrite echo data synchronization The previous code would run out of sync if there was packet lost or clock skews. When that happened, the echo cancellation feature would completely stop working. As this is crucial for audio calls, this patch re-implement synchronization completely. Instead of letting it drift until next discont, we now synchronize against the record data at every iteration. This way we simply never let the stream drift for longer then 10ms period. We also shorter the delay by using the latency up the probe (basically excluding the sink latency. This is a decent delay to avoid starving in the probe queue. https://bugzilla.gnome.org/show_bug.cgi?id=768009
2016-06-29 19:56:47 +00:00
GST_DEBUG_OBJECT (self, "We have a latency of %" GST_TIME_FORMAT
" and delay of %ims", GST_TIME_ARGS (latency),
(gint) (upstream_latency / GST_MSECOND));
webrtcdsp: Add WebRTC Audio Processing support This DSP library can be used to enhance voice signal for real time communication call. In implements multiple filters like noise reduction, high pass filter, echo cancellation, automatic gain control, etc. The webrtcdsp element can be used along, or with the help of the webrtcechoprobe if echo cancellation is enabled. The echo probe should be placed as close as possible to the audio sink, while the DSP is generally place close to the audio capture. For local testing, one can use an echo loop pipeline like the following: autoaudiosrc ! webrtcdsp ! webrtcechoprobe ! autoaudiosink This pipeline should produce a single echo rather then repeated echo. Those elements works if they are placed in the same top level pipeline. https://bugzilla.gnome.org/show_bug.cgi?id=767800
2016-06-06 19:10:05 +00:00
break;
default:
break;
}
return klass->src_event (btrans, event);
}
static GstFlowReturn
gst_webrtc_echo_probe_transform_ip (GstBaseTransform * btrans,
GstBuffer * buffer)
{
GstWebrtcEchoProbe *self = GST_WEBRTC_ECHO_PROBE (btrans);
GstBuffer *newbuf = NULL;
GST_WEBRTC_ECHO_PROBE_LOCK (self);
newbuf = gst_buffer_copy (buffer);
/* Moves the buffer timestamp to be in Running time */
GST_BUFFER_PTS (newbuf) = gst_segment_to_running_time (&btrans->segment,
GST_FORMAT_TIME, GST_BUFFER_PTS (buffer));
gst_adapter_push (self->adapter, newbuf);
if (gst_adapter_available (self->adapter) > MAX_ADAPTER_SIZE)
gst_adapter_flush (self->adapter,
gst_adapter_available (self->adapter) - MAX_ADAPTER_SIZE);
GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
return GST_FLOW_OK;
}
static void
gst_webrtc_echo_probe_finalize (GObject * object)
{
GstWebrtcEchoProbe *self = GST_WEBRTC_ECHO_PROBE (object);
G_LOCK (gst_aec_probes);
gst_aec_probes = g_list_remove (gst_aec_probes, self);
G_UNLOCK (gst_aec_probes);
gst_object_unref (self->adapter);
self->adapter = NULL;
G_OBJECT_CLASS (gst_webrtc_echo_probe_parent_class)->finalize (object);
}
static void
gst_webrtc_echo_probe_init (GstWebrtcEchoProbe * self)
{
self->adapter = gst_adapter_new ();
gst_audio_info_init (&self->info);
g_mutex_init (&self->lock);
webrtcdsp: Rewrite echo data synchronization The previous code would run out of sync if there was packet lost or clock skews. When that happened, the echo cancellation feature would completely stop working. As this is crucial for audio calls, this patch re-implement synchronization completely. Instead of letting it drift until next discont, we now synchronize against the record data at every iteration. This way we simply never let the stream drift for longer then 10ms period. We also shorter the delay by using the latency up the probe (basically excluding the sink latency. This is a decent delay to avoid starving in the probe queue. https://bugzilla.gnome.org/show_bug.cgi?id=768009
2016-06-29 19:56:47 +00:00
self->latency = GST_CLOCK_TIME_NONE;
webrtcdsp: Add WebRTC Audio Processing support This DSP library can be used to enhance voice signal for real time communication call. In implements multiple filters like noise reduction, high pass filter, echo cancellation, automatic gain control, etc. The webrtcdsp element can be used along, or with the help of the webrtcechoprobe if echo cancellation is enabled. The echo probe should be placed as close as possible to the audio sink, while the DSP is generally place close to the audio capture. For local testing, one can use an echo loop pipeline like the following: autoaudiosrc ! webrtcdsp ! webrtcechoprobe ! autoaudiosink This pipeline should produce a single echo rather then repeated echo. Those elements works if they are placed in the same top level pipeline. https://bugzilla.gnome.org/show_bug.cgi?id=767800
2016-06-06 19:10:05 +00:00
G_LOCK (gst_aec_probes);
gst_aec_probes = g_list_prepend (gst_aec_probes, self);
G_UNLOCK (gst_aec_probes);
}
static void
gst_webrtc_echo_probe_class_init (GstWebrtcEchoProbeClass * klass)
{
GObjectClass *gobject_class = G_OBJECT_CLASS (klass);
GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
GstBaseTransformClass *btrans_class = GST_BASE_TRANSFORM_CLASS (klass);
GstAudioFilterClass *audiofilter_class = GST_AUDIO_FILTER_CLASS (klass);
gobject_class->finalize = gst_webrtc_echo_probe_finalize;
btrans_class->passthrough_on_same_caps = TRUE;
btrans_class->src_event = GST_DEBUG_FUNCPTR (gst_webrtc_echo_probe_src_event);
btrans_class->transform_ip =
GST_DEBUG_FUNCPTR (gst_webrtc_echo_probe_transform_ip);
btrans_class->stop = GST_DEBUG_FUNCPTR (gst_webrtc_echo_probe_stop);
audiofilter_class->setup = GST_DEBUG_FUNCPTR (gst_webrtc_echo_probe_setup);
gst_element_class_add_static_pad_template (element_class,
&gst_webrtc_echo_probe_src_template);
gst_element_class_add_static_pad_template (element_class,
&gst_webrtc_echo_probe_sink_template);
gst_element_class_set_static_metadata (element_class,
"Accoustic Echo Canceller probe",
"Generic/Audio",
"Gathers playback buffers for webrtcdsp",
"Nicolas Dufresne <nicolas.dufrsesne@collabora.com>");
}
GstWebrtcEchoProbe *
gst_webrtc_acquire_echo_probe (const gchar * name)
{
GstWebrtcEchoProbe *ret = NULL;
GList *l;
G_LOCK (gst_aec_probes);
for (l = gst_aec_probes; l; l = l->next) {
GstWebrtcEchoProbe *probe = GST_WEBRTC_ECHO_PROBE (l->data);
GST_WEBRTC_ECHO_PROBE_LOCK (probe);
if (!probe->acquired && g_strcmp0 (GST_OBJECT_NAME (probe), name) == 0) {
probe->acquired = TRUE;
ret = GST_WEBRTC_ECHO_PROBE (gst_object_ref (probe));
GST_WEBRTC_ECHO_PROBE_UNLOCK (probe);
break;
}
GST_WEBRTC_ECHO_PROBE_UNLOCK (probe);
}
G_UNLOCK (gst_aec_probes);
return ret;
}
void
gst_webrtc_release_echo_probe (GstWebrtcEchoProbe * probe)
{
GST_WEBRTC_ECHO_PROBE_LOCK (probe);
probe->acquired = FALSE;
GST_WEBRTC_ECHO_PROBE_UNLOCK (probe);
gst_object_unref (probe);
}
webrtcdsp: Rewrite echo data synchronization The previous code would run out of sync if there was packet lost or clock skews. When that happened, the echo cancellation feature would completely stop working. As this is crucial for audio calls, this patch re-implement synchronization completely. Instead of letting it drift until next discont, we now synchronize against the record data at every iteration. This way we simply never let the stream drift for longer then 10ms period. We also shorter the delay by using the latency up the probe (basically excluding the sink latency. This is a decent delay to avoid starving in the probe queue. https://bugzilla.gnome.org/show_bug.cgi?id=768009
2016-06-29 19:56:47 +00:00
gint
gst_webrtc_echo_probe_read (GstWebrtcEchoProbe * self, GstClockTime rec_time,
gpointer _frame)
{
webrtc::AudioFrame * frame = (webrtc::AudioFrame *) _frame;
GstClockTimeDiff diff;
gsize avail, skip, offset, size;
gint delay = -1;
GST_WEBRTC_ECHO_PROBE_LOCK (self);
if (!GST_CLOCK_TIME_IS_VALID (self->latency))
goto done;
webrtcdsp: Add delay-agnostic property In this mode, we let WebRTC Audio Processing figure-out the delay. This is useful when the latency reported by the stack cannot be trusted. Note that in this mode, the leaking of echo during packet lost is much worst. It is recommanded to use PLC (e.g. spanplc, or opus built-in plc). In this mode, we don't do any synchronization. Instead, we simply process all the available reverse stream data as it comes.
2016-07-14 03:00:51 +00:00
/* In delay agnostic mode, just return 10ms of data */
if (!GST_CLOCK_TIME_IS_VALID (rec_time)) {
avail = gst_adapter_available (self->adapter);
if (avail < self->period_size)
goto done;
size = self->period_size;
skip = 0;
offset = 0;
goto copy;
}
webrtcdsp: Rewrite echo data synchronization The previous code would run out of sync if there was packet lost or clock skews. When that happened, the echo cancellation feature would completely stop working. As this is crucial for audio calls, this patch re-implement synchronization completely. Instead of letting it drift until next discont, we now synchronize against the record data at every iteration. This way we simply never let the stream drift for longer then 10ms period. We also shorter the delay by using the latency up the probe (basically excluding the sink latency. This is a decent delay to avoid starving in the probe queue. https://bugzilla.gnome.org/show_bug.cgi?id=768009
2016-06-29 19:56:47 +00:00
if (gst_adapter_available (self->adapter) == 0) {
diff = G_MAXINT64;
} else {
GstClockTime play_time;
guint64 distance;
play_time = gst_adapter_prev_pts (self->adapter, &distance);
if (GST_CLOCK_TIME_IS_VALID (play_time)) {
play_time += gst_util_uint64_scale_int (distance / self->info.bpf,
GST_SECOND, self->info.rate);
play_time += self->latency;
diff = GST_CLOCK_DIFF (rec_time, play_time) / GST_MSECOND;
} else {
/* We have no timestamp, assume perfect delay */
diff = self->delay;
}
}
avail = gst_adapter_available (self->adapter);
if (diff > self->delay) {
skip = (diff - self->delay) * self->info.rate / 1000 * self->info.bpf;
skip = MIN (self->period_size, skip);
offset = 0;
} else {
skip = 0;
offset = (self->delay - diff) * self->info.rate / 1000 * self->info.bpf;
offset = MIN (avail, offset);
}
size = MIN (avail - offset, self->period_size - skip);
if (size < self->period_size)
memset (frame->data_, 0, self->period_size);
webrtcdsp: Add delay-agnostic property In this mode, we let WebRTC Audio Processing figure-out the delay. This is useful when the latency reported by the stack cannot be trusted. Note that in this mode, the leaking of echo during packet lost is much worst. It is recommanded to use PLC (e.g. spanplc, or opus built-in plc). In this mode, we don't do any synchronization. Instead, we simply process all the available reverse stream data as it comes.
2016-07-14 03:00:51 +00:00
copy:
webrtcdsp: Rewrite echo data synchronization The previous code would run out of sync if there was packet lost or clock skews. When that happened, the echo cancellation feature would completely stop working. As this is crucial for audio calls, this patch re-implement synchronization completely. Instead of letting it drift until next discont, we now synchronize against the record data at every iteration. This way we simply never let the stream drift for longer then 10ms period. We also shorter the delay by using the latency up the probe (basically excluding the sink latency. This is a decent delay to avoid starving in the probe queue. https://bugzilla.gnome.org/show_bug.cgi?id=768009
2016-06-29 19:56:47 +00:00
if (size) {
gst_adapter_copy (self->adapter, (guint8 *) frame->data_ + skip,
offset, size);
gst_adapter_flush (self->adapter, offset + size);
}
frame->num_channels_ = self->info.channels;
frame->sample_rate_hz_ = self->info.rate;
frame->samples_per_channel_ = self->period_size / self->info.bpf;
delay = self->delay;
done:
GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
return delay;
}
Reference in a new issue Copy permalink