webrtcdsp: add support for using F32/non-interleaved buffers

This is the native format that is in use by the webrtc audio processing
library internally, so this avoids internal {de,}interleaving and
format conversion (S16->F32 and back)

https://bugzilla.gnome.org/show_bug.cgi?id=793605

6 files changed, 247 insertions, 57 deletions

diff --git a/ext/webrtcdsp/Makefile.am b/ext/webrtcdsp/Makefile.am
index 63ebb45a4..16e829ccd 100644
--- a/ext/webrtcdsp/Makefile.am
+++ b/ext/webrtcdsp/Makefile.am
@@ -6,9 +6,12 @@ libgstwebrtcdsp_la_CXXFLAGS = \
 	$(GST_CXXFLAGS) \
 	$(GST_BASE_CFLAGS) \
 	$(GST_PLUGINS_BASE_CFLAGS) \
+	-I$(top_srcdir)/gst-libs \
+	-I$(top_builddir)/gst-libs \
 	$(WEBRTCDSP_CFLAGS)
 libgstwebrtcdsp_la_LIBADD = \
 	-lgstaudio-$(GST_API_VERSION) \
+	$(top_builddir)/gst-libs/gst/audio/libgstbadaudio-$(GST_API_VERSION).la \
 	$(GST_LIBS) $(GST_BASE_LIBS) \
 	$(GST_PLUGINS_BASE_LIBS) \
 	$(WEBRTCDSP_LIBS)
diff --git a/ext/webrtcdsp/gstwebrtcdsp.cpp b/ext/webrtcdsp/gstwebrtcdsp.cpp
index c92072ae2..9d7c8d2a3 100644
--- a/ext/webrtcdsp/gstwebrtcdsp.cpp
+++ b/ext/webrtcdsp/gstwebrtcdsp.cpp
@@ -95,6 +95,11 @@ GST_STATIC_PAD_TEMPLATE ("sink",
         "format = (string) " GST_AUDIO_NE (S16) ", "
         "layout = (string) interleaved, "
         "rate = (int) { 48000, 32000, 16000, 8000 }, "
+        "channels = (int) [1, MAX];"
+        "audio/x-raw, "
+        "format = (string) " GST_AUDIO_NE (F32) ", "
+        "layout = (string) non-interleaved, "
+        "rate = (int) { 48000, 32000, 16000, 8000 }, "
         "channels = (int) [1, MAX]")
     );
 
@@ -106,6 +111,11 @@ GST_STATIC_PAD_TEMPLATE ("src",
         "format = (string) " GST_AUDIO_NE (S16) ", "
         "layout = (string) interleaved, "
         "rate = (int) { 48000, 32000, 16000, 8000 }, "
+        "channels = (int) [1, MAX];"
+        "audio/x-raw, "
+        "format = (string) " GST_AUDIO_NE (F32) ", "
+        "layout = (string) non-interleaved, "
+        "rate = (int) { 48000, 32000, 16000, 8000 }, "
         "channels = (int) [1, MAX]")
     );
 
@@ -230,11 +240,14 @@ struct _GstWebrtcDsp
 
   /* Protected by the object lock */
   GstAudioInfo info;
+  gboolean interleaved;
   guint period_size;
+  guint period_samples;
   gboolean stream_has_voice;
 
   /* Protected by the stream lock */
   GstAdapter *adapter;
+  GstPlanarAudioAdapter *padapter;
   webrtc::AudioProcessing * apm;
 
   /* Protected by the object lock */
@@ -321,20 +334,35 @@ gst_webrtc_dsp_take_buffer (GstWebrtcDsp * self)
   GstBuffer *buffer;
   GstClockTime timestamp;
   guint64 distance;
+  gboolean at_discont;
 
-  timestamp = gst_adapter_prev_pts (self->adapter, &distance);
-  timestamp += gst_util_uint64_scale_int (distance / self->info.bpf,
-      GST_SECOND, self->info.rate);
+  if (self->interleaved) {
+    timestamp = gst_adapter_prev_pts (self->adapter, &distance);
+    distance /= self->info.bpf;
+  } else {
+    timestamp = gst_planar_audio_adapter_prev_pts (self->padapter, &distance);
+  }
 
-  buffer = gst_adapter_take_buffer (self->adapter, self->period_size);
+  timestamp += gst_util_uint64_scale_int (distance, GST_SECOND, self->info.rate);
+
+  if (self->interleaved) {
+    buffer = gst_adapter_take_buffer (self->adapter, self->period_size);
+    at_discont = (gst_adapter_pts_at_discont (self->adapter) == timestamp);
+  } else {
+    buffer = gst_planar_audio_adapter_take_buffer (self->padapter,
+        self->period_samples, GST_MAP_READWRITE);
+    at_discont =
+        (gst_planar_audio_adapter_pts_at_discont (self->padapter) == timestamp);
+  }
 
   GST_BUFFER_PTS (buffer) = timestamp;
   GST_BUFFER_DURATION (buffer) = 10 * GST_MSECOND;
 
-  if (gst_adapter_pts_at_discont (self->adapter) == timestamp && distance == 0) {
+  if (at_discont && distance == 0) {
     GST_BUFFER_FLAG_SET (buffer, GST_BUFFER_FLAG_DISCONT);
-  } else
+  } else {
     GST_BUFFER_FLAG_UNSET (buffer, GST_BUFFER_FLAG_DISCONT);
+  }
 
   return buffer;
 }
@@ -346,6 +374,7 @@ gst_webrtc_dsp_analyze_reverse_stream (GstWebrtcDsp * self,
   GstWebrtcEchoProbe *probe = NULL;
   webrtc::AudioProcessing * apm;
   webrtc::AudioFrame frame;
+  GstBuffer *buf = NULL;
   GstFlowReturn ret = GST_FLOW_OK;
   gint err, delay;
 
@@ -364,7 +393,7 @@ gst_webrtc_dsp_analyze_reverse_stream (GstWebrtcDsp * self,
     rec_time = GST_CLOCK_TIME_NONE;
 
 again:
-  delay = gst_webrtc_echo_probe_read (probe, rec_time, (gpointer) &frame);
+  delay = gst_webrtc_echo_probe_read (probe, rec_time, (gpointer) &frame, &buf);
   apm->set_stream_delay_ms (delay);
 
   if (delay < 0)
@@ -379,15 +408,31 @@ again:
     goto done;
   }
 
-  if ((err = apm->AnalyzeReverseStream (&frame)) < 0)
-    GST_WARNING_OBJECT (self, "Reverse stream analyses failed: %s.",
-        webrtc_error_to_string (err));
+  if (buf) {
+    webrtc::StreamConfig config (frame.sample_rate_hz_, frame.num_channels_,
+        false);
+    GstAudioBuffer abuf;
+    float * const * data;
+
+    gst_audio_buffer_map (&abuf, &self->info, buf, GST_MAP_READWRITE);
+    data = (float * const *) abuf.planes;
+    if ((err = apm->ProcessReverseStream (data, config, config, data)) < 0)
+      GST_WARNING_OBJECT (self, "Reverse stream analyses failed: %s.",
+          webrtc_error_to_string (err));
+    gst_audio_buffer_unmap (&abuf);
+    gst_buffer_replace (&buf, NULL);
+  } else {
+    if ((err = apm->AnalyzeReverseStream (&frame)) < 0)
+      GST_WARNING_OBJECT (self, "Reverse stream analyses failed: %s.",
+          webrtc_error_to_string (err));
+  }
 
   if (self->delay_agnostic)
       goto again;
 
 done:
   gst_object_unref (probe);
+  gst_buffer_replace (&buf, NULL);
 
   return ret;
 }
@@ -418,23 +463,34 @@ static GstFlowReturn
 gst_webrtc_dsp_process_stream (GstWebrtcDsp * self,
     GstBuffer * buffer)
 {
-  GstMapInfo info;
+  GstAudioBuffer abuf;
   webrtc::AudioProcessing * apm = self->apm;
-  webrtc::AudioFrame frame;
   gint err;
 
-  frame.num_channels_ = self->info.channels;
-  frame.sample_rate_hz_ = self->info.rate;
-  frame.samples_per_channel_ = self->period_size / self->info.bpf;
-
-  if (!gst_buffer_map (buffer, &info, (GstMapFlags) GST_MAP_READWRITE)) {
+  if (!gst_audio_buffer_map (&abuf, &self->info, buffer,
+          (GstMapFlags) GST_MAP_READWRITE)) {
     gst_buffer_unref (buffer);
     return GST_FLOW_ERROR;
   }
 
-  memcpy (frame.data_, info.data, self->period_size);
+  if (self->interleaved) {
+    webrtc::AudioFrame frame;
+    frame.num_channels_ = self->info.channels;
+    frame.sample_rate_hz_ = self->info.rate;
+    frame.samples_per_channel_ = self->period_samples;
+
+    memcpy (frame.data_, abuf.planes[0], self->period_size);
+    err = apm->ProcessStream (&frame);
+    if (err >= 0)
+      memcpy (abuf.planes[0], frame.data_, self->period_size);
+  } else {
+    float * const * data = (float * const *) abuf.planes;
+    webrtc::StreamConfig config (self->info.rate, self->info.channels, false);
+
+    err = apm->ProcessStream (data, config, config, data);
+  }
 
-  if ((err = apm->ProcessStream (&frame)) < 0) {
+  if (err < 0) {
     GST_WARNING_OBJECT (self, "Failed to filter the audio: %s.",
         webrtc_error_to_string (err));
   } else {
@@ -446,10 +502,9 @@ gst_webrtc_dsp_process_stream (GstWebrtcDsp * self,
 
       self->stream_has_voice = stream_has_voice;
     }
-    memcpy (info.data, frame.data_, self->period_size);
   }
 
-  gst_buffer_unmap (buffer, &info);
+  gst_audio_buffer_unmap (&abuf);
 
   return GST_FLOW_OK;
 }
@@ -467,10 +522,16 @@ gst_webrtc_dsp_submit_input_buffer (GstBaseTransform * btrans,
   if (is_discont) {
     GST_DEBUG_OBJECT (self,
         "Received discont, clearing adapter.");
-    gst_adapter_clear (self->adapter);
+    if (self->interleaved)
+      gst_adapter_clear (self->adapter);
+    else
+      gst_planar_audio_adapter_clear (self->padapter);
   }
 
-  gst_adapter_push (self->adapter, buffer);
+  if (self->interleaved)
+    gst_adapter_push (self->adapter, buffer);
+  else
+    gst_planar_audio_adapter_push (self->padapter, buffer);
 
   return GST_FLOW_OK;
 }
@@ -480,8 +541,15 @@ gst_webrtc_dsp_generate_output (GstBaseTransform * btrans, GstBuffer ** outbuf)
 {
   GstWebrtcDsp *self = GST_WEBRTC_DSP (btrans);
   GstFlowReturn ret;
+  gboolean not_enough;
+
+  if (self->interleaved)
+    not_enough = gst_adapter_available (self->adapter) < self->period_size;
+  else
+    not_enough = gst_planar_audio_adapter_available (self->padapter) <
+        self->period_samples;
 
-  if (gst_adapter_available (self->adapter) < self->period_size) {
+  if (not_enough) {
     *outbuf = NULL;
     return GST_FLOW_OK;
   }
@@ -545,13 +613,21 @@ gst_webrtc_dsp_setup (GstAudioFilter * filter, const GstAudioInfo * info)
   GST_OBJECT_LOCK (self);
 
   gst_adapter_clear (self->adapter);
+  gst_planar_audio_adapter_clear (self->padapter);
+
   self->info = *info;
+  self->interleaved = (info->layout == GST_AUDIO_LAYOUT_INTERLEAVED);
   apm = self->apm;
 
+  if (!self->interleaved)
+    gst_planar_audio_adapter_configure (self->padapter, info);
+
   /* WebRTC library works with 10ms buffers, compute once this size */
-  self->period_size = info->bpf * info->rate / 100;
+  self->period_samples = info->rate / 100;
+  self->period_size = self->period_samples * info->bpf;
 
-  if ((webrtc::AudioFrame::kMaxDataSizeSamples * 2) < self->period_size)
+  if (self->interleaved &&
+      (webrtc::AudioFrame::kMaxDataSizeSamples * 2) < self->period_size)
     goto period_too_big;
 
   if (self->probe) {
@@ -676,6 +752,7 @@ gst_webrtc_dsp_stop (GstBaseTransform * btrans)
   GST_OBJECT_LOCK (self);
 
   gst_adapter_clear (self->adapter);
+  gst_planar_audio_adapter_clear (self->padapter);
 
   if (self->probe) {
     gst_webrtc_release_echo_probe (self->probe);
@@ -840,6 +917,7 @@ gst_webrtc_dsp_finalize (GObject * object)
   GstWebrtcDsp *self = GST_WEBRTC_DSP (object);
 
   gst_object_unref (self->adapter);
+  gst_object_unref (self->padapter);
   g_free (self->probe_name);
 
   G_OBJECT_CLASS (gst_webrtc_dsp_parent_class)->finalize (object);
@@ -849,6 +927,7 @@ static void
 gst_webrtc_dsp_init (GstWebrtcDsp * self)
 {
   self->adapter = gst_adapter_new ();
+  self->padapter = gst_planar_audio_adapter_new ();
   gst_audio_info_init (&self->info);
 }
 
diff --git a/ext/webrtcdsp/gstwebrtcdsp.h b/ext/webrtcdsp/gstwebrtcdsp.h
index a2519dfac..e0e808f13 100644
--- a/ext/webrtcdsp/gstwebrtcdsp.h
+++ b/ext/webrtcdsp/gstwebrtcdsp.h
@@ -28,6 +28,9 @@
 #include <gst/base/gstbasetransform.h>
 #include <gst/audio/audio.h>
 
+#define GST_USE_UNSTABLE_API
+#include <gst/audio/gstplanaraudioadapter.h>
+
 G_BEGIN_DECLS
 
 #define GST_TYPE_WEBRTC_DSP            (gst_webrtc_dsp_get_type())
diff --git a/ext/webrtcdsp/gstwebrtcechoprobe.cpp b/ext/webrtcdsp/gstwebrtcechoprobe.cpp
index a34fa5201..ad350e78b 100644
--- a/ext/webrtcdsp/gstwebrtcechoprobe.cpp
+++ b/ext/webrtcdsp/gstwebrtcechoprobe.cpp
@@ -49,6 +49,11 @@ GST_STATIC_PAD_TEMPLATE ("sink",
         "format = (string) " GST_AUDIO_NE (S16) ", "
         "layout = (string) interleaved, "
         "rate = (int) { 48000, 32000, 16000, 8000 }, "
+        "channels = (int) [1, MAX];"
+        "audio/x-raw, "
+        "format = (string) " GST_AUDIO_NE (F32) ", "
+        "layout = (string) non-interleaved, "
+        "rate = (int) { 48000, 32000, 16000, 8000 }, "
         "channels = (int) [1, MAX]")
     );
 
@@ -60,6 +65,11 @@ GST_STATIC_PAD_TEMPLATE ("src",
         "format = (string) " GST_AUDIO_NE (S16) ", "
         "layout = (string) interleaved, "
         "rate = (int) { 48000, 32000, 16000, 8000 }, "
+        "channels = (int) [1, MAX];"
+        "audio/x-raw, "
+        "format = (string) " GST_AUDIO_NE (F32) ", "
+        "layout = (string) non-interleaved, "
+        "rate = (int) { 48000, 32000, 16000, 8000 }, "
         "channels = (int) [1, MAX]")
     );
 
@@ -80,11 +90,17 @@ gst_webrtc_echo_probe_setup (GstAudioFilter * filter, const GstAudioInfo * info)
   GST_WEBRTC_ECHO_PROBE_LOCK (self);
 
   self->info = *info;
+  self->interleaved = (info->layout == GST_AUDIO_LAYOUT_INTERLEAVED);
+
+  if (!self->interleaved)
+    gst_planar_audio_adapter_configure (self->padapter, info);
 
   /* WebRTC library works with 10ms buffers, compute once this size */
-  self->period_size = info->bpf * info->rate / 100;
+  self->period_samples = info->rate / 100;
+  self->period_size = self->period_samples * info->bpf;
 
-  if ((webrtc::AudioFrame::kMaxDataSizeSamples * 2) < self->period_size)
+  if (self->interleaved &&
+      (webrtc::AudioFrame::kMaxDataSizeSamples * 2) < self->period_size)
     goto period_too_big;
 
   GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
@@ -107,6 +123,7 @@ gst_webrtc_echo_probe_stop (GstBaseTransform * btrans)
 
   GST_WEBRTC_ECHO_PROBE_LOCK (self);
   gst_adapter_clear (self->adapter);
+  gst_planar_audio_adapter_clear (self->padapter);
   GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
 
   return TRUE;
@@ -163,11 +180,24 @@ gst_webrtc_echo_probe_transform_ip (GstBaseTransform * btrans,
   /* 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);
+  if (self->interleaved) {
+    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);
+  } else {
+    gsize available;
+
+    gst_planar_audio_adapter_push (self->padapter, newbuf);
+    available =
+        gst_planar_audio_adapter_available (self->padapter) * self->info.bpf;
+    if (available > MAX_ADAPTER_SIZE)
+      gst_planar_audio_adapter_flush (self->padapter,
+          (available - MAX_ADAPTER_SIZE) / self->info.bpf);
+  }
+
   GST_WEBRTC_ECHO_PROBE_UNLOCK (self);
 
   return GST_FLOW_OK;
@@ -183,7 +213,9 @@ gst_webrtc_echo_probe_finalize (GObject * object)
   G_UNLOCK (gst_aec_probes);
 
   gst_object_unref (self->adapter);
+  gst_object_unref (self->padapter);
   self->adapter = NULL;
+  self->padapter = NULL;
 
   G_OBJECT_CLASS (gst_webrtc_echo_probe_parent_class)->finalize (object);
 }
@@ -192,6 +224,7 @@ static void
 gst_webrtc_echo_probe_init (GstWebrtcEchoProbe * self)
 {
   self->adapter = gst_adapter_new ();
+  self->padapter = gst_planar_audio_adapter_new ();
   gst_audio_info_init (&self->info);
   g_mutex_init (&self->lock);
 
@@ -268,7 +301,7 @@ gst_webrtc_release_echo_probe (GstWebrtcEchoProbe * probe)
 
 gint
 gst_webrtc_echo_probe_read (GstWebrtcEchoProbe * self, GstClockTime rec_time,
-    gpointer _frame)
+    gpointer _frame, GstBuffer ** buf)
 {
   webrtc::AudioFrame * frame = (webrtc::AudioFrame *) _frame;
   GstClockTimeDiff diff;
@@ -281,31 +314,39 @@ gst_webrtc_echo_probe_read (GstWebrtcEchoProbe * self, GstClockTime rec_time,
       !GST_AUDIO_INFO_IS_VALID (&self->info))
     goto done;
 
+  if (self->interleaved)
+    avail = gst_adapter_available (self->adapter) / self->info.bpf;
+  else
+    avail = gst_planar_audio_adapter_available (self->padapter);
+
   /* 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)
+    if (avail < self->period_samples)
       goto done;
 
-    size = self->period_size;
+    size = self->period_samples;
     skip = 0;
     offset = 0;
 
     goto copy;
   }
 
-  if (gst_adapter_available (self->adapter) == 0) {
+  if (avail == 0) {
     diff = G_MAXINT64;
   } else {
     GstClockTime play_time;
     guint64 distance;
 
-    play_time = gst_adapter_prev_pts (self->adapter, &distance);
+    if (self->interleaved) {
+      play_time = gst_adapter_prev_pts (self->adapter, &distance);
+      distance /= self->info.bpf;
+    } else {
+      play_time = gst_planar_audio_adapter_prev_pts (self->padapter, &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 += gst_util_uint64_scale_int (distance, GST_SECOND,
+          self->info.rate);
       play_time += self->latency;
 
       diff = GST_CLOCK_DIFF (rec_time, play_time) / GST_MSECOND;
@@ -315,33 +356,91 @@ gst_webrtc_echo_probe_read (GstWebrtcEchoProbe * self, GstClockTime rec_time,
     }
   }
 
-  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);
+    skip = (diff - self->delay) * self->info.rate / 1000;
+    skip = MIN (self->period_samples, skip);
     offset = 0;
   } else {
     skip = 0;
-    offset = (self->delay - diff) * self->info.rate / 1000 * self->info.bpf;
+    offset = (self->delay - diff) * self->info.rate / 1000;
     offset = MIN (avail, offset);
   }
 
-  size = MIN (avail - offset, self->period_size - skip);
-
-  if (size < self->period_size)
-    memset (frame->data_, 0, self->period_size);
+  size = MIN (avail - offset, self->period_samples - skip);
 
 copy:
-  if (size) {
-    gst_adapter_copy (self->adapter, (guint8 *) frame->data_ + skip,
-        offset, size);
-    gst_adapter_flush (self->adapter, offset + size);
+  if (self->interleaved) {
+    skip *= self->info.bpf;
+    offset *= self->info.bpf;
+    size *= self->info.bpf;
+
+    if (size < self->period_size)
+      memset (frame->data_, 0, self->period_size);
+
+    if (size) {
+      gst_adapter_copy (self->adapter, (guint8 *) frame->data_ + skip,
+          offset, size);
+      gst_adapter_flush (self->adapter, offset + size);
+    }
+  } else {
+    GstBuffer *ret, *taken, *tmp;
+
+    if (size) {
+      gst_planar_audio_adapter_flush (self->padapter, offset);
+
+      /* we need to fill silence at the beginning and/or the end of each
+       * channel plane in order to have exactly period_samples in the buffer */
+      if (size < self->period_samples) {
+        GstAudioMeta *meta;
+        gint bps = self->info.finfo->width / 8;
+        gsize padding = self->period_samples - (skip + size);
+        gint c;
+
+        taken = gst_planar_audio_adapter_take_buffer (self->padapter, size,
+            GST_MAP_READ);
+        meta = gst_buffer_get_audio_meta (taken);
+        ret = gst_buffer_new ();
+
+        for (c = 0; c < meta->info.channels; c++) {
+          /* need some silence at the beginning */
+          if (skip) {
+            tmp = gst_buffer_new_allocate (NULL, skip * bps, NULL);
+            gst_buffer_memset (tmp, 0, 0, skip * bps);
+            ret = gst_buffer_append (ret, tmp);
+          }
+
+          tmp = gst_buffer_copy_region (taken, GST_BUFFER_COPY_MEMORY,
+              meta->offsets[c], size * bps);
+          ret = gst_buffer_append (ret, tmp);
+
+          /* need some silence at the end */
+          if (padding) {
+            tmp = gst_buffer_new_allocate (NULL, padding * bps, NULL);
+            gst_buffer_memset (tmp, 0, 0, padding * bps);
+            ret = gst_buffer_append (ret, tmp);
+          }
+        }
+
+        gst_buffer_unref (taken);
+        gst_buffer_add_audio_meta (ret, &self->info, self->period_samples,
+            NULL);
+      } else {
+        ret = gst_planar_audio_adapter_take_buffer (self->padapter, size,
+          GST_MAP_READWRITE);
+      }
+    } else {
+      ret = gst_buffer_new_allocate (NULL, self->period_size, NULL);
+      gst_buffer_memset (ret, 0, 0, self->period_size);
+      gst_buffer_add_audio_meta (ret, &self->info, self->period_samples,
+          NULL);
+    }
+
+    *buf = ret;
   }
 
   frame->num_channels_ = self->info.channels;
   frame->sample_rate_hz_ = self->info.rate;
-  frame->samples_per_channel_ = self->period_size / self->info.bpf;
+  frame->samples_per_channel_ = self->period_samples;
 
   delay = self->delay;
 
diff --git a/ext/webrtcdsp/gstwebrtcechoprobe.h b/ext/webrtcdsp/gstwebrtcechoprobe.h
index 7c8a24022..a8a889907 100644
--- a/ext/webrtcdsp/gstwebrtcechoprobe.h
+++ b/ext/webrtcdsp/gstwebrtcechoprobe.h
@@ -28,6 +28,9 @@
 #include <gst/base/gstbasetransform.h>
 #include <gst/audio/audio.h>
 
+#define GST_USE_UNSTABLE_API
+#include <gst/audio/gstplanaraudioadapter.h>
+
 G_BEGIN_DECLS
 
 #define GST_TYPE_WEBRTC_ECHO_PROBE            (gst_webrtc_echo_probe_get_type())
@@ -62,11 +65,14 @@ struct _GstWebrtcEchoProbe
   /* Protected by the lock */
   GstAudioInfo info;
   guint period_size;
+  guint period_samples;
   GstClockTime latency;
   gint delay;
+  gboolean interleaved;
 
   GstSegment segment;
   GstAdapter *adapter;
+  GstPlanarAudioAdapter *padapter;
 
   /* Private */
   gboolean acquired;
@@ -82,7 +88,7 @@ GType gst_webrtc_echo_probe_get_type (void);
 GstWebrtcEchoProbe *gst_webrtc_acquire_echo_probe (const gchar * name);
 void gst_webrtc_release_echo_probe (GstWebrtcEchoProbe * probe);
 gint gst_webrtc_echo_probe_read (GstWebrtcEchoProbe * self,
-    GstClockTime rec_time, gpointer frame);
+    GstClockTime rec_time, gpointer frame, GstBuffer ** buf);
 
 G_END_DECLS
 #endif /* __GST_WEBRTC_ECHO_PROBE_H__ */
diff --git a/ext/webrtcdsp/meson.build b/ext/webrtcdsp/meson.build
index 9d02a7faf..678d128bd 100644
--- a/ext/webrtcdsp/meson.build
+++ b/ext/webrtcdsp/meson.build
@@ -12,7 +12,7 @@ if webrtc_dep.found()
     cpp_args : gst_plugins_bad_args,
     link_args : noseh_link_args,
     include_directories : [configinc],
-    dependencies : [gstbase_dep, gstaudio_dep, webrtc_dep],
+    dependencies : [gstbase_dep, gstaudio_dep, gstbadaudio_dep, webrtc_dep],
     install : true,
     install_dir : plugins_install_dir,
   )