diff --git a/ext/vorbis/gstvorbisenc.c b/ext/vorbis/gstvorbisenc.c
index ab309af161..eae99a1ae9 100644
--- a/ext/vorbis/gstvorbisenc.c
+++ b/ext/vorbis/gstvorbisenc.c
@@ -248,6 +248,8 @@ gst_vorbis_enc_stop (GstAudioEncoder * enc)
   g_slist_foreach (vorbisenc->headers, (GFunc) gst_buffer_unref, NULL);
   vorbisenc->headers = NULL;
 
+  gst_tag_setter_reset_tags (GST_TAG_SETTER (enc));
+
   return TRUE;
 }
 
diff --git a/gst-libs/gst/audio/gstaudiodecoder.c b/gst-libs/gst/audio/gstaudiodecoder.c
index c9a1ab6cd0..6fb3ebed51 100644
--- a/gst-libs/gst/audio/gstaudiodecoder.c
+++ b/gst-libs/gst/audio/gstaudiodecoder.c
@@ -585,47 +585,95 @@ gst_audio_decoder_setup (GstAudioDecoder * dec)
   dec->priv->agg = ! !res;
 }
 
-/* mini aggregator combining output buffers into fewer larger ones,
- * if so allowed/configured */
 static GstFlowReturn
-gst_audio_decoder_output (GstAudioDecoder * dec, GstBuffer * buf)
+gst_audio_decoder_push_forward (GstAudioDecoder * dec, GstBuffer * buf)
 {
   GstAudioDecoderClass *klass;
   GstAudioDecoderPrivate *priv;
   GstAudioDecoderContext *ctx;
   GstFlowReturn ret = GST_FLOW_OK;
-  GstBuffer *inbuf = NULL;
 
   klass = GST_AUDIO_DECODER_GET_CLASS (dec);
   priv = dec->priv;
   ctx = &dec->priv->ctx;
 
+  g_return_val_if_fail (ctx->info.bpf != 0, GST_FLOW_ERROR);
+
+  if (G_UNLIKELY (!buf)) {
+    g_assert_not_reached ();
+    return GST_FLOW_OK;
+  }
+
+  GST_LOG_OBJECT (dec, "clipping buffer of size %d with ts %" GST_TIME_FORMAT
+      ", duration %" GST_TIME_FORMAT, gst_buffer_get_size (buf),
+      GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (buf)),
+      GST_TIME_ARGS (GST_BUFFER_DURATION (buf)));
+
+  /* clip buffer */
+  buf = gst_audio_buffer_clip (buf, &dec->segment, ctx->info.rate,
+      ctx->info.bpf);
+  if (G_UNLIKELY (!buf)) {
+    GST_DEBUG_OBJECT (dec, "no data after clipping to segment");
+    goto exit;
+  }
+
+  /* decorate */
+  if (G_UNLIKELY (priv->discont)) {
+    GST_LOG_OBJECT (dec, "marking discont");
+    GST_BUFFER_FLAG_SET (buf, GST_BUFFER_FLAG_DISCONT);
+    priv->discont = FALSE;
+  }
+
+  /* track where we are */
+  if (G_LIKELY (GST_BUFFER_TIMESTAMP_IS_VALID (buf))) {
+    /* duration should always be valid for raw audio */
+    g_assert (GST_BUFFER_DURATION_IS_VALID (buf));
+    dec->segment.position =
+        GST_BUFFER_TIMESTAMP (buf) + GST_BUFFER_DURATION (buf);
+  }
+
+  if (klass->pre_push) {
+    /* last chance for subclass to do some dirty stuff */
+    ret = klass->pre_push (dec, &buf);
+    if (ret != GST_FLOW_OK || !buf) {
+      GST_DEBUG_OBJECT (dec, "subclass returned %s, buf %p",
+          gst_flow_get_name (ret), buf);
+      if (buf)
+        gst_buffer_unref (buf);
+      goto exit;
+    }
+  }
+
+  GST_LOG_OBJECT (dec, "pushing buffer of size %d with ts %" GST_TIME_FORMAT
+      ", duration %" GST_TIME_FORMAT, gst_buffer_get_size (buf),
+      GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (buf)),
+      GST_TIME_ARGS (GST_BUFFER_DURATION (buf)));
+
+  ret = gst_pad_push (dec->srcpad, buf);
+
+exit:
+  return ret;
+}
+
+/* mini aggregator combining output buffers into fewer larger ones,
+ * if so allowed/configured */
+static GstFlowReturn
+gst_audio_decoder_output (GstAudioDecoder * dec, GstBuffer * buf)
+{
+  GstAudioDecoderPrivate *priv;
+  GstFlowReturn ret = GST_FLOW_OK;
+  GstBuffer *inbuf = NULL;
+
+  priv = dec->priv;
+
   if (G_UNLIKELY (priv->agg < 0))
     gst_audio_decoder_setup (dec);
 
   if (G_LIKELY (buf)) {
-    g_return_val_if_fail (ctx->info.bpf != 0, GST_FLOW_ERROR);
-
-    GST_LOG_OBJECT (dec,
-        "output buffer of size %" G_GSIZE_FORMAT " with ts %" GST_TIME_FORMAT
+    GST_LOG_OBJECT (dec, "output buffer of size %d with ts %" GST_TIME_FORMAT
         ", duration %" GST_TIME_FORMAT, gst_buffer_get_size (buf),
         GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (buf)),
         GST_TIME_ARGS (GST_BUFFER_DURATION (buf)));
-
-    /* clip buffer */
-    buf = gst_audio_buffer_clip (buf, &dec->segment, ctx->info.rate,
-        ctx->info.bpf);
-    if (G_UNLIKELY (!buf)) {
-      GST_DEBUG_OBJECT (dec, "no data after clipping to segment");
-    } else {
-      GST_LOG_OBJECT (dec,
-          "buffer after segment clipping has size %" G_GSIZE_FORMAT " with ts %"
-          GST_TIME_FORMAT ", duration %" GST_TIME_FORMAT,
-          gst_buffer_get_size (buf), GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (buf)),
-          GST_TIME_ARGS (GST_BUFFER_DURATION (buf)));
-    }
-  } else {
-    GST_DEBUG_OBJECT (dec, "no output buffer");
   }
 
 again:
@@ -675,39 +723,8 @@ again:
   }
 
   if (G_LIKELY (buf)) {
-
-    if (G_UNLIKELY (priv->discont)) {
-      GST_LOG_OBJECT (dec, "marking discont");
-      GST_BUFFER_FLAG_SET (buf, GST_BUFFER_FLAG_DISCONT);
-      priv->discont = FALSE;
-    }
-
-    if (G_LIKELY (GST_BUFFER_TIMESTAMP_IS_VALID (buf))) {
-      /* duration should always be valid for raw audio */
-      g_assert (GST_BUFFER_DURATION_IS_VALID (buf));
-      dec->segment.position =
-          GST_BUFFER_TIMESTAMP (buf) + GST_BUFFER_DURATION (buf);
-    }
-
-    if (klass->pre_push) {
-      /* last chance for subclass to do some dirty stuff */
-      ret = klass->pre_push (dec, &buf);
-      if (ret != GST_FLOW_OK || !buf) {
-        GST_DEBUG_OBJECT (dec, "subclass returned %s, buf %p",
-            gst_flow_get_name (ret), buf);
-        if (buf)
-          gst_buffer_unref (buf);
-        goto exit;
-      }
-    }
-
-    GST_LOG_OBJECT (dec, "pushing buffer of size %d with ts %" GST_TIME_FORMAT
-        ", duration %" GST_TIME_FORMAT, gst_buffer_get_size (buf),
-        GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (buf)),
-        GST_TIME_ARGS (GST_BUFFER_DURATION (buf)));
-
     if (dec->segment.rate > 0.0) {
-      ret = gst_pad_push (dec->srcpad, buf);
+      ret = gst_audio_decoder_push_forward (dec, buf);
       GST_LOG_OBJECT (dec, "buffer pushed: %s", gst_flow_get_name (ret));
     } else {
       ret = GST_FLOW_OK;
@@ -715,7 +732,6 @@ again:
       GST_LOG_OBJECT (dec, "buffer queued");
     }
 
-  exit:
     if (inbuf) {
       buf = inbuf;
       goto again;
@@ -1202,6 +1218,7 @@ gst_audio_decoder_flush_decode (GstAudioDecoder * dec)
 {
   GstAudioDecoderPrivate *priv = dec->priv;
   GstFlowReturn res = GST_FLOW_OK;
+  GstClockTime timestamp;
   GList *walk;
 
   walk = priv->decode;
@@ -1239,9 +1256,28 @@ gst_audio_decoder_flush_decode (GstAudioDecoder * dec)
   gst_audio_decoder_drain (dec);
 
   /* now send queued data downstream */
+  timestamp = GST_CLOCK_TIME_NONE;
   while (priv->queued) {
     GstBuffer *buf = GST_BUFFER_CAST (priv->queued->data);
 
+    /* duration should always be valid for raw audio */
+    g_assert (GST_BUFFER_DURATION_IS_VALID (buf));
+
+    /* interpolate (backward) if needed */
+    if (G_LIKELY (timestamp != -1))
+      timestamp -= GST_BUFFER_DURATION (buf);
+
+    if (!GST_BUFFER_TIMESTAMP_IS_VALID (buf)) {
+      GST_LOG_OBJECT (dec, "applying reverse interpolated ts %"
+          GST_TIME_FORMAT, GST_TIME_ARGS (timestamp));
+      GST_BUFFER_TIMESTAMP (buf) = timestamp;
+    } else {
+      /* track otherwise */
+      timestamp = GST_BUFFER_TIMESTAMP (buf);
+      GST_LOG_OBJECT (dec, "tracking ts %" GST_TIME_FORMAT,
+          GST_TIME_ARGS (timestamp));
+    }
+
     if (G_LIKELY (res == GST_FLOW_OK)) {
       GST_DEBUG_OBJECT (dec, "pushing buffer %p of size %u, "
           "time %" GST_TIME_FORMAT ", dur %" GST_TIME_FORMAT, buf,
@@ -1252,7 +1288,7 @@ gst_audio_decoder_flush_decode (GstAudioDecoder * dec)
       /* avoid stray DISCONT from forward processing,
        * which have no meaning in reverse pushing */
       GST_BUFFER_FLAG_UNSET (buf, GST_BUFFER_FLAG_DISCONT);
-      res = gst_pad_push (dec->srcpad, buf);
+      res = gst_audio_decoder_push_forward (dec, buf);
     } else {
       gst_buffer_unref (buf);
     }
diff --git a/gst-libs/gst/audio/gstaudioencoder.c b/gst-libs/gst/audio/gstaudioencoder.c
index 485087228f..81f61b5f1b 100644
--- a/gst-libs/gst/audio/gstaudioencoder.c
+++ b/gst-libs/gst/audio/gstaudioencoder.c
@@ -1087,6 +1087,13 @@ gst_audio_encoder_sink_setcaps (GstAudioEncoder * enc, GstCaps * caps)
     if (res)
       ctx->info = state;
 
+    /* invalidate state to ensure no casual carrying on */
+    if (!res) {
+      GST_DEBUG_OBJECT (enc, "subclass did not accept format");
+      gst_audio_info_init (&state);
+      goto exit;
+    }
+
     /* notify if new latency */
     GST_OBJECT_LOCK (enc);
     if ((ctx->min_latency > 0 && ctx->min_latency != old_min_latency) ||
diff --git a/gst-libs/gst/rtp/gstrtcpbuffer.h b/gst-libs/gst/rtp/gstrtcpbuffer.h
index 2a9d41ae3d..da8dce26d1 100644
--- a/gst-libs/gst/rtp/gstrtcpbuffer.h
+++ b/gst-libs/gst/rtp/gstrtcpbuffer.h
@@ -63,10 +63,17 @@ typedef enum
  * GstRTCPFBType:
  * @GST_RTCP_FB_TYPE_INVALID: Invalid type
  * @GST_RTCP_RTPFB_TYPE_NACK: Generic NACK
+ * @GST_RTCP_RTPFB_TYPE_TMMBR: Temporary Maximum Media Stream Bit Rate Request
+ * @GST_RTCP_RTPFB_TYPE_TMMBN: Temporary Maximum Media Stream Bit Rate
+ *    Notification
  * @GST_RTCP_PSFB_TYPE_PLI: Picture Loss Indication
  * @GST_RTCP_PSFB_TYPE_SLI: Slice Loss Indication
  * @GST_RTCP_PSFB_TYPE_RPSI: Reference Picture Selection Indication
  * @GST_RTCP_PSFB_TYPE_AFB: Application layer Feedback
+ * @GST_RTCP_PSFB_TYPE_FIR: Full Intra Request Command
+ * @GST_RTCP_PSFB_TYPE_TSTR: Temporal-Spatial Trade-off Request
+ * @GST_RTCP_PSFB_TYPE_TSTN: Temporal-Spatial Trade-off Notification
+ * @GST_RTCP_PSFB_TYPE_VBCN: Video Back Channel Message
  *
  * Different types of feedback messages.
  *
@@ -78,11 +85,19 @@ typedef enum
   GST_RTCP_FB_TYPE_INVALID    = 0,
   /* RTPFB types */
   GST_RTCP_RTPFB_TYPE_NACK    = 1,
+  /* RTPFB types assigned in RFC 5104 */
+  GST_RTCP_RTPFB_TYPE_TMMBR   = 3,
+  GST_RTCP_RTPFB_TYPE_TMMBN   = 4,
   /* PSFB types */
   GST_RTCP_PSFB_TYPE_PLI      = 1,
   GST_RTCP_PSFB_TYPE_SLI      = 2,
   GST_RTCP_PSFB_TYPE_RPSI     = 3,
-  GST_RTCP_PSFB_TYPE_AFB      = 15
+  GST_RTCP_PSFB_TYPE_AFB      = 15,
+  /* PSFB types assigned in RFC 5104 */
+  GST_RTCP_PSFB_TYPE_FIR      = 4,
+  GST_RTCP_PSFB_TYPE_TSTR     = 5,
+  GST_RTCP_PSFB_TYPE_TSTN     = 6,
+  GST_RTCP_PSFB_TYPE_VBCN     = 7,
 } GstRTCPFBType;
 
 /**