Hi,

my hardware supports only aac lc profile. I have added that to the caps of the sink, but still the sink is used instead of faad to decode the audio.
It seems that the reason is that no profile fields are generated by matroska demux for aac audio data.
I have generated a patch which works for me. I'm no expert, so I'm not quite sure whether this patch
diff --git a/gst/matroska/matroska-demux.c b/gst/matroska/matroska-demux.c
index c586f55..6e23d75 100644
--- a/gst/matroska/matroska-demux.c
+++ b/gst/matroska/matroska-demux.c
@@ -5442,6 +5442,8 @@ gst_matroska_demux_audio_caps (GstMatroskaTrackAudioContext *
           "stream-format", G_TYPE_STRING, "raw", NULL);
       gst_caps_set_simple (caps, "codec_data", GST_TYPE_BUFFER, priv, NULL);
       *codec_name = g_strdup_printf ("MPEG-%d AAC audio", mpegversion);
+      if (context && context->codec_priv_size > 0)
+        gst_codec_utils_aac_caps_set_level_and_profile(caps, context->codec_priv, context->codec_priv_size);
       gst_buffer_unref (priv);
     }
   } else if (!strcmp (codec_id, GST_MATROSKA_CODEC_ID_AUDIO_TTA)) {

or this patch
diff --git a/gst/matroska/matroska-demux.c b/gst/matroska/matroska-demux.c
index c586f55..afc2e2c 100644
--- a/gst/matroska/matroska-demux.c
+++ b/gst/matroska/matroska-demux.c
@@ -5444,6 +5444,9 @@ gst_matroska_demux_audio_caps (GstMatroskaTrackAudioContext *
       *codec_name = g_strdup_printf ("MPEG-%d AAC audio", mpegversion);
       gst_buffer_unref (priv);
     }
+
+    if (context && context->codec_priv_size > 0)
+      gst_codec_utils_aac_caps_set_level_and_profile(caps, context->codec_priv, context->codec_priv_size);
   } else if (!strcmp (codec_id, GST_MATROSKA_CODEC_ID_AUDIO_TTA)) {
     caps = gst_caps_new_simple ("audio/x-tta",
         "width", G_TYPE_INT, audiocontext->bitdepth, NULL);

is correct.