diff --git a/daemon/call_interfaces.c b/daemon/call_interfaces.c
index 6fdeecdb8..d917b039e 100644
--- a/daemon/call_interfaces.c
+++ b/daemon/call_interfaces.c
@@ -1778,6 +1778,9 @@ const char *call_play_media_ng(bencode_item_t *input, bencode_item_t *output) {
 	else
 		goto out;
 
+	if (monologue->player->duration)
+		bencode_dictionary_add_integer(output, "duration", monologue->player->duration);
+
 	err = NULL;
 
 out:
diff --git a/daemon/media_player.c b/daemon/media_player.c
index a4a4312cb..f7427e504 100644
--- a/daemon/media_player.c
+++ b/daemon/media_player.c
@@ -220,6 +220,8 @@ found:
 	if (!mp->ssrc_out)
 		return -1;
 
+	mp->duration = avs->duration * 1000 * avs->time_base.num / avs->time_base.den;
+
 	return 0;
 }
 
@@ -334,6 +336,9 @@ found:
 
 // call->master_lock held in W
 static void media_player_play_start(struct media_player *mp) {
+	// needed to have usable duration for some formats. ignore errors.
+	avformat_find_stream_info(mp->fmtctx, NULL);
+
 	mp->next_run = rtpe_now;
 	// give ourselves a bit of a head start with decoding
 	timeval_add_usec(&mp->next_run, -50000);
diff --git a/include/media_player.h b/include/media_player.h
index d08bceaf2..8546447e7 100644
--- a/include/media_player.h
+++ b/include/media_player.h
@@ -33,6 +33,7 @@ struct media_player {
 	struct timeval next_run;
 
 	AVFormatContext *fmtctx;
+	unsigned long duration; // in milliseconds
 	AVPacket pkt;
 	struct codec_handler *handler;
 	struct ssrc_ctx *ssrc_out;