diff --git a/src/client.rs b/src/client.rs
index 39703e55a..2f745b70c 100644
--- a/src/client.rs
+++ b/src/client.rs
@@ -1661,8 +1661,9 @@ impl LoginConfigHandler {
 
 /// Media data.
 pub enum MediaData {
-    VideoFrame,
-    AudioFrame(AudioFrame),
+    VideoQueue,
+    VideoFrame(Box<VideoFrame>),
+    AudioFrame(Box<AudioFrame>),
     AudioFormat(AudioFormat),
     Reset,
     RecordScreen(bool, i32, i32, String),
@@ -1692,7 +1693,12 @@ where
         loop {
             if let Ok(data) = video_receiver.recv() {
                 match data {
-                    MediaData::VideoFrame => {
+                    MediaData::VideoFrame(vf) => {
+                        if let Ok(true) = video_handler.handle_frame(*vf) {
+                            video_callback(&mut video_handler.rgb);
+                        }
+                    }
+                    MediaData::VideoQueue => {
                         if let Some(vf) = video_queue.pop() {
                             if let Ok(true) = video_handler.handle_frame(vf) {
                                 video_callback(&mut video_handler.rgb);
@@ -1727,7 +1733,7 @@ pub fn start_audio_thread() -> MediaSender {
             if let Ok(data) = audio_receiver.recv() {
                 match data {
                     MediaData::AudioFrame(af) => {
-                        audio_handler.handle_frame(af);
+                        audio_handler.handle_frame(*af);
                     }
                     MediaData::AudioFormat(f) => {
                         log::debug!("recved audio format, sample rate={}", f.sample_rate);
diff --git a/src/client/io_loop.rs b/src/client/io_loop.rs
index 6b0e139e2..b0bddc82e 100644
--- a/src/client/io_loop.rs
+++ b/src/client/io_loop.rs
@@ -816,6 +816,18 @@ impl<T: InvokeUiSession> Remote<T> {
         }
     }
 
+    fn contains_key_frame(vf: &VideoFrame) -> bool {
+        match &vf.union {
+            Some(vf) => match vf {
+                video_frame::Union::Vp9s(f) => f.frames.iter().any(|e| e.key),
+                video_frame::Union::H264s(f) => f.frames.iter().any(|e| e.key),
+                video_frame::Union::H265s(f) => f.frames.iter().any(|e| e.key),
+                _ => false,
+            },
+            None => false,
+        }
+    }
+
     async fn handle_msg_from_peer(&mut self, data: &[u8], peer: &mut Stream) -> bool {
         if let Ok(msg_in) = Message::parse_from_bytes(&data) {
             match msg_in.union {
@@ -834,8 +846,15 @@ impl<T: InvokeUiSession> Remote<T> {
                             ..Default::default()
                         })
                     };
-                    self.video_queue.force_push(vf);
-                    self.video_sender.send(MediaData::VideoFrame).ok();
+                    if Self::contains_key_frame(&vf) {
+                        while let Some(_) = self.video_queue.pop() {}
+                        self.video_sender
+                            .send(MediaData::VideoFrame(Box::new(vf)))
+                            .ok();
+                    } else {
+                        self.video_queue.force_push(vf);
+                        self.video_sender.send(MediaData::VideoQueue).ok();
+                    }
                 }
                 Some(message::Union::Hash(hash)) => {
                     self.handler
@@ -1222,7 +1241,9 @@ impl<T: InvokeUiSession> Remote<T> {
                 }
                 Some(message::Union::AudioFrame(frame)) => {
                     if !self.handler.lc.read().unwrap().disable_audio.v {
-                        self.audio_sender.send(MediaData::AudioFrame(frame)).ok();
+                        self.audio_sender
+                            .send(MediaData::AudioFrame(Box::new(frame)))
+                            .ok();
                     }
                 }
                 Some(message::Union::FileAction(action)) => match action.union {
diff --git a/src/server/connection.rs b/src/server/connection.rs
index 6b0cbefc3..0c17fd176 100644
--- a/src/server/connection.rs
+++ b/src/server/connection.rs
@@ -1669,7 +1669,7 @@ impl Connection {
                 Some(message::Union::AudioFrame(frame)) => {
                     if !self.disable_audio {
                         if let Some(sender) = &self.audio_sender {
-                            allow_err!(sender.send(MediaData::AudioFrame(frame)));
+                            allow_err!(sender.send(MediaData::AudioFrame(Box::new(frame))));
                         } else {
                             log::warn!(
                                 "Processing audio frame without the voice call audio sender."