fix video_mode

google-gemini · Dec 20, 2024 · 0ddb0b5 · 0ddb0b5
1 parent e9ecfaa
commit 0ddb0b5
Show file tree

Hide file tree

Showing 2 changed files with 9 additions and 7 deletions.
diff --git a/gemini-2/live_api_starter.py b/gemini-2/live_api_starter.py
@@ -244,9 +244,9 @@ async def run(self):
                 send_text_task = tg.create_task(self.send_text())
                 tg.create_task(self.send_realtime())
                 tg.create_task(self.listen_audio())
-                if self.MODE == "camera":
+                if self.video_mode == "camera":
                     tg.create_task(self.get_frames())
-                elif self.MODE == "screen":
+                elif self.video_mode == "screen":
                     tg.create_task(self.get_screen())
 
                 tg.create_task(self.receive_audio())
@@ -269,7 +269,7 @@ async def run(self):
         type=str,
         default=DEFAULT_MODE,
         help="pixels to stream from",
-        choices=["camera", "screen"],
+        choices=["camera", "screen", "none"],
     )
     args = parser.parse_args()
     main = AudioLoop(video_mode=args.mode)

diff --git a/gemini-2/websockets/live_api_starter.py b/gemini-2/websockets/live_api_starter.py
@@ -84,6 +84,7 @@
 
 class AudioLoop:
     def __init__(self, video_mode=DEFAULT_MODE):
+        self.video_mode=video_mode
         self.audio_in_queue = None
         self.out_queue = None
 
@@ -174,8 +175,9 @@ async def get_screen(self):
                 break
 
             await asyncio.sleep(1.0)
-
-            await self.out_queue.put(frame)
+
+            msg = {"realtime_input": {"media_chunks": frame}}
+            await self.out_queue.put(msg)
 
     async def send_realtime(self):
         while True:
@@ -268,9 +270,9 @@ async def run(self):
 
                 tg.create_task(self.send_realtime())
                 tg.create_task(self.listen_audio())
-                if MODE == "camera":
+                if self.video_mode == "camera":
                     tg.create_task(self.get_frames())
-                elif MODE == "screen":
+                elif self.video_mode == "screen":
                     tg.create_task(self.get_screen())
                 tg.create_task(self.receive_audio())
                 tg.create_task(self.play_audio())