Update app.py

app.py

@@ -203,8 +203,10 @@ def _generate(message: dict, history: list[dict], system_prompt: str = "", max_n
 @spaces.GPU(time_limit=120)
 def generate(data: WebRTCData, history: list[dict], system_prompt: str = "", max_new_tokens: int = 512, image=None):
     files = []
+    audio_file = None
     if data.audio is not None and data.audio[1].size > 0:
-        files.append(save_audio_to_cache(data.audio[1], data.audio[0], format="mp3", cache_dir=get_upload_folder()))
+        audio_file = save_audio_to_cache(data.audio[1], data.audio[0], format="mp3", cache_dir=get_upload_folder())
+        files.append(audio_file)
     if image is not None:
         files.append(image)
     message = {
@@ -212,6 +214,10 @@ def generate(data: WebRTCData, history: list[dict], system_prompt: str = "", max
         "files": files,
     }
     print("message", message)
+    if image is not None:
+        history.append({"role": "user", "content": (image,)})
+    if audio_file is not None:
+        history.append({"role": "user", "content": (audio_file,)})
     history.append({"role": "user", "content": data.textbox})
     print("history", history)
     yield AdditionalOutputs(history)