Update app.py

app.py

@@ -1,41 +1,16 @@
 import gradio as gr
 import cv2
-import tempfile
 import numpy as np
 import os
-import time
 import requests
 import base64
-import threading
-import pygame
+import base64
+import io
+import soundfile as sf
 
 # Backend server URL
 backend_server_url = "https://0416-2600-1017-a410-36b8-2357-52be-1318-959b.ngrok-free.app"
 
-send_thread = None # To keep track of ongoing threads
-
-# Audio playback
-def play_audio(audio_base64):
-    """
-    Play audio file using pygame mixer.
-    
-    Args:
-        audio_path: Path to audio file
-    """
-    audio_bytes = base64.b64decode(audio_base64)
-    try:
-        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
-            temp_audio.write(audio_bytes)
-            temp_audio_path = temp_audio.name
-
-        pygame.mixer.init()
-        pygame.mixer.music.load(temp_audio_path)
-        pygame.mixer.music.play()
-        while pygame.mixer.music.get_busy():
-            pygame.time.Clock().tick(10)
-    except Exception as e:
-        print(f"Audio error: {e}")
-
 # Backend interaction
 def send_to_backend(frame):
     try:
@@ -64,29 +39,24 @@ def send_to_backend(frame):
     except Exception as e:
         return {"error": f"Exception: {str(e)}"}
 
-def thread_sendToBackend(frame):
-    """ Starts a thread to send the frame to the backend. """
-    global send_thread
-    if send_thread is None:
-        send_thread = threading.Thread(target=send_to_backend, args=(frame,), daemon=True)
-        send_thread.start()
-
-
 # # Gradio processing function
 def process_webcam(image):
     if image is None:
-        return None, "No frame", None
+        return None, None
 
     frame = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
     result = send_to_backend(frame)
-    print(len(result))
-    caption = result['caption']
-    audio_base64 = result['audio_base64']
+
+    caption = result.get("caption", "No caption")
+    audio_base64 = result.get("audio_base64", None)
 
     if audio_base64:
-        threading.Thread(target=play_audio, args=(audio_base64,), daemon=True).start()
+        audio_bytes = base64.b64decode(audio_base64)
+        audio_buffer = io.BytesIO(audio_bytes)
+        audio_array, sample_rate = sf.read(audio_buffer)
+        return caption, (sample_rate, audio_array)
 
-    return caption
+    return caption, None
 
 
 # Gradio interface
@@ -95,6 +65,7 @@ demo = gr.Interface(
     inputs=gr.Image(sources=["upload", "webcam"]),    
     outputs=[
         gr.Textbox(label="Caption"),
+        gr.Audio(label="Audio Output")
     ],
     live=True,
     title="SpokenVision",
@@ -102,4 +73,4 @@ demo = gr.Interface(
     allow_flagging="never"
 )
 
-demo.launch()
+demo.launch()