Update app.py

app.py

@@ -4,15 +4,16 @@ import soundfile as sf
 import tempfile
 import os
 
-# URL of your dedicated processing server.
-# Adjust the port or endpoint path if needed.
+# URL of your dedicated processing server (adjust port/path if needed)
 SERVER_URL = "http://204.12.245.139:5000/process_audio"
 
 def process_audio(audio):
     """
-    Receives audio input from the browser (as a tuple: (sample_rate, numpy_array)),
-    writes it to a temporary WAV file, then sends the file to your dedicated server.
-    Expects a JSON response with a key 'transcription' or 'response' that is returned.
+    Process the input audio:
+      - Expects 'audio' as a tuple: (sample_rate, numpy_array)
+      - Writes the audio data to a temporary WAV file
+      - Sends the file via POST to your server endpoint
+      - Returns the transcription (or server response) as text
     """
     if audio is None:
         return "No audio provided. Please record something."
@@ -26,11 +27,10 @@ def process_audio(audio):
     try:
         with open(wav_path, "rb") as f:
             files = {"file": f}
-            # Timeout can be adjusted as needed.
             response = requests.post(SERVER_URL, files=files, timeout=30)
         if response.status_code == 200:
             json_data = response.json()
-            # Try to get a 'transcription' key first, then 'response'
+            # Expect server to return a key 'transcription' or 'response'
             result = json_data.get("transcription") or json_data.get("response")
             if not result:
                 result = "Server processed the audio, but did not return a result."
@@ -40,21 +40,22 @@ def process_audio(audio):
         result = f"Exception during processing: {e}"
     finally:
         os.remove(wav_path)
+    
     return result
 
-# Create a Gradio interface that captures audio via the browser microphone.
+# Create a Gradio interface.
+# Note: We removed "source" from gr.Audio to ensure compatibility with your Gradio version.
 iface = gr.Interface(
     fn=process_audio,
-    inputs=gr.Audio(source="microphone", type="numpy", label="Record Your Voice"),
+    inputs=gr.Audio(type="numpy", label="Record Your Voice"),
     outputs=gr.Textbox(label="Server Response"),
     title="Live AI Call Agent – Browser Mic Frontend",
     description=(
-        "Record audio using your browser microphone. The audio will be sent to our dedicated "
-        "server for processing with GPU acceleration. Your server should return a transcription or "
-        "an AI-generated response."
+        "Record audio using your browser microphone. The audio will be sent to our dedicated server "
+        "for processing with GPU acceleration. Your server should return a transcription or generated response."
     )
 )
 
 if __name__ == "__main__":
-    # Launch the app so that it listens on all interfaces. Hugging Face Spaces uses these settings.
+    # Launch the Gradio app to listen on all interfaces. Hugging Face Spaces will assign a public URL.
     iface.launch(server_name="0.0.0.0", server_port=7860)