Update app.py

app.py

@@ -10,17 +10,9 @@ device = 0 if torch.cuda.is_available() else -1
 asr_pipeline = pipeline(model="openai/whisper-small", device=device)
 
 # Function to handle the transcription process
-def transcribe_audio(audio_file):
-    # Create a temporary file to save the uploaded audio
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio_file:
-        temp_audio_file.write(audio_file.read())
-        temp_file_path = temp_audio_file.name
-
-    # Perform the transcription
-    transcription = asr_pipeline(temp_file_path)
-    
-    # Remove the temporary file
-    os.remove(temp_file_path)
+def transcribe_audio(audio_file_path):
+    # Perform the transcription using the audio file path
+    transcription = asr_pipeline(audio_file_path)
 
     # Return the transcription result
     return transcription['text']
@@ -28,7 +20,7 @@ def transcribe_audio(audio_file):
 # Create Gradio interface
 interface = gr.Interface(
     fn=transcribe_audio,  # The function to call when audio is uploaded
-    inputs=gr.Audio(type="file"),  # Removed the 'source' argument
+    inputs=gr.Audio(type="filepath"),  # Use 'filepath' to get the path to the audio file
     outputs="text",  # Output type: text (transcription)
     title="Whisper Audio Transcription",  # Title of the Gradio interface
     description="Upload an audio file to get a transcription using OpenAI's Whisper model"