Update app.py

app.py

@@ -105,16 +105,33 @@ def render_file_uploader():
 def save_uploaded_file(uploaded_file):
     """Save uploaded file to a temporary file and return the path."""
     try:
-        with tempfile.NamedTemporaryFile(delete=False, suffix=os.path.splitext(uploaded_file.name)[1]) as tmp_file:
-            tmp_file.write(uploaded_file.getvalue())
+        #Validate file type
+        allowed_types = ["wav", "mp3", "ogg"]
+        file_ext = os.path.splitext(uploaded_file.name)[1].lower()
+        if file_ext not in allowed_types:
+            raise ValueError(f"Unsupported file type: {file_ext}. Allowed: {', '.join(allowed_types)}")
+        
+        #Create temp file
+        with tempfile.NamedTemporaryFile(delete=False, suffix=f".{file_ext}") as tmp_file:
+            #Read files in chunks to handle large files
+            for chunk in uploaded_file.chunks(4096):
+                tmp_file.write(chunk)
             return tmp_file.name
     except Exception as e:
         st.error(f"Error saving file: {str(e)}")
-        return None
+        if 'tmp_file' in locals() and os.path.exists(tmp_file.name):
+            try:
+                os.unlink(tmp_file.name)
+            except:
+                pass
+        raise
 
 def transcribe_audio(file_path, model_name):
     """Transcribe audio using the specified model."""
     try:
+        #debug info
+        st.sidebar.info("ℹ️ Starting transcription...")
+
         # Debug: Show authentication status
         hf_token = os.getenv('HUGGINGFACE_TOKEN') or os.getenv('HF_TOKEN')
         st.sidebar.info(f"🔑 Using model: {model_name}")
@@ -124,29 +141,29 @@ def transcribe_audio(file_path, model_name):
         
         # Read the audio file
         try:
-            audio_data, sample_rate = librosa.load(file_path, sr=16000)
-            st.sidebar.info(f"🔊 Loaded audio: {len(audio_data)} samples @ {sample_rate}Hz")
+            audio_data, sample_rate = librosa.load(file_path, sr=16000, mono=True)
+            st.sidebar.info(f"🔊 Loaded audio: {len(audio_data)/sample_rate:.2f} seconds @ {sample_rate}Hz")
+
         except Exception as e:
-            st.error(f"❌ Error loading audio file: {str(e)}")
-            return None
-            
+            raise Exception(f"❌ Error loading audio file: {str(e)}")
+                    
         # Transcribe
         with st.spinner("🔄 Transcribing audio..."):
             result = transcriber.transcribe_audio(audio_data, sample_rate)
             
         if not result or 'text' not in result:
-            st.error("❌ No transcription results returned. The model might not be accessible.")
-            return None
-            
+            raise ValueError("❌ No transcription results returned. The model might not be accessible.")
+                    
         return result
         
     except Exception as e:
-        st.error(f"❌ Transcription failed: {str(e)}")
-        st.error("This might be due to:")
-        st.error("1. Invalid or missing Hugging Face token")
-        st.error("2. Insufficient permissions for the model")
-        st.error("3. Network connectivity issues")
-        st.error("4. Model not found or not accessible")
+        st.sidebar.error("❌ Transcription failed: {str(e)}")
+        st.sidebar.error("This might be due to:")
+        st.sidebar.error("1. Invalid or missing Hugging Face token")
+        st.sidebar.error("2. Insufficient permissions for the model")
+        st.sidebar.error("3. Network connectivity issues")
+        st.sidebar.error("4. Model not found or not accessible")
+        raise
         
         # Add debug info
         st.sidebar.error("🔍 Debug Info:")