Update app.py

app.py

@@ -171,6 +171,37 @@ def generate_speech(text, voice1, voice2, temperature, top_p, repetition_penalty
         return None
     
     try:
+        genai.configure(api_key=api_key)
+        model = genai.GenerativeModel('gemini-2.5-pro-preview-03-25')
+        
+        combined_content = prompt or ""
+        
+        if uploaded_file is not None:
+            file_bytes = io.BytesIO(uploaded_file)
+            
+            # Try to detect the file type based on content
+            file_bytes.seek(0)
+            if file_bytes.read(4) == b'%PDF':
+                # It's a PDF file
+                file_bytes.seek(0)
+                pdf_reader = PyPDF2.PdfReader(file_bytes)
+                file_content = "\n".join([page.extract_text() for page in pdf_reader.pages])
+            else:
+                # Try as text file first
+                file_bytes.seek(0)
+                try:
+                    file_content = file_bytes.read().decode('utf-8')
+                except UnicodeDecodeError:
+                    # If it's not a text file, try as a docx
+                    file_bytes.seek(0)
+                    try:
+                        doc = Document(file_bytes)
+                        file_content = "\n".join([para.text for para in doc.paragraphs])
+                    except:
+                        raise ValueError("Unsupported file type or corrupted file")
+
+            combined_content += "\n" + file_content if combined_content else file_content
+            
         progress(0.1, "Processing text...")
         paragraphs = text.split('\n\n')  # Split by double newline
         audio_samples = []
@@ -234,8 +265,7 @@ with gr.Blocks(title="Orpheus Text-to-Speech") as demo:
                 interactive=True,
                 container=True
             )
-            uploaded_file = gr.File(label="Upload File", type="binary")
-
+        uploaded_file = gr.File(label="Upload File", type="binary")
         with gr.Column(scale=2):
             duration = gr.Slider(minimum=1, maximum=60, value=5, step=1, label="Duration (minutes)")
             num_hosts = gr.Radio(["1", "2"], label="Number of Hosts", value="1")