Update app.py

app.py

@@ -5,6 +5,7 @@ import gradio as gr
 import asyncio
 import logging
 import subprocess
+import shutil
 from serpapi import GoogleSearch
 from pydantic import BaseModel
 from autogen_agentchat.agents import AssistantAgent
@@ -20,7 +21,7 @@ import traceback
 import soundfile as sf
 import tempfile
 from pydub import AudioSegment
-!huggingface-cli download coqui/XTTS-v2 --repo-type model --local-dir XTTS-v2
+
 # Set up logging
 logging.basicConfig(
     level=logging.DEBUG,
@@ -172,6 +173,41 @@ async def validate_and_convert_speaker_audio(speaker_audio):
         logger.error("Failed to validate or convert speaker audio %s: %s", speaker_audio, str(e))
         return None
 
+# Helper function to download XTTS-v2 model
+def download_xtts_model():
+    if os.path.exists(XTTS_MODEL_DIR):
+        logger.info("XTTS-v2 directory already exists: %s", XTTS_MODEL_DIR)
+        return True
+    
+    if not shutil.which("huggingface-cli"):
+        logger.error("huggingface-cli not installed")
+        return False
+    
+    cmd = [
+        "huggingface-cli",
+        "download",
+        "coqui/XTTS-v2",
+        "--repo-type", "model",
+        "--local-dir", XTTS_MODEL_DIR
+    ]
+    logger.info("Downloading XTTS-v2 model: %s", " ".join(cmd))
+    
+    try:
+        result = subprocess.run(
+            cmd,
+            capture_output=True,
+            text=True,
+            check=True
+        )
+        logger.info("XTTS-v2 download succeeded: %s", result.stdout)
+        return True
+    except subprocess.CalledProcessError as e:
+        logger.error("Failed to download XTTS-v2: %s\n%s", e.stderr, e.stdout)
+        return False
+    except Exception as e:
+        logger.error("Unexpected error downloading XTTS-v2: %s", str(e))
+        return False
+
 # Helper function to generate audio using XTTS-v2 CLI
 def generate_xtts_audio(text, speaker_wav, output_path):
     try:
@@ -578,6 +614,16 @@ Example for 1 content slide:
             """
             return
         
+        # Download XTTS-v2 model if needed
+        if not download_xtts_model():
+            yield f"""
+            <div style="display: flex; flex-direction: column; justify-content: center; align-items: center; height: 100%; min-height: 700px; padding: 20px; text-align: center; border: 1px solid #ddd; border-radius: 8px;">
+                <h2 style="color: #d9534f;">Failed to download XTTS-v2 model</h2>
+                <p style="margin-top: 20px;">Please install huggingface_hub (`pip install huggingface_hub`) and ensure internet connectivity, then try again.</p>
+            </div>
+            """
+            return
+        
         # Verify XTTS-v2 model files
         required_files = [
             os.path.join(XTTS_MODEL_DIR, "model_se.pth.tar"),
@@ -589,7 +635,7 @@ Example for 1 content slide:
                 yield f"""
                 <div style="display: flex; flex-direction: column; justify-content: center; align-items: center; height: 100%; min-height: 700px; padding: 20px; text-align: center; border: 1px solid #ddd; border-radius: 8px;">
                     <h2 style="color: #d9534f;">Missing XTTS-v2 model files</h2>
-                    <p style="margin-top: 20px;">Please ensure XTTS-v2 is downloaded to {XTTS_MODEL_DIR} and try again.</p>
+                    <p style="margin-top: 20px;">Failed to download XTTS-v2 to {XTTS_MODEL_DIR}. Please run `huggingface-cli download coqui/XTTS-v2 --repo-type model --local-dir XTTS-v2` manually and try again.</p>
                 </div>
                 """
                 return
@@ -776,6 +822,7 @@ with gr.Blocks(title="Agent Feynman") as demo:
                 serpapi_key = gr.Textbox(label="SerpApi Key", type="password", placeholder="Enter your SerpApi key")
                 num_slides = gr.Slider(1, 20, step=1, label="Number of Content Slides", value=3)
                 speaker_audio = gr.Audio(label="Speaker sample audio (MP3 or WAV)", type="filepath", elem_id="speaker-audio")
+                audio_preview = gr.Audio(label="Audio Preview", interactive=False)
                 generate_btn = gr.Button("Generate Lecture")
         with gr.Column(scale=2):
             default_slide_html = """
@@ -788,7 +835,8 @@ with gr.Blocks(title="Agent Feynman") as demo:
     
     speaker_audio.change(
         fn=update_audio_preview,
-        inputs=speaker_audio
+        inputs=speaker_audio,
+        outputs=audio_preview
     )
     
     generate_btn.click(