Update app.py

app.py

@@ -1,19 +1,35 @@
 import os
 import gradio as gr
-from TTS.api import TTS
 import torch
-import spaces 
+import spaces # ZeroGPU Magic
 
-# License accept
+# 1. Terms Accept
 os.environ["COQUI_TOS_AGREED"] = "1"
 
-# Model load (Wait for 2-3 mins first time)
+# 2. Model Installation Check
+print("⏳ Importing TTS...")
+try:
+    from TTS.api import TTS
+except ImportError:
+    print("❌ TTS Install Failed. Check requirements.txt")
+    raise
+
+# 3. Load Model (GPU Support)
 device = "cuda" if torch.cuda.is_available() else "cpu"
+print(f"⚙️ Device Selected: {device}")
+
+# Model ko global variable mein rakhenge
 tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
 
-@spaces.GPU(duration=60) 
+# 4. Voice Cloning Function (With ZeroGPU Decorator)
+@spaces.GPU(duration=120) 
 def clone_voice(text, language, speaker_audio):
+    if not text or not speaker_audio:
+        return None
+    
     output_path = "output.wav"
+    
+    # Generate Audio
     tts.tts_to_file(
         text=text,
         file_path=output_path,
@@ -22,13 +38,18 @@ def clone_voice(text, language, speaker_audio):
     )
     return output_path
 
+# 5. UI Setup
 iface = gr.Interface(
     fn=clone_voice,
     inputs=[
-        gr.Textbox(label="Text", value="Namaste, kaise hain aap?"),
+        gr.Textbox(label="Text to Speak", value="Namaste, ye meri cloned aawaz hai.", lines=2),
         gr.Dropdown(label="Language", choices=["hi", "en"], value="hi"),
-        gr.Audio(label="Reference Audio", type="filepath")
+        gr.Audio(label="Upload Speaker Voice (WAV/MP3)", type="filepath")
     ],
-    outputs=gr.Audio(label="Cloned Voice")
+    outputs=gr.Audio(label="Generated Audio"),
+    title="🚀 Shubham's Super Fast XTTS (H200 GPU)",
+    description="ZeroGPU powered Hindi Voice Cloning."
 )
-iface.launch()
+
+if __name__ == "__main__":
+    iface.launch()