diff --git a/demo.py b/demo.py
new file mode 100644
index 0000000..b716781
--- /dev/null
+++ b/demo.py
@@ -0,0 +1,143 @@
+import gradio as gr
+import numpy as np
+import soundfile as sf
+import tempfile
+import os
+from zipvoice.luxvoice import LuxTTS
+
+# Initialize LuxTTS model (only once at startup)
+print("Loading LuxTTS model...")
+lux_tts = LuxTTS('YatharthS/LuxTTS', device='cuda', threads=2)  # Use device='cpu' for CPU
+print("Model loading complete")
+
+def generate(audio, text, rms, num_steps, t_shift):
+    if audio is None:
+        return None, "Error: Please upload an audio file"
+    
+    if not text or text.strip() == "":
+        return None, "Error: Please enter text"
+    
+    try:
+        sample_rate, audio_data = audio
+        
+        print(f"Received text: {text}")
+        print(f"Sample rate: {sample_rate}")
+        print(f"Audio data shape: {audio_data.shape}")
+        print(f"Parameters - RMS: {rms}, Num steps: {num_steps}, T-shift: {t_shift}")
+        
+        # Save audio to temporary file (LuxTTS requires file path)
+        with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp_file:
+            tmp_path = tmp_file.name
+            sf.write(tmp_path, audio_data, sample_rate)
+        
+        # Encode audio (first run may take ~10 seconds)
+        print("Encoding audio...")
+        encoded_prompt = lux_tts.encode_prompt(tmp_path, rms=rms)
+        
+        # Generate speech
+        print("Generating speech...")
+        final_wav = lux_tts.generate_speech(text, encoded_prompt, num_steps=num_steps, t_shift=t_shift)
+        
+        # Convert to numpy array
+        final_wav = final_wav.numpy().squeeze()
+        
+        # Delete temporary file
+        os.unlink(tmp_path)
+        
+        print("Speech generation complete")
+        
+        # Return with 48000Hz sample rate
+        return (48000, final_wav), "✓ Speech generation complete"
+        
+    except Exception as e:
+        print(f"Error occurred: {str(e)}")
+        import traceback
+        traceback.print_exc()
+        return None, f"Error: {str(e)}"
+
+
+# Create Gradio interface
+with gr.Blocks() as demo:
+    gr.Markdown("# LuxTTS Voice Cloning")
+    gr.Markdown("Upload a reference audio and enter text to generate speech")
+    
+    with gr.Row():
+        with gr.Column():
+            audio_input = gr.Audio(
+                label="Reference Audio (WAV/MP3)",
+                type="numpy",
+                sources=["upload", "microphone"]
+            )
+            text_input = gr.Textbox(
+                label="Text to Generate",
+                placeholder="Enter the text you want to generate as speech",
+                lines=3
+            )
+            
+            # Parameter controls
+            gr.Markdown("### Generation Parameters")
+            rms_slider = gr.Slider(
+                minimum=0.001,
+                maximum=0.1,
+                value=0.01,
+                step=0.001,
+                label="RMS (Volume normalization)",
+                info="Root Mean Square for audio normalization (default: 0.01)"
+            )
+            num_steps_slider = gr.Slider(
+                minimum=1,
+                maximum=20,
+                value=4,
+                step=1,
+                label="Number of Steps",
+                info="Inference steps - higher values may improve quality but take longer (default: 4)"
+            )
+            t_shift_slider = gr.Slider(
+                minimum=0.0,
+                maximum=1.0,
+                value=0.9,
+                step=0.05,
+                label="T-Shift",
+                info="Temperature shift parameter (default: 0.9)"
+            )
+            
+            submit_btn = gr.Button("Generate Speech", variant="primary")
+        
+        with gr.Column():
+            audio_output = gr.Audio(
+                label="Generated Speech (WAV)",
+                type="numpy"
+            )
+            status_output = gr.Textbox(
+                label="Status",
+                interactive=False
+            )
+    
+    # Handle button click
+    submit_btn.click(
+        fn=generate,
+        inputs=[audio_input, text_input, rms_slider, num_steps_slider, t_shift_slider],
+        outputs=[audio_output, status_output]
+    )
+    
+    # Usage instructions
+    gr.Markdown("""
+    ## How to Use
+    1. **Reference Audio**: Upload a WAV or MP3 file of the voice you want to clone, or record from microphone
+    2. **Text Input**: Enter the text you want to generate as speech (English recommended)
+    3. **Adjust Parameters** (optional):
+       - **RMS**: Controls volume normalization (0.001-0.1, default: 0.01)
+       - **Number of Steps**: Inference quality/speed tradeoff (1-20, default: 4)
+       - **T-Shift**: Temperature parameter affecting generation (0.0-1.0, default: 0.9)
+    4. Click the **Generate Speech** button
+    5. Once processing is complete, the generated audio will appear on the right
+    
+    **Notes**: 
+    - First run may take ~10 seconds for audio encoding initialization
+    - Clear reference audio with minimal background noise is recommended
+    - Use `device='cuda'` for GPU or `device='cpu'` for CPU in the code
+    - Higher num_steps values may improve quality but increase processing time
+    """)
+
+if __name__ == "__main__":
+    demo.launch()