Create app.py

app.py

@@ -0,0 +1,51 @@
+import gradio as gr
+import torch
+import librosa
+import numpy as np
+
+# Assuming you have a model file for voice conversion
+from model import load_model, convert_voice
+
+# Load the pre-trained voice conversion model
+model = load_model("path_to_pretrained_model")  # Adjust this based on the actual RVC model
+
+def voice_conversion(source_audio, target_voice):
+    """
+    Function to perform voice conversion from source to target voice style
+    """
+    # Convert input audio to the desired format (this may vary depending on your model)
+    y, sr = librosa.load(source_audio)
+    input_audio = torch.tensor(y).unsqueeze(0)
+
+    # Use model for voice conversion
+    converted_audio = convert_voice(model, input_audio, target_voice)
+    
+    # Convert output tensor back to numpy for playback
+    converted_audio_np = converted_audio.detach().cpu().numpy()
+    
+    # Save to file or return as numpy array
+    output_file = "output_converted.wav"
+    librosa.output.write_wav(output_file, converted_audio_np, sr)
+    
+    return output_file
+
+# Define the Gradio interface
+def infer(source_audio, target_voice):
+    # Call the voice conversion function
+    result_audio = voice_conversion(source_audio, target_voice)
+    return result_audio
+
+# Gradio interface with inputs and outputs
+iface = gr.Interface(
+    fn=infer, 
+    inputs=[
+        gr.Audio(source="microphone", type="filepath", label="Source Audio"),
+        gr.Dropdown(["Voice1", "Voice2", "Voice3"], label="Target Voice")  # Dropdown for target voice options
+    ], 
+    outputs=gr.Audio(type="file", label="Converted Audio"),
+    title="Retrieval-based Voice Conversion",
+    description="Convert voice from a source audio to a target voice style."
+)
+
+if __name__ == "__main__":
+    iface.launch()