app.py

import gradio as gr
import librosa
import soundfile as sf
import numpy as np

def convert_voice(source, target):
    src_audio, sr = librosa.load(source, sr=None)
    tgt_audio, _ = librosa.load(target, sr=None)

    # simple placeholder mixing (replace with RVC inference)
    min_len = min(len(src_audio), len(tgt_audio))
    output = (src_audio[:min_len] * 0.5) + (tgt_audio[:min_len] * 0.5)

    out_path = "output.wav"
    sf.write(out_path, output, sr)

    return out_path


with gr.Blocks() as demo:
    gr.Markdown("# RVC Voice Converter")

    source = gr.Audio(type="filepath", label="Source Voice")
    target = gr.Audio(type="filepath", label="Target Voice")

    btn = gr.Button("Convert")

    output = gr.Audio(label="Converted Voice")

    btn.click(convert_voice, inputs=[source, target], outputs=output)

demo.launch()