app.py

import gradio as gr
import os
import shlex

os.system('wget https://www.dropbox.com/s/luro5o8kjotkn70/pretrained.pt -P synthesizer/saved_models/pretrained')
os.system('wget https://www.dropbox.com/s/dv0ymnlqillecfw/pretrained.pt -P encoder/saved_models')
os.system('wget https://www.dropbox.com/s/aiym2qfv7087bsc/pretrained.pt -P vocoder/saved_models/pretrained')

def inference(audio, text):
    os.system("python demo_cli.py --audio_path "+audio.name+" --text "+shlex.quote(text.strip()))
    return 'demo_output_1.wav'

iface = gr.Interface(inference, inputs=[gr.inputs.Audio(type="file"),"text"], outputs=gr.outputs.Audio(type="file"))

iface.launch()