build.py

import os, stat
from zipfile import ZipFile

# Use never ffmpeg binary for Ubuntu20 to use denoising for microphone input
print("Export newer ffmpeg binary for denoise filter")
ZipFile("ffmpeg.zip").extractall()
print("Make ffmpeg binary executable")
st = os.stat("ffmpeg")
os.chmod("ffmpeg", st.st_mode | stat.S_IEXEC)

# This will trigger downloading model
print("Downloading if not downloaded Coqui XTTS V2")
from TTS.utils.manage import ModelManager

model_name = "tts_models/multilingual/multi-dataset/xtts_v2"
ModelManager().download_model(model_name)
print("XTTS downloaded")