from transformers import VitsModel, AutoTokenizer
import torch

model = VitsModel.from_pretrained("SeyedAli/Arabic-Speech-synthesis")
tokenizer = AutoTokenizer.from_pretrained("SeyedAli/Arabic-Speech-synthesis")

text = "السلام عليكم ورحمة الله وبركاته"
inputs = tokenizer(text, return_tensors="pt")

with torch.no_grad():
    output = model(**inputs).waveform