import gradio as gr
import torch
import soundfile as sf
import spaces
import os
import numpy as np
import re
from transformers import SpeechT5Processor, SpeechT5ForTextToSpeech, SpeechT5HifiGan, AutoTokenizer
from speechbrain.pretrained import EncoderClassifier
from datasets import load_dataset
from huggingface_hub import hf_hub_download
import uuid
import wave
import io
import tempfile
import shutil



import importlib.util
from huggingface_hub import hf_hub_download


device = "cuda" if torch.cuda.is_available() else "cpu"


auth_token = os.environ.get("hf_token") or True


tts_path = hf_hub_download(repo_id="fahadqazi/private-code", filename="sindhi_tts_module.py", use_auth_token=auth_token)

spec = importlib.util.spec_from_file_location("sindhi_tts_module", tts_path)
tts_module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(tts_module)

SindhiTTSClass = tts_module.SindhiTTS

SindhiTTS = SindhiTTSClass(
    auth_token=auth_token
)

    
iface = gr.Interface(
    fn=SindhiTTS.synthesize,
    inputs=[
        gr.Textbox(label="Enter Sindhi text to convert to speech", value="سنڌي ماڻھو ڏَکڻ ايشيا ۾ آباد ماڻھن جو هڪ نسلي گروھ آهي جيڪو دراوڙ، آريائي ۽ سامي نسلن سان تعلق رکي ٿو۔", rtl=True),
        gr.Radio(
            choices=[("Female", 1), ("Male", 0)],
            value=1,
            label="Choose Gender"
        )
    ],
    outputs=[
        gr.Audio(label="Generated Speech", type="numpy")
    ],
    title="Sindhi Text-to-Speech Demo",
    description="Enter Sindhi text, and listen to the generated speech. Use shorter messages for better results."
)

iface.launch(share=True)