import tempfile
import gradio as gr
# import numpy as np
from TTS.utils.manage import ModelManager
from TTS.utils.synthesizer import Synthesizer


title = "Mandarin Text-to-Speech (TTS)"
description = "Generate mandarin speech from text using a Tachotron2 model with Coqui TTS 🐸, " \
              "a deep learning toolkit for Text-to-Speech."
article = "<p style='text-align: center'><a href='https://news.machinelearning.sg/posts/mandarin_text_to_speech_with_coqui_tts/' target='blank' class='footer'>Blog</a> | <a href='https://github.com/eugenesiow/practical-ml' target='_blank' class='footer'>Github Repo</a></p>"
examples = [
    ["语音合成是通过机械的、电子的方法产生人造语音的技术。"],
    ["李显龙总理表示，我国要达到像意大利的开放程度，几乎回到冠病疫情前的生活，还需要一段时间。"]
]


manager = ModelManager()
model_path, config_path, model_item = manager.download_model("tts_models/zh-CN/baker/tacotron2-DDC-GST")
synthesizer = Synthesizer(
    model_path, config_path, None, None, None,
)


def inference(text: str):
    wavs = synthesizer.tts(text)
    # output = (synthesizer.output_sample_rate, np.array(wavs).astype(np.float32))
    # return output
    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
        synthesizer.save_wav(wavs, fp)
        return fp.name


gr.Interface(
    fn=inference,
    inputs=[
        gr.inputs.Textbox(
            label="Input",
            default="你好吗？我很好。",
        ),
    ],
    outputs=gr.outputs.Audio(label="Output"),
    title=title,
    description=description,
    article=article,
    examples=examples,
    enable_queue=True,
    allow_flagging=False,
    ).launch(debug=False)