import gradio as gr
from TTS.api import TTS

# Init TTS
tts = TTS(model_name="tts_models/multilingual/multi-dataset/your_tts", progress_bar = False, gpu=False)
zh_tts = TTS(model_name="tts_models/zh-CN/baker/tacotron2-DDC-GST", progress_bar=False, gpu=False)
# en_tts = TTS(model_name = "tts_models/en/ljspeech/vits", gpu=False)
# fr_tts = TTS(model_name = "tts_models/fr/css10/vits", gpu=False)
# de_tts = TTS(model_name = "tts_models/de/thorsten/vits", gpu=False)

def text_to_speech(text: str, speaker_wav, language: str):
    file_path = "output.wav"
    if language == "zh-CN":
        # if speaker_wav is not None:
        #     zh_tts.tts_to_file(text, speaker_wav=speaker_wav, file_path=file_path)
        # else:
        zh_tts.tts_to_file(text, file_path=file_path)
    # elif language == "de":
    #     if speaker_wav is not None:
    #         de_tts.tts_to_file(text, speaker_wav=speaker_wav, file_path=file_path)
    #     else:
    #         de_tts.tts_to_file(text, file_path=file_path)
    # elif language == "en":
    #     if speaker_wav is not None:
    #         en_tts.tts_to_file(text, speaker_wav=speaker_wav, file_path=file_path)
    #     else:
    #         en_tts.tts_to_file(text, file_path=file_path)
    else:
        if speaker_wav is not None:
            tts.tts_to_file(text, speaker_wav=speaker_wav, language=language, file_path=file_path)
        else:
            tts.tts_to_file(text, speaker=tts.speakers[0], language=language, file_path=file_path)
    return file_path

inputs = [gr.Textbox(label="Input the text", value="", max_lines=3),
          gr.Audio(label="Voice to clone", source="microphone", type="filepath"), 
            gr.Radio(label="Language", choices=["en", "zh-CN", "fr-fr"], value="en")]
outputs = gr.Audio(label="Output")

demo = gr.Interface(fn=text_to_speech, inputs=inputs, outputs=outputs)

demo.launch()