import sys
import os
# By using XTTS you agree to CPML license https://coqui.ai/cpml
os.environ["COQUI_TOS_AGREED"] = "1"

import gradio as gr
from TTS.api import TTS

model_names = TTS().list_models()
m = model_names[1]
tts = TTS(m, gpu=False)
tts.to("cpu")
#tts.to("cuda") # cuda only

def predict(prompt, language, audio_file_pth, mic_file_path, use_mic, agree):
    if agree == True:
        if use_mic == True:
            if mic_file_path is not None:
                speaker_wav=mic_file_path
            else:
                gr.Warning("Please record your voice with Microphone, or uncheck Use Microphone to use reference audios")
                return (
                    None,
                    None,
                ) 
                
        else:
            speaker_wav=audio_file_pth

        if len(prompt)<2:
            gr.Warning("Please give a longer prompt text")
            return (
                    None,
                    None,
                )
        if len(prompt)>10000:
            gr.Warning("Text length limited to 10000 characters for this demo, please try shorter text")
            return (
                    None,
                    None,
                )  
        try:
            if language == "fr":
                if m.find("your") != -1:
                    language = "fr-fr"
            if m.find("/fr/") != -1:
                language = None
            tts.tts_to_file(
                text=prompt,
                file_path="output.wav",
                speaker_wav=speaker_wav,
                language=language
            )
        except RuntimeError as e :
            if "device-assert" in str(e):
                # cannot do anything on cuda device side error, need tor estart
                gr.Warning("Unhandled Exception encounter, please retry in a minute")
                print("Cuda device-assert Runtime encountered need restart")
                sys.exit("Exit due to cuda device-assert")
            else:
                raise e
            
        return (
            gr.make_waveform(
                audio="output.wav",
            ),
            "output.wav",
        )
    else:
        gr.Warning("Please accept the Terms & Condition!")
        return (
                None,
                None,
            ) 

title = "Voice Clone"

gr.Interface(
    fn=predict,
    inputs=[
        gr.Textbox(
            label="Text Prompt",
            info="One or two sentences at a time is better",
            value="Hello, World !, here is an example of light voice cloning. Try to upload your best audio samples quality",
        ),
        gr.Dropdown(
            label="Language",
            info="Select an output language for the synthesised speech",
            choices=[
                "en",
                "es",
                "fr",
                "de",
                "it",
                "pt",
                "pl",
                "tr",
                "ru",
                "nl",
                "cs",
                "ar",
                "zh-cn",
            ],
            max_choices=1,
            value="en",
        ),
        gr.Audio(
            label="Reference Audio",
            info="Click on the ✎ button to upload your own target speaker audio",
            type="filepath",
            value="examples/female.wav",
        ),
        gr.Audio(source="microphone",
                 type="filepath",
                 info="Use your microphone to record audio",
                 label="Use Microphone for Reference"),
        gr.Checkbox(label="Check to use Microphone as Reference",
                    value=False,
                    info="Notice: Microphone input may not work properly under traffic",),
        gr.Checkbox(
            label="Agree",
            value=True,
            info="I agree to the terms of the Coqui Public Model License at https://coqui.ai/cpml",
        ),
    ],
    outputs=[
        gr.Video(label="Waveform Visual"),
        gr.Audio(label="Synthesised Audio"),
    ],
    title=title,
).queue().launch(debug=True)