from subprocess import call
import gradio as gr
import os
from TTS.api import TTS
from TTS.utils.synthesizer import Synthesizer

# List available 🐸TTS models and choose the first one
# all_models = TTS.list_models()
# for model in all_models:
#     print(model)

# print("Using model: ", all_models[0])
# model = all_models[0]
# Init TTS

MAX_TXT_LEN = 500


print("Downloading model...", '')

voiceCloneModel = TTS('tts_models/multilingual/multi-dataset/your_tts')

# tts = TTS("tts_models/de/thorsten/tacotron2-DDC")


def run_cmd(command):
    try:
        print(command)
        call(command)
    except KeyboardInterrupt:
        print("Process interrupted")
        sys.exit(1)


def inference(text, speaker):
    if len(text) > MAX_TXT_LEN:
        text = text[:MAX_TXT_LEN]
        print(
            f"Input text was cutoff since it went over the {MAX_TXT_LEN} character limit.")
    if (speaker == 'Speaker-1'):
        speaker = 'input/amitabh.mp3'
    elif (speaker == 'Speaker-2'):
        speaker = 'input/amrish.mp3'
    elif (speaker == 'Speaker-3'):
        speaker = 'input/obama.mp3'
    elif (speaker == 'Speaker-4'):
        speaker = 'input/trump.wav'
    elif (speaker == 'Rock'):
        speaker = 'input/sample/Dwayne-Johnson-sample.wav'
    elif (speaker == 'Elon'):
        speaker = 'input/sample/Elon-Sample.wav'
    elif (speaker == 'Obama'):
        speaker = 'input/sample/Obama-sample.wav'
    elif (speaker == 'Tony'):
        speaker = 'input/sample/tony-stark-2.wav'
    elif (speaker == 'Madara Uchiha'):
        speaker = 'input/sample/Madara.mp3'
    else:
        speaker = 'input/z-default.wav'
    # print("speaker: ", speaker)
    # cmd = ['tts', '--text', text, '--out_path', 'output/tts_output.wav']
    # run_cmd(cmd)
    # Text to speech to a file
    # tts = TTS(model="tts_models/multilingual/multi-dataset/your_tts",
    #           progress_bar=False, gpu=True)
    voiceCloneModel.tts_to_file(text, speaker_wav=speaker,
                                language="en", file_path="output/output.wav")

    # for i in range(len(model.languages)):
    #     model.tts_to_file(text=text,
    #                       speaker=model.speakers[i], language=model.languages[0], file_path='output/output-'+str(i)+'.wav')

    return 'output/output.wav'


sampleInput = ['Speaker-1', 'Speaker-2', 'Speaker-3',
               'Speaker-4', 'Rock', 'Elon', 'Obama', 'Tony', 'Madara Uchiha']
inputs = [gr.Textbox(lines=5, label="Input Text"),
          gr.Dropdown(
              sampleInput, label="Sample Input", default="Speaker-1",)
          ]
outputs = gr.Audio(type="filepath", label="Output Audio")
title = "Text To Speech"
description = "An example of using TTS to generate speech from text."
article = ""
examples = [
    ["This is an open-source library that generates synthetic speech"]
]
gr.Interface(
    inference,
    inputs,
    outputs,
    verbose=True,
    title=title,
    description=description,
    article=article,
    examples=examples,
    enable_queue=True,
    allow_flagging="never",

).launch(debug=True)