import os
import scipy
import nltk
import tempfile
import numpy as np
from bark.generation import preload_models, SAMPLE_RATE
from bark import generate_audio
from scipy.io import wavfile
from IPython.display import Audio
import gradio as gr
nltk.download('punkt')
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
preload_models()



def generate_audio_from_text(text,language_prompt,gender_prompt):
    if language_prompt == "english":
        if gender_prompt == "male":
            history_prompt = "v2/en_speaker_8"
        else:
            history_prompt = "v2/en_speaker_9"
    elif language_prompt == "french":
        if gender_prompt == "male":
            history_prompt = "v2/fr_speaker_0"
        else:
            history_prompt = "v2/fr_speaker_1"
    elif language_prompt =="german":
        if gender_prompt=="male":
            history_prompt = "v2/de_speaker_2"
        else:
            history_prompt="v2/de_speaker_3"
    elif language_prompt =="hindi":
        if gender_prompt=="male":
            history_prompt = "v2/hi_speaker_8"
        else:
            history_prompt="v2/hi_speaker_3"
    elif language_prompt =="chinese":
        if gender_prompt=="male":
            history_prompt = "v2/zh_speaker_1"
        else:
            history_prompt="v2/zh_speaker_4"
    elif language_prompt =="italian":
        if gender_prompt=="male":
            history_prompt = "v2/it_speaker_4"
        else:
            history_prompt="v2/it_speaker_7"
    elif language_prompt =="japanese":
        if gender_prompt=="male":
            history_prompt = "v2/ja_speaker_2"
        else:
            history_prompt="v2/ja_speaker_0"
    else:
        raise ValueError("Invalid language or gender selection")

    sentences = nltk.sent_tokenize(text)
    # silence = np.zeros(int(0.25 * SAMPLE_RATE))  # quarter second of silence

    pieces = []
    for sentence in sentences:
        audio_array = generate_audio(sentence, history_prompt=history_prompt)
        pieces += [audio_array]

    # Concatenate the audio pieces
    final_audio = np.concatenate(pieces)

    # Save the audio to a WAV file
    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_wav:
        wavfile.write(temp_wav.name, SAMPLE_RATE, final_audio)

    # Return the saved audio file
    return temp_wav.name

# Define lists of language and gender options
language_options = [
    "english",
    "french",
    "german",
    "hindi",
    "chinese",
    "italian",
    "japanese",


]

gender_options = [
    "male",
    "female",
]

# Create a Gradio interface with text input and dropdown menus for language and gender
iface = gr.Interface(
    fn=generate_audio_from_text,
    inputs=[
        gr.Textbox(text="Enter text to convert to speech:"),
        gr.Dropdown(choices=language_options, label="Select language:"),
        gr.Dropdown(choices=gender_options, label="Select gender:"),
    ],
    outputs=gr.outputs.File(label="Download WAV File"),
    title="Text-to-Speech App Vertical Solution",
    timeout=300,
)

# Launch the Gradio app with sharing enabled
iface.launch(share=True, debug=True, enable_queue=True)