import gradio as gr
import edge_tts
import asyncio
import os

# Vozes disponíveis
VOICES = {
    "Antonio": "pt-BR-AntonioNeural",
    "Andrew (Multilingual)": "en-US-AndrewMultilingualNeural",
    "Brian (Multilingual)": "en-US-BrianMultilingualNeural",
    "Francisca": "pt-BR-FranciscaNeural",
    "Thalita": "pt-BR-ThalitaNeural",
    "Ava (Multilingual)": "en-US-AvaMultilingualNeural",
    "Emma (Multilingual)": "en-US-EmmaMultilingualNeural"
}

async def text_to_speech(text, voice):
    voice_key = VOICES[voice]
    output_file = f"output_{hash(text)}.mp3"
    communicate = edge_tts.Communicate(text, voice_key)
    await communicate.save(output_file)
    return output_file

async def process_tts(text, voice):
    return await text_to_speech(text, voice)

css = """
.gradio-container {
    background: white !important;
}

.label {
    color: black !important;
}

.gr-box {
    background: white !important;
    color: black !important;
}

.gr-input {
    color: black !important;
}

.gr-button {
    background: #2196F3 !important;
    color: white !important;
}

.gr-form {
    background: white !important;
    color: black !important;
}

.footer {
    text-align: center;
    margin-top: 50px;
    color: #666 !important;
    font-size: 0.9em;
}
"""

with gr.Blocks(theme=gr.themes.Light(), css=css) as iface:
    with gr.Column():
        gr.Markdown("# TSM - Texto em Voz")
        gr.Markdown("Converta texto em fala usando vozes em português e multilíngues.")
        
        with gr.Row():
            with gr.Column(scale=2):
                text_input = gr.Textbox(
                    label="Digite o texto para converter em fala",
                    placeholder="Escreva seu texto aqui...",
                    lines=5
                )
            with gr.Column(scale=1):
                voice = gr.Radio(
                    choices=list(VOICES.keys()),
                    label="Escolha a voz",
                    value="Francisca"
                )
        
        convert_button = gr.Button("Converter para Áudio")
        audio_output = gr.Audio(label="Áudio Gerado", interactive=True)
        
        gr.Markdown(
            """
            <div class="footer">
                <p>Desenvolvido por [TSM LTDA] © 2022-2024</p>
                <p>Powered by Azure Text-to-Speech</p>
            </div>
            """
        )
    
    convert_button.click(process_tts, inputs=[text_input, voice], outputs=[audio_output])

iface.launch(show_api=False)