app.py

import gradio as gr
import edge_tts
import asyncio
import os
import datetime
from pathlib import Path
import hashlib

# Crea una directory per i file audio
AUDIO_DIR = Path("audio_files")
AUDIO_DIR.mkdir(exist_ok=True)

async def get_voices():
    voices = await edge_tts.list_voices()
    return {f"{v['ShortName']} - {v['Locale']} ({v['Gender']})": v['ShortName'] for v in voices}

async def text_to_speech(text, voice, rate, pitch):
    if not text.strip():
        return None, "Please enter text to convert."
    if not voice:
        return None, "Please select a voice."
    
    voice_short_name = voice.split(" - ")[0]
    rate_str = f"{rate:+d}%"
    pitch_str = f"{pitch:+d}Hz"
    communicate = edge_tts.Communicate(text, voice_short_name, rate=rate_str, pitch=pitch_str)
    
    # Crea un nome file univoco basato sul contenuto
    content_hash = hashlib.md5(f"{text}{voice}{rate}{pitch}".encode()).hexdigest()
    output_path = AUDIO_DIR / f"{content_hash}.mp3"
    
    # Genera l'audio solo se non esiste già
    if not output_path.exists():
        await communicate.save(str(output_path))
    
    return str(output_path), None

# Funzione di pulizia per rimuovere i file più vecchi
def cleanup_old_files(directory: Path, max_files: int = 100, max_age_hours: int = 24):
    try:
        files = list(directory.glob("*.mp3"))
        
        # Rimuovi i file più vecchi di max_age_hours
        current_time = datetime.datetime.now()
        for file in files:
            try:
                file_age = current_time - datetime.datetime.fromtimestamp(file.stat().st_mtime)
                if file_age.total_seconds() > (max_age_hours * 3600):
                    try:
                        file.unlink()
                    except (PermissionError, OSError):
                        continue
            except (OSError, ValueError):
                continue
        
        # Se ci sono ancora troppi file, rimuovi i più vecchi
        files = list(directory.glob("*.mp3"))
        if len(files) > max_files:
            files.sort(key=lambda x: x.stat().st_mtime)
            for file in files[:-max_files]:
                try:
                    file.unlink()
                except (PermissionError, OSError):
                    continue
    except Exception as e:
        print(f"Error during cleanup: {e}")

async def tts_interface(text, voice, rate, pitch):
    # Esegui la pulizia prima di generare un nuovo file
    cleanup_old_files(AUDIO_DIR)
    
    audio, warning = await text_to_speech(text, voice, rate, pitch)
    if warning:
        return audio, gr.Warning(warning)
    return audio, None

async def create_demo():
    voices = await get_voices()
    
    description = """
    Convert text to speech using Microsoft Edge TTS. Adjust speech rate and pitch: 0 is default, positive values increase, negative values decrease.
    Original Space by innoai
    """
    
    demo = gr.Interface(
        fn=tts_interface,
        inputs=[
            gr.Textbox(label="Input Text", lines=5),
            gr.Dropdown(choices=[""] + list(voices.keys()), label="Select Voice", value=""),
            gr.Slider(minimum=-50, maximum=50, value=0, label="Speech Rate Adjustment (%)", step=1),
            gr.Slider(minimum=-20, maximum=20, value=0, label="Pitch Adjustment (Hz)", step=1)
        ],
        outputs=[
            gr.Audio(label="Generated Audio", type="filepath"),
            gr.Markdown(label="Warning", visible=False)
        ],
        title="Edge TTS Text-to-Speech",
        description=description,
        article="Experience the power of Edge TTS for text-to-speech conversion!",
        analytics_enabled=False,
        allow_flagging="manual",
        api_name="predict"
    )
    return demo

async def main():
    demo = await create_demo()
    demo.queue(default_concurrency_limit=25)
    demo.launch(show_api=True)

if __name__ == "__main__":
    asyncio.run(main())