audio.py

import aiofiles
import hashlib
import os
from fastapi import APIRouter
from fastapi.responses import JSONResponse, FileResponse
from TTS.api import TTS
from pydantic import BaseModel

audio_router = APIRouter()

# Define the input model
class TTSInput(BaseModel):
    input_text: str = 'olà tia divina '
    emotion: str = "Happy"
    language: str
    speed: float = 1.5
    key: str


async def generate_audio_file(input_data: TTSInput):
    model_name = TTS.list_models()[0]
    
    # Initialize the TTS model
    tts = TTS(model_name=model_name)
    
    # Create a string with the input values for hashing
    hash_input = f"{input_data.input_text}{input_data.emotion}{input_data.language}{input_data.speed}"
    
    # Calculate the MD5 hash based on the input values
    md5_hash = hashlib.md5(hash_input.encode()).hexdigest()
    
    # Check if the audio file already exists
    audio_file_path = os.path.join("audio", f"{md5_hash}.wav")
    if not os.path.exists(audio_file_path):
        # Generate TTS audio and save to a file
        tts.tts_to_file(
            text=input_data.input_text,
            speaker=tts.speakers[5],
            language=tts.languages[2],
            file_path=audio_file_path,
            gpu=True,
            emotion=input_data.emotion,
            speed=input_data.speed,
            progress_bar=True
        )
    
    return audio_file_path, md5_hash

@audio_router.post("/generate_audio", response_class=JSONResponse)
async def generate_audio(input_data: TTSInput):
    audio_file, md5_hash = await generate_audio_file(input_data)
    return {"message": "Audio generated successfully", "audio_file": audio_file, "md5_hash": md5_hash}

@audio_router.get("/download_audio/{md5_hash}")
async def download_audio(md5_hash: str):
    audio_file = os.path.join("audio", f"{md5_hash}.wav")
    
    if os.path.exists(audio_file):
        return FileResponse(audio_file, media_type="audio/wav")
    else:
        return JSONResponse("message", "Audio not found")