handler.py

from typing import Dict, Any
import tempfile
import torchaudio
import soundfile as sf
import re
from num2words import num2words
from f5_tts.model import DiT
from f5_tts.infer.utils_infer import (
    load_vocoder,
    load_model,
    preprocess_ref_audio_text,
    infer_process,
    remove_silence_for_generated_wav,
)
import base64
import io
import numpy as np
from huggingface_hub import hf_hub_download
import traceback

class EndpointHandler:
    def __init__(self, path=""):
        self.vocoder = load_vocoder()
        model_cfg = dict(dim=1024, depth=22, heads=16, ff_mult=2, text_dim=512, conv_layers=4)
        
        model_path = hf_hub_download(
            repo_id="jpgallegoar/F5-Spanish",
            filename="model_1200000.safetensors"
        )
        
        self.ema_model = load_model(DiT, model_cfg, model_path)

    def traducir_numero_a_texto(self, texto):
        texto_separado = re.sub(r'([A-Za-z])(\d)', r'\1 \2', texto)
        texto_separado = re.sub(r'(\d)([A-Za-z])', r'\1 \2', texto_separado)
        def reemplazar_numero(match):
            numero = match.group()
            return num2words(int(numero), lang='es')
        return re.sub(r'\b\d+\b', reemplazar_numero, texto_separado)

    def __call__(self, data: Dict[str, Any]) -> Dict[str, Any]:
        try:
            ref_audio_base64 = data.get("ref_audio")
            if not ref_audio_base64:
                return {
                    "success": False,
                    "error": "Missing required field: 'ref_audio'"
                }

            # Decode base64 audio and write to temp file
            try:
                audio_bytes = base64.b64decode(ref_audio_base64)
                with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio_file:
                    temp_audio_file.write(audio_bytes)
                    temp_audio_path = temp_audio_file.name
            except Exception as e:
                return {
                    "success": False,
                    "error": f"Invalid audio data: {type(e).__name__}: {str(e)}"
                }

            ref_text = data.get("ref_text", "")
            gen_text = data.get("gen_text", "")
            if not gen_text:
                return {
                    "success": False,
                    "error": "Missing required field: 'gen_text'"
                }

            remove_silence = data.get("remove_silence", True)
            cross_fade_duration = data.get("cross_fade_duration", 0.15)
            speed = data.get("speed", 1.0)

            ref_audio, ref_text = preprocess_ref_audio_text(temp_audio_path, ref_text, show_info=print)

            if not gen_text.startswith(" "):
                gen_text = " " + gen_text
            if not gen_text.endswith(". "):
                gen_text += ". "
            gen_text = self.traducir_numero_a_texto(gen_text.lower())

            final_wave, final_sample_rate, _ = infer_process(
                ref_audio,
                ref_text,
                gen_text,
                self.ema_model,
                self.vocoder,
                cross_fade_duration=cross_fade_duration,
                speed=speed,
                show_info=print,
            )

            if remove_silence:
                with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
                    sf.write(f.name, final_wave, final_sample_rate)
                    remove_silence_for_generated_wav(f.name)
                    final_wave, _ = torchaudio.load(f.name)
                final_wave = final_wave.squeeze().cpu().numpy()

            with io.BytesIO() as buffer:
                sf.write(buffer, final_wave, final_sample_rate, format="WAV")
                buffer.seek(0)
                encoded_audio = base64.b64encode(buffer.read()).decode("utf-8")

            return {
                "success": True,
                "audio_base64": encoded_audio
                }

        except Exception as e:
            print("==== Exception Traceback ====")
            traceback.print_exc()
            print("==== End Traceback ====")
            return {
                "success": False,
                "error": f"{type(e).__name__}: {str(e)}"
                }