Files
BotVPS/audio_handler.py

64 lines
2.4 KiB
Python

import os
import speech_recognition as sr
from pydub import AudioSegment
from gtts import gTTS
import uuid
import re
from elevenlabs import generate, save, Voice, VoiceSettings
from elevenlabs.client import ElevenLabs
def transcribe_audio(file_path: str) -> str:
"""Converte áudio (qualquer formato compatível com pydub) para WAV e transcreve com Google Speech."""
recognizer = sr.Recognizer()
# Se não for wav, converte usando pydub (precisa de ffmpeg na VPS)
temp_wav = f"/tmp/{uuid.uuid4()}.wav"
try:
audio = AudioSegment.from_file(file_path)
audio.export(temp_wav, format="wav")
with sr.AudioFile(temp_wav) as source:
audio_data = recognizer.record(source)
text = recognizer.recognize_google(audio_data, language="pt-BR")
return text
finally:
if os.path.exists(temp_wav):
os.remove(temp_wav)
def text_to_speech(text: str) -> str:
"""Sintetiza texto em áudio MP3 usando ElevenLabs (voz masculina) ou gTTS como fallback."""
# Limpeza para narração
texto_limpo = text.replace("🤖", "").replace("🧑‍🏫", "").replace("*", "").replace("`", "")
texto_limpo = re.sub(r'<REFINED>.*?</REFINED>', '', texto_limpo, flags=re.DOTALL).strip()
if not texto_limpo:
texto_limpo = "Relatório processado. Os detalhes estão no painel visual."
filename = f"audio_reply_{uuid.uuid4().hex[:8]}.mp3"
filepath = os.path.join("/tmp", filename)
api_key = os.getenv("ELEVENLABS_API_KEY")
voice_id = os.getenv("ELEVENLABS_VOICE_ID", "ErMmoak87FvRAs60I6g0") # Marco (Male, PT-BR) as default
if api_key:
try:
client = ElevenLabs(api_key=api_key)
audio = client.generate(
text=texto_limpo,
voice=voice_id,
model="eleven_multilingual_v2"
)
# Salvando o resultado para arquivo
with open(filepath, "wb") as f:
for chunk in audio:
if chunk:
f.write(chunk)
return filename
except Exception as e:
print(f"[VOICE] Erro ElevenLabs: {e}. Usando fallback gTTS.")
# Fallback gTTS (Voz feminina lenta mas gratuita)
tts = gTTS(text=texto_limpo, lang='pt-br', tld='com.br', slow=False)
tts.save(filepath)
return filename