third_party_tools/text_to_audio_file.py

import os
import base64
from elevenlabs.client import ElevenLabs

def text_to_audio_file(text: str, path: str = "output.mp3") -> str | None:
    """
    Generate audio and save as a file. Returns:
    - (filepath) if ok
    - (None) if failed
    """
    if not text:
        return None

    api_key = os.getenv("ELEVEN_LABS_API_KEY")
    if not api_key:
        print("API KEY MISSING")
        return None

    try:
        client = ElevenLabs(api_key=api_key)

        audio_stream = client.text_to_speech.convert(
            text=text,
            voice_id="NOpBlnGInO9m6vDvFkFC",
            model_id="eleven_multilingual_v2",
            output_format="mp3_44100_128"
        )
        audio_bytes = b"".join(audio_stream)

        # FOR TESTING LOCAL FILE:
        # with open("output.mp3", "rb") as f:
        #     audio_bytes = f.read()

        with open(path, "wb") as f:
            f.write(audio_bytes)
        print('path', path)
        return path

    except Exception as e:
        print("Audio generation failed:", e)
        return None