The python API is not the most intuitive, maybe use standardized formats? E.g.

from pathlib import Path

from openai import OpenAI

client = OpenAI()

speech_file_path = Path(__file__).parent / "speech.mp3"

response = client.audio.speech.create(

model="tts-1",

voice="alloy",

input="Today is a wonderful day to build something people love!"

)

response.stream_to_file(speech_file_path)