Audio Generation frequently returns audio as None

API Bugs
1

I’m using Audio Generation feature of chat completion which puts modalities as [‘audio’, ‘text’] in API. For the record, i’m using openai official python sdk.

Here is the code

            completion = await self.openai_client.chat.completions.create(
                model="gpt-4o-audio-preview",
                modalities=["text", "audio"],
                audio={"voice": "alloy", "format": "pcm16"},
                messages=prompt,
            )

            audio = completion.choices[0].message.audio

            if audio is None:
                logger.info("Audio generated from openai audio generation api is None")

I put assistant message as audio id like below

  {
    "role": "user",
    "content": [
      {
        "type": "text",
        "text": "Hello"
      }
    ]
  },
  {
    "role": "assistant",
    "audio": {
      "id": "audio_672d6aeabb9c81909f950c75e781aaf5"
    }
  }

Frequently, completion.choices[0].message.audio is None.
If this is related to the bug of Python SDK, please let me know so that i can report bug to SDK offical github.

Thanks : )