[Bug]: Error with Deepgram Model for Transcription: Unsupported Type for audio_file

[TechBeme]

What happened?

I am attempting to use the Deepgram transcription API in my Python code, but I am encountering an error related to the audio_file parameter. When I make a request to the transcription endpoint, I receive the following error message:

litellm.APIConnectionError: Unsupported type for audio_file: <class '_io.BytesIO'>

I have tested all available models (e.g., nova-2), and the error persists for every request. The goal is to integrate Deepgram with OpenWebUI using OpenAI Endpoints, but this error prevents successful transcription.

Relevant log output

{
    "error": {
        "message": "litellm.APIConnectionError: Unsupported type for audio_file: <class '_io.BytesIO'>\nTraceback (most recent call last):\n  File \"/usr/lib/python3.13/site-packages/litellm/main.py\", line 4911, in atranscription\n    init_response = await loop.run_in_executor(None, func_with_context)\n                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"/usr/lib/python3.13/concurrent/futures/thread.py\", line 59, in run\n    result = self.fn(*self.args, **self.kwargs)\n  File \"/usr/lib/python3.13/site-packages/litellm/utils.py\", line 949, in wrapper\n    result = original_function(*args, **kwargs)\n  File \"/usr/lib/python3.13/site-packages/litellm/main.py\", line 5090, in transcription\n    response = base_llm_http_handler.audio_transcriptions(\n        model=model,\n    ...<19 lines>...\n        headers={},\n    )\n  File \"/usr/lib/python3.13/site-packages/litellm/llms/custom_httpx/llm_http_handler.py\", line 926, in audio_transcriptions\n    binary_data = self.handle_audio_file(audio_file)\n  File \"/usr/lib/python3.13/site-packages/litellm/llms/custom_httpx/llm_http_handler.py\", line 881, in handle_audio_file\n    raise TypeError(f\"Unsupported type for audio_file: {type(audio_file)}\")\nTypeError: Unsupported type for audio_file: <class '_io.BytesIO'>\n. Received Model Group=nova-2\nAvailable Model Group Fallbacks=None LiteLLM Retried: 1 times, LiteLLM Max Retries: 2",
        "type": "None",
        "param": "None",
        "code": "500"
    }
}

Code Snippet

Here’s the Python code I’m using to call the API:

import requests

file_path = "D:/test.mp3"
filename = "test.mp3"

with open(file_path, "rb") as file:
    response = requests.post(
        url="http://0.0.0.0:4000/v1/audio/transcriptions",
        headers={
            "Authorization": "Bearer sk-xxxxxxxx"
        },
        files={"file": (filename, file, "audio/mp3")},
        data={"model": "nova-2"},
    )

print(response.text)

config.yml file

model_list:
  - model_name: nova-2
    litellm_params:
      model: deepgram/nova-2
      api_key: os.environ/DEEPGRAM_API_KEY
    model_info:
      mode: audio_transcription

Are you a ML Ops Team?

No

What LiteLLM version are you on ?

v1.63.3

Twitter / LinkedIn details

No response