How to make audio file input with gpt-4o

Question

Hi! I was trying to use Azure AI Studio gpt4-o model with audio input. It looks like this:

messages = [{"role": "user",

                 "content": [{"type": "input_audio", "input_audio": {

                        "data": encoded_string,

                        "format": "wav"}}]

                }]
This code was taken from OpenAI documentation https://platform.openai.com/docs/guides/audio . Although in AI studio it says that audio channel is available, I get the following response: "Error Error code: 400 - {'error': {'message': "Invalid value: 'input_audio'. Supported values are: 'text', 'image_url', 'audio_url', and 'refusal'.", 'type': 'invalid_request_error', 'param': 'messages[1].content[0].type', 'code': 'invalid_value'}}". I'm using "2024-10-01-preview" api version.

Is usage of audio files, not urls not available for now and will it be available in future?

Answer 1

Hello @Vitalii Rentiuk , Thanks for using Microsoft Q&A Platform.

If you are working with SDK then the supported audio formats are listed in the sample GitHub repository. Here are the supported formats with python SDK:

To learn more about how to use the Audio file as input here is the reference sample code: https://github.com/azure-samples/aoai-realtime-audio-sdk

Is this something you are looking for?