GPT-4o mini Audio is an OpenAI audio-enabled model based on GPT-4o mini. It accepts audio in prompts and returns audio directly, avoiding separate transcription and text-to-speech steps. GPT-4o Audio is more capable, but GPT-4o mini Audio is the economical choice for direct voice interactions. section 8.1.3
Use GPT-4o mini Audio in a Spring AI application by:
- Adding Spring AI's OpenAI starter and setting spring.ai.openai.chat.options.model to gpt-4o-mini-audio-preview. section 1.2.1
Board Game Buddy audio questions
An /audioAsk endpoint accepts a game title and an MP3 question, passes the audio to GPT-4o mini Audio through ChatClient, and writes the returned answer to answer.mp3.

