Please provide a detailed explanation on how to connect to and utilize the OpenAI real-time audio model. Include steps for setting up the environment, any necessary API keys, example code snippets, and tips for effective usage. Additionally, discuss any potential limitations or best practices to consider while working with the model.