問題文
An agent must accept spoken questions from a call center. The team already has a text-based agent that works well. What is the minimal integration that keeps the existing agent unchanged?
選択肢
- Place a speech-to-text stage in front of the agent and pass the transcript on.
- Rebuild the agent with a model that accepts spoken audio input natively so no extra stage is needed.
- Send the audio directly to the agent and let it decide from the sample rate how it should be interpreted.
- Have the caller type the question into a web form while they are on the line, so that the existing text agent receives exactly the input format that it was built for and no additional processing stage has to be introduced anywhere.