Dialogue system architecture

Report a typo

Imagine you address an AI assistant with a question. It will go through several stages. In what order should these stages be placed?

Tip: Everything starts and ends with the speech phase: first, the assistant should recognize your question and then answer out loud. There are three more steps sandwiched between the input and output, the dialogue manager being probably the most unclear. So here is a hint: place it in the very middle as it controls the entire course of conversation (you can read more on Wikipedia).

Put the items in the correct order
Automatic speech recognition
Natural language understanding
Dialogue manager
Speech synthesis
Natural language generation
___

Create a free account to access the full topic