Overview of conversational agents
A conversational agent interacts with people using speech or text. Facebook Messenger would be an example of a text-based agent while Alexa and Siri are examples of agents that interact through speech. In either case, the agent needs to understand the user's intent and respond accordingly. Hence, the core part of the agent would be a natural language understanding (NLU) module. This module would interface with a natural language generation (NLG) module to supply a response back to the user. Voice agents differ from text-based agents in having an additional module that converts voice to text and vice versa. We can imagine the system having the following logical structure for a voice-activated agent:
Figure 9.1: Conceptual architecture of a conversational AI system
The main difference between a speech-based system and a text-based system is how the users communicate with the system. All the other parts to the right of the Speech Recognition...
