How Does A Spoken Dialogue System Work?
We illustrate how a Spoken Dialogue System works using a simple example: appointment scheduling.
Step 1. The User Talks to The System
User: I really need to see the physiotherapist soon – my back is so sore! When is the next available appointment?
The Speech Recognition component of the dialogue system takes this utterance and determines the words that were spoken. The most likely sequence of words is then passed on to the Natural Language Understanding component. This translates the speech into a representation that the computer can understand, for example: utterance type 'appointment request', requested person 'physiotherapist', time: 'as soon as possible'.
Step 2. The System Processes The User's Request
The representation generated in the previous step is then passed on to the Dialogue Manager. This initiates the next step: looking up the next available appointment for the physiotherapist in the calendar. The calendar reports back that the next available physiotherapy appointment is on Thursday at 2PM. Since an appointment could be found, the Dialogue Manager decides that the next step is to offer the user an appointment with the physiotherapist on Thursday at 2PM, and to ask the user whether that appointment is acceptable.
Step 3. The System Answers The User
The Natural Language Generation system creates a sentence that offers an appointment with the physiotherapist at the next available time, on Thursday at 2PM. This sentence is then passed on to the Speech Synthesis system, which reads the sentence out to the user.
System: The next available appointment with the physiotherapist is on Thursday at 2PM. Is that acceptable to you?
