Posted On: Feb 8, 2018
You can now define a response for your Amazon Lex chatbot directly from the AWS Management Console. A response consists of messages dynamically selected from a group of pre-defined messages, populated by the developer. Messages in these message groups can be simple text, or you can use custom markup. In addition, you can also display a response card as part of the response. The responses capability simplifies building dynamic conversations with Amazon Lex. In addition, you can use Speech Synthesis Markup Language (SSML) in a text response.
To define responses in Amazon Lex, by example, if we are building a chatbot that takes car and hotel reservations, you might build three message groups. The first message group could contain different ways the bot could greet the user: “Hello”, “Hi”, “Greetings”. The second message group could contain the different forms of introduction: “My name is ReservationBot”, “I am ReservationBot”, “This is ReservationBot”. And the third message group could communicate the capabilities: “I can help with car rentals and hotel reservations”, “You can do your car rentals and hotel reservations with me”, “Currently I can support car rentals and hotel reservations.” Amazon Lex can use one message from each of these message groups to dynamically build a response in a conversation. A sample response, then, would be, “Hello! My name is ReservationBot. I can help with car rentals and hotel reservations”.
In addition to supporting SSML in speech response, Amazon Lex now supports SSML in a text response. You can use SSML response as input to a Text-to- Speech (TTS) engine to personalize conversations with end users. Amazon Connect customers can now use SSML when leveraging an Amazon Lex chatbot as part of their contact center experience to control aspects of speech, such as pronunciation, volume, and speech rate.
The responses and SSML functionalities for Amazon Lex are available today in the US East (N. Virginia) and EU West (Ireland) AWS Regions. For more information please refer to the Amazon Lex documentation. For a complete list of supported SSML tags, visit the documentation.