Greg Bulmash from Amazon spoke at OSCON 2017 Conference last week about developing your own virtual assistant applications using Amazon's Lex and Polly technologies. He talked about how to plan and execute voice interfaces for these browser-based "Do it Yourself Virtual Assistant" (DIYVA) applications that listen and respond in natural language. They are based on Alexa's voice recognition, Artificial Intelligence (AI), and speech synthesis technologies.
Amazon Lex is used for building conversational interfaces into applications using voice and text. It provides the deep learning functionalities of automatic speech recognition (ASR) for converting speech to text, and natural language understanding (NLU) to recognize the intent of the text.
Amazon Polly is an Amazon AI service that uses deep learning technologies to synthesize speech that sounds like a human voice.
Bulmash showed how to build a cross-platform Cordova based application consuming Alexa services from Amazon, that works on both mobile and desktop clients to perform different tasks like ordering flowers. The sample application includes the following three functions:
- Get user credentials
- Get audio message from user
- Send audio to Lex
You can add security to the application by creating a security profile on Amazon Developer website and setting up an AWS Cognito account for federated identity and access control. Select a user role and then attach the "AmazonLexRunBotsOnly" policy to the role. Then you can setup Amazon Lex Bot by using a custom slot and picking the Bot settings like language, output voice, and session timeout.
Bulmash concluded the presentation by mentioning the upcoming AWS ChatbotChallenge. Contestants can build conversational, intelligent chatbots with Amazon Lex and AWS Lambda technologies and the submissions are due by July 18.