Today, the technology has changed our lives to its best and has been contributing to make it even better. Gone are the days when the computer systems were restricted to only arithmetic calculations. Nowadays, with the invention of the artificial intelligence (AI), computers can talk, listen and even respond, isn’t it great? Powered by one of the great competitors in the cloud computing market, the Amazon Web Services (AWS), Amazon Lex is one such example!
Amazon Lex is a web service, which enables its users to create the conversational interfaces into any applications using voice and texts. It uses the same conversational engine as Amazon Alexa and equips with the deep learning technologies viz. Automatic Speech Recognition (ASR) that converts a speech to text and Natural Language Understanding (NLU) that understands the spoken language and responds accordingly.
First introduced in November, the Amazon Lex Bot is a perfect solution for chatbots in messaging, especially in e-commerce bots, where the chatbots have limited ability to respond. What makes AWS Lex more sophisticated and classy than the ones already available in the market is its unique feature and top-notch technology. Amazon Lex enables the developers to build such bots that do not only convert the speech to text but also identify the purpose of the text that makes the process more conversational. Let us find out more about the technology and see Amazon Polly:
Amazon Lex is equipped with several AWS services that give it a great exposure to work efficiently. The Amazon Lex let the developers use AWS Lambda that recognizes the intent of the input, Amazon Cognito that helps authenticates the users, Amazon Polly that converts the input text to speech. Let us understand how Amazon Lex works:
Developers are required to follow some steps to work with Amazon Lex, such as creating a bot and configuring it with any number of intents you want, to make the Lex understand the client’s intent (objective). Further, it requires the developers to test the bot using the test window client that Amazon Lex console provides, create an alias, and deploy the bot on mobile applications or messaging platforms like the Facebook messenger.
Let us also find out the Amazon Lex Guidelines required to follow while using AWS Lex here!
Signing Requests – The Amazon Lex model-building and runtime API operations uses signature V4, which is a process to add verification of the information, to validate requests made by the users.
PostContent – The developers need to use the unsigned payload option for PostContent, which sends the user input to Amazon Lex. The developers, here, require using the literal string “UNSIGNED-PAYLOAD” for hashing payload. It also requires developers to include a header viz. x-amz-content-sha256 and the value as “UNSIGNED-PAYLOAD” while processing the PostContent request.
Though Amazon Lex is going to be the next big thing in the field of artificial intelligence, for sure, it has some limitations for now. Let us take a sneak peek into these Amazon Lex limits:
Availability: At present, the AWS Lex is providing its Model Building and Runtime Services in the US East (N. Virginia) and EU (Ireland) only.
Amazon Lex Supported Languages: As of now, the Amazon Lex supports the US English language only.
|Speech Input to PostContent||15 Seconds|
|Text Input to PostText||1024 Unicode Characters|
|Maximum size of PostContent Headers||16 KB|
|Maximum size of Request and Session Headers||12 KB|
|Maximum Input Size to Lambda function||12 KB|
|Maximum Output Size to Lambda function||25 KB (12 KB for Session Attributes)|
However, the best part of the AWS Lex Bot is that Amazon Services powers the Lex and hence, you need to pay as per the number of text or voice requests that your bot processes. The company has set the Amazon Lex pricing in a way that you will be charged at the cost of $0.004 per voice request, and $.00075 per text request.
Amazon Lex Pricing Sample
|Input Requests Types||Cost per request||Number of requests||Total|
|4,000 Speech Requests||$0.004||4,000 requests||$16.00|
|1,000 Text Requests||$0.00075||1,000 requests||$0.75|