Amazon Polly Challenge

In this challenge, we'll be exploring Amazon Polly to convert text to audio and then building the exact case study show below:

Helpful Links

.NET AWS Polly SDK Docs .NET AWS SDK Docs

Prerequisites

The following tools and accounts are required to complete these instructions.

• Install .NET Core 3.1
• LambdaSharp Tool
• AWS Account
• GitHub Account

Install SDK & Tools

Make sure the following tools are installed.

• Download and install the .NET Core SDK
• Download and install Git Command Line Interface

Setup AWS Account

The challenge requires an AWS account. AWS provides a Free Tier, which is sufficient for most challenges.

• Create an AWS Account

Clone GitHub Repository

Next, you will need to clone this repo into your working directory:

git clone https://github.com/LambdaSharp/Polly-Challenge.git

Setup LambdaSharp Deployment Tier

The following command uses the dotnet CLI to install the LambdaSharp CLI.

dotnet tool install --global LambdaSharp.Tool

NOTE: if you have installed LambdaSharp.Tool in the past, you will need to run dotnet tool update -g LambdaSharp.Tool instead.

The following command uses the LambdaSharp CLI to create a new deployment tier on the default AWS account. Specify another account using --aws-profile ACCOUNT_NAME.

lash init --quick-start

Level 0 - Checkout Amazon Polly in the AWS Console

• Search for "Amazon Polly" in AWS Console
• Once in the Amazon Polly console, add some text to the "Plain Text" input area then click "Listen to speech".
• Try different "Voices" and "Language and Region" then click "Listen to speech" again.
• Ensure you can play the audio file locally by clicking "Download MP3" and opening the file.

Level 1 - Deploy Challenge to AWS

• In the cloned repo working directory, Polly-Challenge, deploy LambdaSharp by running the following command: lash deploy. This will take a minute.

• Fill out your phone number to receive SMS messages! No spaces or dashes i.e 15551234567

  *** NOTIFICATION SETTINGS ***
  |=> NotificationSms [String]: Phone number for PollyFunction to send SMS messages for Polly audio files:

• Once completed, navigate to the API Gateway Deployment section in the AWS Console. Here is a link to the API Gateway console in US-EAST-1: https://console.aws.amazon.com/apigateway/main/apis?region=us-east-1

• Test Amazon Polly by sending a web request to the API Gateway endpoint like this: curl -d '{"Content": "Hello world! This is some test content.", "FileName": "test.mp3"}' -H "Content-Type: application/json" -X POST https://REPLACEME.execute-api.us-east-1.amazonaws.com/LATEST/text-to-speech

  • Be sure to replace the REPLACEME with the actual subdomain or entire url. Works with Git Bash, *Nix, and Postman.

You can now navigate to the AWS S3 console ArticlesBucket for the saved file test.mp3

You can also do another curl request to get a list of files from that bucket curl -X GET https://REPLACEME.execute-api.us-east-1.amazonaws.com/LATEST/files

Level 2 - Let's Get Notified

Let's get notified when the converted text-to-speech file is ready. Update the SMS message in the SNS message to have a link to the MP3 file. Some setup has already been completed for you in the Logic.cs file in the PollyFunction folder.

Docs: Amazon SNS Publish

Level 3 - Change Polly's Voice

We want the user to be able to choose a language, voice style, and get audio based on that selection.

Use Amazon Polly's built in localization methods to describe the voices available for that language code.

Pick a voice from that list and use it to synthesize the text to speech in the chosen language.

Using the curl request from Level 1, add a property for LanguageCode and VoiceId. Be sure to update the ConvertTextRequest to have these fields to parse in the Logic.cs file in the method AddTextToSpeechFileToBucket.

NOTE: This should not require any text translation

[Docs] https://docs.aws.amazon.com/sdkfornet/v3/apidocs/index.html?page=Polly/MPollyDescribeVoicesDescribeVoicesRequest.html&tocid=Amazon_Polly_AmazonPollyClient

[Language_Codes] https://docs.aws.amazon.com/polly/latest/dg/SupportedLanguage.html

Level 4 - No Rework

We don't want to expend processing power on duplicate files! If the content of the incoming article is identical to one that has already been saved, then do not send for audio processing but still send a link to the MP3 file in a SMS notification.

Level 5 - News Feed Summary

We want to listen to a news feed summary instead of reading them. Make an API request to a news feeds service, get the title and date (parsing required), create an audio file then send a SMS with a link to listen to it! Update the existing method AddNewsFeedAudioToBucket in the Logic.cs file.

Here is a few APIs to look at:

• Hacker news feed: https://api.hnpwa.com/v0/news/1.json
• Techcrunch API Call: https://api.rss2json.com/v1/api.json?rss_url=https%3A%2F%2Ftechcrunch.com%2Ffeed%2F

You can trigger this using: curl -X GET https://REPLACEME.execute-api.us-east-1.amazonaws.com/LATEST/news-feed.

Boss - Translate

We want to be able to listen to the audio in the language of our choice. Modify the method in the previous level to use Amazon Translate to translate the text into another language before sending it to Amazon Polly.

Resource: https://docs.aws.amazon.com/translate/latest/dg/examples-polly.html