Skill Flow Builder Polly Integration

Warning DEPRECATED The Alexa Games team will no longer support or maintain this official distribution of Skill Flow Builder. Thank you to all the folks who have used SFB to make great Alexa skills over the years!

The Skill Flow Builder Polly Integration. Enables Text To Speech in different voices & languages with Amazon Polly in both the SFB editor, and Skill Flow Builder built skills. Handles the making of calls out to Polly, the caching the audio files with S3, and the mixing of the audio responses using FFmpeg.

Visit Skill Flow Builder on Github for more information.

Getting Started

Prerequisites

The following needs to be installed and configured:

Node.js (with npm) # Note: Requires Node.js version >= 10.15.
Yarn

Compiling

yarn install && yarn compile

The compiled code is built into the dist/ directory. The Skill Flow Builder Polly integration requires other core modules to be built. To build all the modules in the packages/ directory at once, run yarn build-modules.

Testing

yarn test

This will run all unit tests in the test/ directory.

Package Structure

The SFB Polly Integration package structure looks like this.

packages/sfb-polly/
└── audioAccessor/ # Loaders for audio files
└── test/ # Tests
└── audioMixer.ts # Mixer to concat and interleave audio files
└── pollyUtil.ts # Polly client to load audio from text
└─── ...

Contributing

The SFB Polly Integration interacts with the AWS Polly service to generate speech from text, and helps with loading it’s audio locally and remotely. You can locally recompile the source code with yarn compile, but you will have to run yarn build-modules, in order for your changes to show appropriately in other modules.