API.AI Speech Context

Generates a Google Cloud Speech API SpeechContext for your API.AI agent.

Usage

generateSpeechContext takes two parameters:

• summary: An API.AI agent summary object with the keys name, entities, and intents.
• blacklist: An array of words to blacklist from the context.

The summary can be created with api-ai-agent-downloader.

It returns an array containing words and phrases based on your agent's entities and intents. Non-blacklisted single words are given highest priority, then phrases that contain at least one non-blacklisted word. Entities within intent templates (e.g. '@sys.any') are removed.

The array comports with the Google Cloud Speech API's SpeechContext usage limits.

Example

const agentDownloader = require('api-ai-agent-downloader');
const { generateSpeechContext } = require('api-ai-speech-context');
const fs = require('fs');

const agentName = 'agent_foo';
const developerToken = 'abc123';

const blacklist = ['bad', 'words', 'and phrases'];
const removeCommonWords = true;  // true by default

agentDownloader.getSummary(agentName, developerToken)
    .then(summary => {
      const context = generateSpeechContext(summary, blacklist, removeCommonWords);
      fs.writeFileSync('./speechContext.json', JSON.stringify(context));
    })
    .catch(error => console.error(error));