Encoding Speech (before and after) Channels

A speech channel enocdes a natural language sound before (speechBefore) or after (speechAfter) the main tone. These channels works for discrete (only) and relatively timed (ideally) tones. The speech rate is controled by the config object.

speechBefore usage pattern

JSON

{
  ...
  "encoding" : {
    "speechBefore": {
      "field": "name",
      "type": "nominal",
      "language": "en-US"
    }
  }
  ...
  "config": {
    "speechRate": 1.75 // default
  }
}

JavaScript

let stream = new Erie.Stream();
...
stream.encoding.speechBefore.field("name", "nominal");
stream.encoding.speechBefore.language("en-US");
...
stream.config("speechRate", 1.75);

Language

Using the language property (in BCP-47 format), one can set the language of a speech. The default language is determined by the browser document’s language. For example if a web page states <html lang="en">, then the default language can be en-GB (British English). Because this channel uses Web Speech API (see this MDN documentation), supported languages include Arabic (Soudi Arabia), Bangla (Bangladesh, India), Czech, Danish, German (Germany, Austria, Switzerland), Greek, English (Australia, Canada, UK, Ireland, India, New Zealand, United States, South Africa), Spanish (Spain, Argentina, Chile, Columbia, Mexico, United States), Finnish, French (France, Belgium, Canada, Switzerland), Hebrew, Hindi, Hungarian, Indonesian, Italian (Italy, Switzerland), Japanese, Korean, Dutch (The Netherlands, Belgium), Norwegian, Plish, Portugese (Portugal, Brazil), Romanian, Russian, Slovak, Swedish, Tamil (India, Sri Lanka), Thai, Turkish, Chinese-Mandarin (China, Taiwan) and Chinese-Cantonese (Hong Kong).

© Hyeok Kim