919775 - [AccessFu] speech output data structure needs to be reworked

Status: RESOLVED INVALID

(Core :: Disability Access APIs, defect)

Description

[Eitan Isaacson [:eeejay]]

Currently, we send a list of "actions". An action could have a method of 'speak' or 'playEarcon'. The 'speak' method could have a flag called 'enqueue', if it is false or absent, the speech synthesis is cancelled, and the new utterance spoken. When the pivot changes, we get an earcon followed by a speak with an enqueue of false. This is incorrect since the earcon is what should be interrupting the synth first.

Furthermore, when a speak is preceded by an earcon there is an opportunity to precache some synthesized text while the earcon is being played in order to cut some of the latency from the speech engine.

Here is a new structure proposal:

{
  earcon: 'tick', // name of earcon
  text: ['list of utterances to queue'],
  interrupt: true // interrupt the speech synthesizer from a previous call
}

Comment 1

[Eitan Isaacson [:eeejay]]

We stopped using AccessFu.