In high school, I remember reading Pygmalion by George Bernard Shaw. In it, not to spoil the plot, there’s a character, Professor Higgins, who can determine someone’s place of upbringing, vocation, and wealth by how they speak. I’ve been wondering… can Amazon do the same?
Amazon has data on potentially billions of voice samples and knows roughly the age, gender, vocation, and wealth, and location of all of the Alexa users. Could it create a model that could determine within 5–8 word utterances of the person’s wealth, health, and upbringing?
Beyond being a stand in for the next run of My Fair Lady, what would be the other benefits for Alexa? Maybe it could adjust its own TTS to match that of the speaker? Maybe it could better market to Alexa users through Alexa endpoints? Or, it could use culturally appropriate pronunciations or terminology, y’all.