Alexa’s ability to act on spoken requests depends on statistical models that translate speech to text and text to actions. Historically, the models’ decisions were one-size-fits-all: the same utterance would produce the same action, regardless of context.Read More