Understanding spoken natural language queries has been a common defining trait of “computers of the future” ever since computing technology itself originated in the 1950’s.
From the hyper-intelligent machines of Heinlein’s The Moon Is a Harsh Mistress and Asimov’s The Last Question, to the iconic robotic sidekicks R2-D2 and C-3PO, to the darker visions of Skynet and HAL, science fiction is replete with computers, robots, and spaceships that demonstrate a human-like comprehension of spoken language.
Unlike some of sci-fi’s other fanciful ideas, like teleportation devices and time machines, there have been real concerted efforts over many decades to bring a rich understanding of spoken language to the world of machines. Progress has been slow, and many time estimates overly optimistic – Steve Jobs figured we’d have speech recognition squared away by 1993 – but, despite it all, the dream of the voice interface stayed alive.
In 2015, we’re finally here. Over the past year alone, speech recognition accuracy has skyrocketed, eclipsing all progress made over the previous twenty years combined. Improvements in natural language understanding – one of the chief areas our MindMeld team works on – have followed closely on speech rec’s heels.
A recent piece in Electronics360 charts the evolution of voice interfaces, explaining how this change so suddenly came about. “Very soon,” the article states, “the most natural form of communication—speech—will dominate human-machine interactions, and the pace of this change is taking everyone’s breath away.” We agree, and are excited as machines learn to communicate with humans fluently in our own language, just as science fiction has always predicted.
