Language is one of the biggest barriers that humans face as the world continues to become more closely knit with each other than ever before. It would be a great boon for all of us and help reduce misunderstandings if we could somehow make it possible for people to understand what others from around the world were saying. A Bluetooth earpiece called ‘Pilot’ is attracting a lot of attention on crowdfunding site Indiegogo as it promises to solve real-time language translation. The way on which the product works, according to the developers at least, is that Pilot comes with a set of earpieces, one of which goes in your ear and the other in the ear of the person you are attempting to talk to.
Once the earpieces are in place, the idea is that both the people will be able to talk in their language and the other person will hear a version translated to their language in real time. Sounds amazing? It is and also something that makes me a little skeptical. The Pilot will launch with real-time translation for English, Hebrew, Italian, French, Spanish and Portuguese somewhere around 2017 and will then continue to add languages from other regions over a period of time.
Machine learning has come a long way thanks to the much larger amount of data processing power as well as the combined knowledge in the field of speech recognition built over decades of work. Unfortunately, though, there is still no answer as to how machines will be able to understand the idiosyncrasies of language itself.
The way in which companies approach the problem of language translation is to treat a particular language as a data set and then learn to recognize the various ways in which this data can be arranged. The nature of language itself though is unique, in that it keeps on evolving over time, varies from place to place and depends a lot on the emotional inference from subtle cues used by people subconsciously.
A computer program needs specific rules to follow and cannot understand these minor variations. Microsoft with Skype has been pioneering work in real-time translation as well and a couple of languages are already online for use between users speaking different languages.
A program like Skype though has access to the computing power in the cloud, something that can be time-consuming between smaller, portable processors as those being used in upcoming gadgets. Even Google’s own translator app is still limited to translating audio into written text that can be converted into a large number of languages.
Researchers though are positive that as the IoT picks up and billions of sensors are deployed all over the world, many of them voice activated, a quantum leap to accelerate machine understanding of language is bound to happen. Even then, though, they say that it is unlikely to be absolutely perfect or infallible.
Too good to be true?
Despite the advances that language translation technology has made, it seems improbable that a device like the Pilot can actually deliver seamless real-time conversation at the moment. Of course, it would be a great joy to be proved wrong and to actually be able to experience a Star Trek kind of experience where everyone seems to speak your language. Immediately and perfectly.