In The Hitchhiker’s Guide to the Galaxy, one of the bibles of technology enthusiasts, author Douglas Adams conceived of the Babel Fish, a creature that when introduced into the ear canal of a person listens to the surrounding conversations and whispers a perfect translation in the native language of its host. In the real world, technological versions of this science fiction creature are beginning to arrive on the market: wireless headphones that, from within the ear, carry out the simultaneous translation of different languages.
The first was the Pilot, a prototype from Waverly Labs released in 2016, which currently sells for about $200. The device uses a sophisticated set of microphones, along with noise cancellation algorithms, to listen to the words spoken by the user and those spoken around them. “These words go to the cloud, where they are processed with the voice recognition system, automatic translation and speech synthesis, and then sent to the user and any other person whose Pilot device is synchronised in the conversation. All in a matter of milliseconds,” explains Andrew Ochoa, CEO of Waverly Labs.
In October 2017, Google launched its Pixel Buds, a set of Bluetooth earbuds capable of simultaneous translation of 40 languages using a smartphone. “It’s like having your own personal translator. If you’re in Italy, for example, and you want to order pasta like a professional, you just need to say: ‘Help me speak Italian,’” explained Adam Champy, product manager at Google. A few months earlier, the German company Bragui had already released the Dash Pro, a similar device that uses the iTranslate application of the iPhone to also translate up to 40 languages. On the other side of the world, the Australian company Lingmo International launched TranslateOne2One, capable of translating eight languages.
A translator for humanitarian actions
Many experts believe that this technology will revolutionise how we communicate. Jonathan Luff, a consultant in technological innovations, is one of them. “Imagine what impact this will have. You can talk, understand, learn and do business with anyone, anywhere in the world and at any time,” he says.
Beyond tourism or the business world, Dan Simonson, a computational linguist at Georgetown University (USA), highlights the potential of technology for humanitarian actions. “This work is often performed by soldiers who don’t have the time or the resources to learn the language of the place they’re destined for. For people in this situation, having access to even poor translation tools—in languages for which there is not much data available to create translation tools— could immediately improve the efficiency of such relief efforts, saving thousands of lives as a result,” he points out.
But while the idea of tools like the Babel Fish is exciting, reality still presents many challenges for this technological utopia to be fulfilled. Experts agree that the current state of machine translation is still very primitive. Two months after its premiere, the reviews of influential media specialised in technology showed that Google’s Pixel Buds did not fulfil what was promised. Wired UK wrote: “They’re not only bad, but totally useless.” The Guardian considered the earbuds “defective” and “a missed opportunity,” and Gizmodo said “they’re not even close to being good.” And where Google fails, others are unlikely to succeed.
“The best we have is in line with something like Google Translate. It’s fast and much better than the previous ones, but it’s still not comparable to what a human translator can do,” says Simonson. More than translation per se, the linguist explains that the main difficulty is to interpret the spirit and cultural intention of a discourse and adapt it from one language to another, regardless of what the words mean in their literal sense. To do so, the software would need to interact with a branch of linguistics called discourse analysis. But if artificial intelligence has now begun to excel in board games and video games, dealing with something as dynamic as human communication is far more complex.
Besides the fact that discourse analysis has not been applied in the field of machine translation, another factor is that human speech has a voice, a tone and a certain audible colour in its expressions that help the listener to intuit the meaning of a message. “It’s possible to translate words and syntax correctly, that is, the order of words and the grammatical patterns of a language, but that still doesn’t mean adequately translating a whole style of conversation,” explains Simonson.
These obstacles mean that experts do not dare to predict when these tools will become a practical reality, but they continue to develop systems of advanced speech recognition, natural language generation and other technologies so that the utopia of a fluent conversation, in different languages and facilitated by technology can come true. The Babel Fish does seem to be on the way.