ULG's Language Services Blog

Why Real-Time Translation Earpieces May Not Be Worth the Hype

 

 

 

Inevitably, when a new real-time translation earpiece hits the market, media outlets gobble up the news, putting out article after article referring to the new device as the answer to breaking down language barriers, and hailing it as the realization of sci-fi devices found in Star Trek or The Hitchhiker’s Guide to the Galaxy.

These earpieces rely on Machine Translation (MT), which is the process of training a computer to translate content instead of a human. Through MT, developers promise to create earpieces that will completely automate the translation process in real time.

Soon after the release of each new device, the hype dies down as that promise falls short, but that doesn’t stop excitement every time a translation earpiece is announced.

For the time being, the question of whether quality real-time translation earpieces are possible is a resounding no. But why?

The Wrong Type of Content for Machine Translation

The first hurdle for real-time, automated translation earpieces comes from the current limits on MT.

Major strides have been made in the past few years when it comes to MT’s capabilities. However, while MT can work great for some content, it doesn’t perform as well with other types of content.

Internal company documents and technical, simple language are where MT does well. And even then, the programs need to be trained to recognize special terms, phrases, and context. MT still has a long way to go before it can be trusted to perfectly translate such documents.

When it comes to translation earpieces, MT will struggle even further. For one, earpieces will have to account for the fact that conversations often aren’t grammatically correct or consistent. Casual content will use a great deal of idioms and slang, while conversational speech functions with speech disfluency, accents, dialects, idiolects, and more.

 

These all present huge challenges for any MT program, as those problems aren’t usually seen in text. All these contribute to creating quite the problem for real-time translation.

Lack of Voice Recognition and Audio Capabilities

Are translation earpieces really capable of consistent and precise voice recognition? At this stage, while these earpieces might be able to recognize some standard differences in pronunciation, the question remains on how accurately the pronunciation will be picked up and synchronized with MT in real time.

Every single language has a myriad of dialects, and many have slightly or completely different words for particular concepts or object. Not only would the voice recognition have to pick up on all these variances, but the MT technology would have to understand these small differences and accurately translate it.

Another problem comes from potential audio issues. To have an effective translation earpiece, the device will have to pick up voices from a conversational distance, and background noises have already proven to be an issue. While one solution is to use noise-canceling earbuds, they have to be equipped to differentiate between background noise and the other speaker’s voice.

Are Real-Time Translation Earpieces Truly Possible?

Ultimately, the current iterations of real-time translation earpieces are all stylish, but have little viable use. While the earpieces can work to a certain degree in controlled environments, these devices aren’t yet ready to dismantle language barriers single-handedly.

Could real-time translation earpieces become viable in the future? That will depend on what future developments are in store for MT.

MT’s limitations when it comes to complicated, nuanced language will take time to improve, and that’s just when it comes to text. Speech will take longer. On this front, human translators and interpreters are still the best option for overcoming language barriers, but that isn’t a solution for translation earpieces.

There’s no way around it: the demand for translation earpieces means that companies will continue to try and create a quality product, and if MT progresses far enough, this could be a globally transformative device. But the science fiction notion of perfect real-time translation through these earpieces is just that for now—science fiction.