Voice Over Internet Protocol Translation
Imagine being able to call someone in another country and hold a conversation, even though neither of you knows the other one’s language. With voice translation on VoiP, this will soon become possible.
Skype Translator offers translation of voice calls in seven languages. Some reports have called this “real-time,” but it doesn’t mean translation as you’re speaking, the way Star Trek’s Universal Translator works. That’s just impossible, because word order is different in different languages. German, for instance, tends to put the verb at the end of a sentence. What it means is translation quickly following the original.
Skype uses machine learning so that the translator familiarizes itself with your voice over time, giving better translation.
A startup company called Waverly Labs is offering a translator in a earpiece, called the Pilot. It’s oddly reminiscent of the Babel Fish in Douglas Adams’ novel, The Hitchhiker’s Guide to the Galaxy. That fish, once inserted in your ear, would instantly translate between your language and the language of any alien you spoke with. The Pilot even looks a little like a fish.
It’s no longer wild speculation to suppose that future VoIP systems will offer voice translation as an option. In fact, it’s hard to imagine that they won’t. There are, however, inherent limits to translation, and it’s never going to happen as smoothly as it does in space opera. Even humans who are fluent in both languages have problems when translating.
One of these problems is that people don’t actually enunciate every word. As a part of normal speech patterns, we slide over some words. “I saw him” often comes out as “I sawim” or “I sawm,” and we don’t even notice. A translator, human or machine, has to be able to deal with this.
Spoken language is full of ambiguity, which only the context of a conversation can resolve. To look at the same example again, “I saw him” might mean something entirely different when spoken by a stage magician explaining a trick. “Singe” in French can mean either “monkey” or “ape,” and the translator has to pick one.
Sometimes people use this ambiguity intentionally, for humor, to convey a dual meaning, or to be intentionally vague. Puns almost always fall flat in translation. Intentional vagueness is likely to become unintentionally specific.
Dialects and accents are another complication. Most languages have a lot of them, especially ones with wide geographic distribution, like English. Translators, human or electronic, can familiarize themselves only with so many of these and will have trouble with the rest. Translators also aren’t good at carrying the tone of a spoken statement into its translation, but at least people will hear the original as well.
Google Translate has a “block offensive words” option for its voice translation. This seems like an excellent idea, just to avoid disastrous mistakes.
Even with these limitations, voice translation can be a valuable tool when people need to get past a language barrier to talk to each other. Hopefully people will recognize the need for extra precision and clarity when using it, and they’ll ask for clarification on anything that doesn’t sound right.
It’s not a substitute for knowing something of the languages you’re trying to communicate in. In fact, it might be most helpful when people do have some knowledge of each other’s languages but aren’t fluent enough to hold a conversation. In this kind of case, people would have some idea of how the system is translating their words, rather than taking it on faith that the device is accurate.
What’s the best way to add VoIP to your business? Contact us to learn more.