Menu

Artificial Speech Translation Is Finally Here – How Effective Is It?

People always wanted to achieve big things. Ever since they’ve seen how far technology can go, they’ve always worked to improve it. Artificial language translation is only one of the things humanity seeks, and it seems that its time has come. However, is it useful just yet?

Marek Kohn has had a discussion with Alex Waibel about artificial translation and the difficulties in achieving it. Although it can be achieved, Waibel talks about how noise can affect the abilities of artificial language translation. Apparently, background noise is a challenge, not to mention the distance between the devices used to perform the translation. While it might be able to work in a meeting room, laboratory or anything of the sort, a big distance won’t allow it to perform fully.

It seems the full development still needs some time, but it’s nice to see what has been achieved up to this point.

Alex Waibel – What Has He Achieved So Far?

Waibel is known to do research on speech recognition and translation. The professor of computer science at the Karlsruhe Institute of Technology has successfully obtained a device that performs artificial speech translation, but it’s still not complete. It needs improvements, given it is unable to perform as long as the users are at a long distance between each other, and there’s background noise too.

Marek Kohn and Professor Waibel were talking to each other while being 10,000 km apart. This made it difficult for them to have Waibel speak in German, and have his words delivered in English.

Of course, the technology idea is not impossible at the Karlsruhe Institute of Technology. Waibel, together with his colleagues, are able to deliver lessons in German, while students can follow them in English thanks to an electronic translator. The device works by translating the texts on laptops. Simply put, the text is generated while the people in the room are talking, working the same way as subtitles do. In order to make it work, though, the speaker has to talk clearly.

Waibel has achieved other things in the past too, though. He was able to come up with a translation system back in 1991, being the first speech translation system ever. The way it worked is by using large workstations and a 500-word vocabulary. However, it took quite a while to process the information it just heard.

So, it’s safe to say that the idea of artificial speech translation isn’t just something recent – on the contrary, it’s been around for quite a while. Waibel is said to have “sort of invented it”. Douglas Adams is said to have done the same around the same time too. A Babel fish was introduced in “The Hitchhiker’s Guide to the Galaxy”. It could have been placed in the ear, giving the user the ability to understand all languages. So, it’s safe to say people have thought about this technology long before it’s been developed.

Is Artificial Speech Translation Available Yet?

We still haven’t reached the point where understanding all languages by simply hearing them through a device is possible. Whereas the concept is worked on and possible, there’s still a lot to do until reaching that point. There are some systems that show how the concept is in the works, such as Google Translate.

The director of product from Google Translate, Barak Turovsky, said that “We’re generally very early in the paradigm of voice-enabled devices, but it’s growing very rapidly, and translation will be one of the key parts of this journey.”

What Needs to Be Improved in Order for Artificial Speech Translation to Become the Norm?

According to Waibel, there are obstacles that they need to overcome before they’re going to be successful in their journey. For example, technology has to be possible to use offline in case you have no internet access. Not to mention, it also has to be able to be simultaneous, thus not affected by the background noises, like the TV, radio or anything of the sort.

Another thing to achieve is making the system able to address people in an appropriate way. Making the system able to address any concerns about how much private speech data accumulates in the cloud is essential too.

Final Thoughts

Artificial speech translation might still seem a technology of the future. Still, when it becomes the norm, it will help people communicate more effectively, breaking language barriers this way. They are meant to make people’s lives easier, allowing them to buy things and understand certain information given by foreign individuals.

Of course, this isn’t better than actually knowing the language they are hearing, as the social value doesn’t decrease when you are able to communicate yourself. It’s much easier to communicate when you speak a language, and you also won’t have to worry about translation cost per word that come with using translation services. Someone who speaks a foreign language will always have an advantage, don’t forget that.

Leave a Reply

Your email address will not be published. Required fields are marked *