With technology traveling at sometimes unmanageable peaks, there is no limit to where it can take us, which routes it can short cut and, at the same time, to which technical detours it can take us. Will our landscape change forever? We provide technical information about voice activation technology in translation and what it can do for us.
Although the rise of phone voice activation systems has been noted with the iPhone’s Siri and Android’s Voice Actions, dictation software has been around for a while through the forms of voice dialing, computer dictation, and driving. automated voice calls. So what has sparked this new urge to ask your phone instead of tapping into a Google search? How do you translate our voice and our need to your system?
Apparently, the programs that help us to dictate words or activate voice commands usually fall into two categories:
Small vocabulary for many users
This deals with the translation of short answers for automated voice calls, such as basic numbers or categories that guide the user to the caller in the end.
Large vocabulary for a limited number of users
These systems typically operate in environments with a smaller user base, that is, in an enterprise environment. This is because the accuracy of the technical translation will be compromised with any other users as it is trained to work with the main users. These programs tend to have tens of thousands of words in their vocabulary with a high accuracy rate.
Now, the technical stuff! This is done by matching or translating sound waves, the vibrations that are created when we speak into a device called an ADC – Analog to Digital Converter. With the phonemes (the smallest unit of language) combined with the context of the other words, you can put the puzzle pieces together in your translation – very elegant!
So how has technology progressed in the 10 odd years it’s been around? One of the main features is the ability to pick up continuous speech. I remember using speech recognition at school in one of the Word programs, just seeing how it worked. If the words were spoken at a general conversational rate, the computer would struggle to understand and generate a series of garbled words on the screen. Since then, speech recognition systems seem to pick up continuous speech much better and with much more accuracy.
