DoCoMo demonstrates spoken language translator for smartphones
(PhysOrg.com) -- Japanese cellular service company NTT DoCoMo, recently demonstrated a smartphone cloud based app that allows users speaking different languages to communicate with one another by translating their conversation into each others language. Using already existing technology from other companies, the service listens to words spoken on one end, coverts those words to text, then translates them to the other persons language, which it spits into another text file; it then uses text-to-speech software to read the results to the person on the other end of the line. When the person responds, the whole process works in reverse. In addition to speaking and hearing the finished result, users can also see the words in both languages on their cell phones as the conversation progresses.
Speaking about the as yet unnamed service at the Wireless Japan 2011 convention, DoCoMo president Ryuji Yamada said, We want a phone that interprets conversations just like in science fiction." Though clearly still not up to Star Trek standards, this precursor to the Universal Translator is definitely a step in the right direction.
In the video made at the demonstration, it is clear that smartphone users are able to communicate with one another (currently only English and Japanese) via the new service; whats not clear is how accurate the system is.
A company spokesman says that though the service is based on the best voice recognition system they could find, they admit its still not at one hundred percent, which means errors are introduced in the very first stage of the translation process. When you combine that with the errors that are almost certainly going to happen as text is converted from one language to another due to idiosyncrasies inherent in different languages (see Google translation) the results wont likely be knocking anyone over anytime soon. However, as new technology is developed for doing both, it can easily be introduced into the system (since its cloud based) with the end users none-the-wiser, except for the increased accuracy.
The company also points out that though the new service might not be for everyone, it likely will be of use to people who dont expect or need a very high level of accuracy in voice translation; friends for example or family speaking long distance, or even business partners who wish to convey very simple information. DoCoMo reps say the new service should be available for trial by the end of the year.
© 2010 PhysOrg.com