3/17/2017 1 Comment Voice Recognition Software #8An interesting, and what will surely be controversial, article appeared recently about how Germany is going to begin using voice recognition for refugees seeking asylum in their country. On the surface, the problem seems to be a simple one, Germany like so many other European countries is dealing with an enormous influx of refugees from war torn nations, in particular Syria. And, the process to determine if in fact these refugees are who they say they are, is a difficult one at best. Germany has clearly become aware of the fact that not all refugees are honest about their point of origin, and that can determine if those people are eligible for asylum. The simplest and most efficient manner in which to deal with this is through a “dialect analysis” which is a speech software that will supposedly define the applicant’s country of origin and determine eligibility to enter Germany. The technology is based upon the use of “speech samples” that have been derived from a library of speech samples that have been collected over the past 20 years.
This process in Germany is actually nothing terribly new, Germany has employed speech linguists for many years who have been responsible for interviewing applicants and authenticating their speech in order to prove the application is genuine. The concept of automating this established process has many worried that both errors will occur, and that people will be discriminated because of incorrect or inaccurate information derived in this manner. It has also been widely reported that the main objective of this program is to detect Arabic speaking persons who may be claiming to be Syrian refugees in order to increase their chance of gaining asylum. Germany has been using a form of speech analysis since 1998, however that existing system is far from sophisticated and clearly needs to be either thoroughly updated, or discarded altogether. The existing system is actually monitored by individuals, and not by a software program as is the new system that will be put into place in the next few weeks. Can a system such as this one actually perform the task with absolute accuracy? For now, that question remains, and all linguists will tell you that language is constantly changing, in all cultures and countries. While languages all have a historical basis, we do know that they all are constantly evolving, just like culture is. To suggest that a person can be fully identified simply through a voice recognition software program is both dangerous, and impedes the basic human right of privacy. What if an individual has been raised as an orphan, and learned only the language spoken in the adopted household? Does that invalidate the actual identity, or does it truly identify who this person may or may not be? Another consideration may be people who have speech impediments, or have suffered injuries that result in changed speech patterns. I highly doubt this program can discern the differences with these types of situations, there are simply too many potential variables. What is clear is that nations are now paying extremely close attention as to how immigrants, refugees and foreign applicants are being examined and processed in order to gain asylum. What is not clear is just how truly effective the mechanisms, such as this type of software will be with accuracy in that process, and how it may affect the future of asylum seeking individuals.
1 Comment
Cammron Keehley
5/23/2017 08:46:15 pm
Hey Luigi,
Reply
Leave a Reply. |
Archives
May 2017
Categories
|