'The Babel Fish is small, yellow and leech-like, and probably the oddest thing in the Universe. It feeds on brainwave energy received not from its own carrier but from those around it. The practical upshot of all this is that if you stick a Babel Fish in your ear you can instantly understand anything said to you in any form of language. The speech patterns you actually hear decode the brainwave matrix which has been fed into your mind by your Babel Fish. Meanwhile, the poor Babel Fish, by effectively removing all barriers to communication between different races and cultures, has caused more and bloodier wars than anything else in the history of creation.' (From The Hitchhiker's Guide to the Galaxy. By Douglas Adams.)


I always here "The computers will bring the world closer", or make us a "global community".  Presently this will never happen, simply because of the tower of Babel.  Bible stories about the tower may not be historically accurate but the truth of the concept or consequence is.  Until we discover or create a Babel Fish, the world will never be a homogeneous community.  Tough issues like, religion, human rights and politics are hard enough to deal with out the confusion of a foreign language.   Although Douglas Adams may have been right in assuming that the fall of the language barrier would cause much stress and conflict, given time I believe the world would be a better place.
	 I had to ask the question, Is anyone working on a 'Babel fish', or is it even possible.  In trying to answer this question I surprised by the scope of people who were working on various parts of the problem. Although there are many companies who have produced a plethora of language translation programs, no one has created a comprehensive one.   I also found many people who are developing voice and more importantly speech recognition systems.  It seems to me that an electronic Babel fish would need four things, speech recognition ability, language translation software, speech synthesis ability and be packed into a small piece of hardware. In order for our Babel fish to work it must be able to understand many languages and translate them to a single language, then repeat, almost instantaneously, the correct translation. 
	Speech Recognition
I found very little information on speech recognition compared to the other parts of this problem.  What I did find was a new technology that is taking off quickly.  There are many groups working on this technology.  With applications ranging from computer input devices for the handicap to easier control of equipment for surgeons.
Speech Recognition is done by the computer first digitally sampling the audio input, then applying a acoustic algorithm to spectrally examine the sound data, this is called Cochlea modeling.  After the sound data has been analyzed its resulting pattern is examined for phonemes.  Phonemes are the building blocks for the sound of words.  Next the computer sets up a relationship with the vocabulary, there are many techniques to do this: Dynamic time wrapping, Hidden Markov modeling and neural networks.  Hidden Markov modeling is the most commonly used techniques.  HMM recognizes speech by estimating the likelihood of each phoneme at contiguous, small regions (frames) of the speech signal.  Each word in the vocabulary is specified by its phonemes.   HMM  compares the phoneme sequence of the spoken word with the most likely match in the vocabulary.  The results are the computers best guess as to what the word is.  Some systems are in the work that will allow the computer to take in to account punctuation and context to further increase the accuracy.  
	Most speech recognition systems currently have about a 90% accuracy.  Accuracy improves when limitations are placed on the situation like limited vocabulary.  Currently many researchers are working on improving the accuracy of their systems; this is the last stumbling block to real time speech recognition.  In the near future we should see speech recognition systems that can be used effectively by many different users.  These systems will also be able do this instantaneously.  Good, the first part of the Babel fish is completed.
	The second part of the Babel fish is by far the simplest, If we now what a word means in a foregn language, we can easily put that information into a Database.  There is software available commercially for many, many languages.  In the Babel fish, it would be necessary to have a large database, that would contain as many words and languages as possible.  Teamed with speech recognition, the computer could easily translate any language it had been programmed with.


Not complete, I will be adding the rest today