The FOPHO Speech Recognition Project

The FOPHO (F_oreign Phonet ic ian) speech recogn i t i on p ro jec t concerns the development of a system to produce a reasonably high q u a l i t y phonetic t r a n s c r i p t i o n output from continuous speech i npu t . The system is developed to perform in a way which approximates the act ions of a phonet ic ian t r y i n g to t ranscr ibe a fo re ign tongue, ( i n the case of FOPHO, Aus t ra l i an Eng l i sh ) . Because of th i s cen t ra l phi losophy, FOPHO is a very i n t e r a c t i v e system and has f a c i l i t i e s for automatic learn ing and analys is of i t s own performance. Good q u a l i t y recogn i t ion is achieved through algor i thms which are very context-dependent and which are sens i t i ve to a v a r i e t y of poss ib le productions of s i m i l a r sounds even though the system i t s e l f is speaker independent.