Handling Diphthong and Triphone Symbols: Useful in Automatic English Text Generation from Pitman Shorthand Language Document

Abstract

The paper describes the recognition of diphthong and triphone signs useful in automatic generation of English text from Pitman Shorthand Language (PSL) document. The PSL is used to note down the dictated/spoken text and is widely practiced in all organizations where English is the transaction medium. This has a built-in practical advantage because of which it is universally acknowledged. This recording medium will continue to exist in spite of considerable developments in Speech Processing Systems, which are not universally established yet. Because of wide usage of PSL and the effort in its automation, PSL processing has emerged as a potential problem for research in the areas of Pattern Recognition, Image Processing, Artificial Intelligence and Document Analysis. There are six long and six short vowels in PSL. These vowels are represented by signs/symbols that over ride a stroke symbol and require recognition for composing English text from phonetic text documented through PSL. The work pertaining to other constructs of PSL is already carried out at word level by the authors [7,8]. But during dictation, it is common that the vowels are joined to form one syllable, called Diphthong. The diphthong appended with a tick-mark is called Triphone and represents any vowel that immediately follows the diphthong in a stroke. The diphthongs are to be cognized and recognized to generate the correct English equivalent text. The present work comprises of the definition of diphthong primitives, creation of knowledge base and the development of an algorithm for their recognition. A suitable shape recognition algorithm is assumed available here. This work is new and this module serves as a prerequisite for the complete recognition of PSL document and generation of an equivalent and correct English text.