Handling Pronunciation Variation Using Hybrid Approach In Continuous Speaker Independent Speech Recognition For Amharic

The problem of modeling promllciation variation lies in accurately predicting the wordrnpronunciations that occur in the test material .In order to achieve this, the pronunciation variantsrnmust first be obtained in some way or other i.e. from pronunciation data or from pre-specifiedrnphonological rules based on linguistic knowledge.rnIn this study, first models was developed using canonical dictionary as a reference with totalrndata set of 950 sentences of which 700 for training and the remaining 250 are used for testingrnmodelrnTwo models were developed using knowledge based and data driven adding variants to thernrespective dictionaries. Lastly another model: hybrid approach was developed. This model isrnsupposed to avoid the short coming of the two models, knowledge based and data driven. In lightrnof this, the model developed using hybrid approach has shown better performance. The betterrnprogress obtained is due to the fact that in case of knowledge based approach, the variant addedrnmayor may not really appear in the text audio. in contrary, during data driven the variantsrnactually appear in audio but it is very difficult to listen and identify all words with variation.rnThus the experiment undertaken in this study revealed that combining the two approachesrn(hybrid) is the best way for handling pronunciation variation.

Get Full Work

Report copyright infringement or plagiarism