Speech recognition is emerging as a contemporary research topic wherein many studies are conducted to optimize the efficiency of speech recognition and thereafter, suggest relevant documents, texts or any other source of information required by the users. With many real world applications, clustering can be efficiently used for the purpose of speech recognition and recommendation system. Many theories and algorithms have been proposed so far with n gram approach, DNN based system etc. The paper will dwell deeper into an efficient algorithm termed as spontaneous clustering in which clustering of documents is done simultaneously as speech data is analysed and recorded. Speech segments are clustered on-the-fly and thereafter, Golay coding will be applied to finally cluster the speech data to recommend relevant documents. This methodology tends to give better results than the existing methodologies as it reduces the error rate of the recommendations and clustering efficiency is refined. Clustering has advantages in plethora of fields and the field of speech recognition is highly benefited to recommend relevant documents. This methodology shall be cutting edge for the future of speech recognition with minimal error rate and better results of recommendation.
Prof. Dr. Bilal BİLGİN