Such as, an algorithm may be fed a smaller sized amount of labeled speech data and afterwards trained over a much larger set of unlabeled speech data to be able to make a machine learning design capable of speech recognition.The saving of data is called Serialization, when restoring the data is referred to as Deserialization. Also, we manage unique