Modified-prior i-vector estimation for language identification of short duration utterances

In this paper, we address the problem of Language Identification (LID) on short duration segments. Current state-of-the-art LID systems typically employ total variability i-Vector modeling for obtaining fixed length representation of utterances. However, when the utterances are short, only a small amount of data is available, and the estimated i-Vector representation will consequently exhibit significant variability, making the identification problem challenging. In this paper, we propose novel techniques to modify the standard normal prior distribution of the i-Vectors, to obtain a more discriminative i-Vector extraction given the small amount of available utterance data. Improved performance was observed by using the proposed i-Vector estimation techniques on short segments of the DARPA RATS corpora, with lengths as small as 3 seconds.