Use of temporal correlation between successive frames in a hidden Markov model based speech recognizer
Kuldip K. Paliwal · IEEE International Conference on Acoustics Speech and Signal Processing · 1993
The temporal correlation between successive frames is incorporated in an HMM (hidden Markov model) based speech recognizer. This is done by making the probability of the current observation vector dependent on the previous observation vectors. Preliminary results show that this approach provides significant improvement in recognition performance (with the use of temporal correlation between two successive frames alone).>