Enhancing HMM Based Malayalam Continuous Speech Recognizer Using Artificial Neural Networks

Conference paper
Part of the Smart Innovation, Systems and Technologies book series (SIST, volume 32)


Improving discrimination in recognition systems is a subject of research in recent years. Neural network classifiers are naturally discriminative and can be easily applied to real-world problems. This paper examines the use of multilayer perceptrons as the emission probability estimator in a hidden Markov model based continuous speech recognizer for Malayalam language. The performance of the system has been compared with a recognizer using Gaussian mixture model as the emission probability estimator. Experimental results show that the proposed neural network based acoustic scoring yields significant gains in recognition accuracy and system compactness.


Continuous speech recognition Malayalam speech recognition Hidden Markov model Gaussian mixture model Artificial neural network 


  1. 1.
    Anuj, M., Nair, K.N.R.: Continuous Malayalam speech recognition using hidden Markov models. In: Proceedings of 1st Amrita-ACM-W Celebration on Women in Computing in India (2010)Google Scholar
  2. 2.
    Anuj, M., Nair, K.N.R.: HMM/ANN hybrid model for continuous Malayalam speech recognition. Procedia Eng. 30, 616–622 (2012)CrossRefGoogle Scholar
  3. 3.
    Kurian, C.: Analysis of unique phonemes and development of automatic speech recognizer for malayalam language. Ph.D. Thesis, Cochin University of Science and Technology, Kerala (2013)Google Scholar
  4. 4.
    Yegnanarayana, B., Kishore, S.P.: AANN: an alternative to GMM for pattern recognition. Neural Networks 15, 459–469 (2002)CrossRefGoogle Scholar
  5. 5.
    Richard, M.D., Lippmann, R.P.: Neural network classifiers estimate Bayesian a posteriori probabilities. Neural Comput. 3, 461–483 (1991)CrossRefGoogle Scholar
  6. 6.
    Bourlard, H., Morgan, N.: Continuous speech recognition by connectionist statistical methods. IEEE Trans. Neural Networks 4, 893–909 (1993)CrossRefGoogle Scholar
  7. 7.
    Seid, H., Gamback, B.: A speaker independent continuous speech recognizer for Amharic. In: INTERSPEECH 2005, 9th European Conference on Speech Communication and Technology, Lisbon (2005)Google Scholar
  8. 8.
    Meinnedo, H., Neto, J.P.: Combination of acoustic models in continuous speech recognition hybrid systems. In: Proceedings of ICSLP, Beijing (2000)Google Scholar
  9. 9.
    Bakis, R.: Continuous speech recognition via centiseconds acoustic states. J. Acoust. Soc. Am. 59(S1) (1976) Google Scholar
  10. 10.
    Baum, L.E.: An inequality and associated maximization technique in statistical estimation for probabilistic functions of Markov processes. Inequalities 3, 1–8 (1972)Google Scholar
  11. 11.
    Rumelhart, D.E., Hinton, G.E., Williams, R.J.: Learning representations by back-propagating errors. Nature 323, 533–536 (1986)CrossRefGoogle Scholar
  12. 12.
    Viterbi, A.J.: Error bounds for convolutional codes and asymptotically optimum decoding algorithms. IEEE Trans. Inf. Theory 13, 260–269 (1982)CrossRefGoogle Scholar

Copyright information

© Springer India 2015

Authors and Affiliations

  1. 1.School of Computer SciencesMahatma Gandhi UniversityKottayamIndia
  2. 2.Department of Computer Science and EngineeringViswajyothi College of Engineering and TechnologyVazhakulam, MuvattupuzhaIndia

Personalised recommendations