Improving Keyword Spotting with a Tandem BLSTM-DBN Architecture
We propose a novel architecture for keyword spotting which is composed of a Dynamic Bayesian Network (DBN) and a bidirectional Long Short-Term Memory (BLSTM) recurrent neural net. The DBN uses a hidden garbage variable as well as the concept of switching parents to discriminate between keywords and arbitrary speech. Contextual information is incorporated by a BLSTM network, providing a discrete phoneme prediction feature for the DBN. Together with continuous acoustic features, the discrete BLSTM output is processed by the DBN which detects keywords. Due to the flexible design of our Tandem BLSTM-DBN recognizer, new keywords can be added to the vocabulary without having to re-train the model. Further, our concept does not require the training of an explicit garbage model. Experiments on the TIMIT corpus show that incorporating a BLSTM network into the DBN architecture can increase true positive rates by up to 10%.
KeywordsKeyword Spotting Long Short-Term Memory Dynamic Bayesian Networks
Unable to display preview. Download preview PDF.
- 1.Bilmes, J. A.: Graphical models and automatic speech recognition. In: Mathematical Foundations of Speech and Language Processing (2003)Google Scholar
- 4.Graves, A., Fernandez, S., Schmidhuber, J.: Bidirectional LSTM networks for improved phoneme classification and recognition. In: Proc. of ICANN, Warsaw, Poland, vol. 18(5-6), pp. 602–610 (2005)Google Scholar
- 5.Graves, A.: Supervised sequence labelling with recurrent neural networks. Phd thesis, Technische Universität München (2008)Google Scholar
- 6.Hermansky, H., Ellis, D.P.W., Sharma, S.: Tandem connectionist feature extraction for conventional HMM systems. In: Proc. of ICASSP, Istanbul, Turkey, vol. 3, pp. 1635–1638 (2000)Google Scholar
- 8.Ketabdar, H., Bourlard, H.: Enhanced phone posteriors for improving speech recognition systems. IDIAP-RR, no. 39 (2008)Google Scholar
- 9.Rose, R.C., Paul, D.B.: A hidden markov model based keyword recognition system. In: Proc. of ICASSP, Albuquerque, NM, USA, vol. 1, pp. 129–132 (1990)Google Scholar
- 11.Wöllmer, M., Eyben, F., Reiter, S., Schuller, B., Cox, C., Douglas-Cowie, E., Cowie, R.: Abandoning Emotion Classes - Towards Continuous Emotion Recognition with Modelling of Long-Range Dependencies. In: Proc. of Interspeech, Brisbane, Australia, pp. 597–600 (2008)Google Scholar
- 12.Wöllmer, M., Eyben, F., Keshet, J., Graves, A., Schuller, B., Rigoll, G.: Robust discriminative keyword spotting for emotionally colored spontaneous speech using bidirectional LSTM networks. In: Proc. of ICASSP, Taipei, Taiwan (2009)Google Scholar
- 13.Wöllmer, M., Eyben, F., Schuller, B., Rigoll, G.: Robust vocabulary independent keyword spotting with graphical models. In: Proc. of ASRU, Merano, Italy (2009)Google Scholar