Abstract
Multi-Layer Perceptron (MLP) features have recently been attracting growing interest for automatic speech recognition due to their complementarity with cepstral features. In this paper the use of MLP features is evaluated in a large vocabulary continuous speech recognition task, exploring different types of MLP features and their combination. Cepstral features and three types of Bottle-Neck MLP features were first evaluated without and with unsupervised model adaptation using models with the same number of parameters. When used with MLLR adaption on a broadcast news Arabic transcription task, Bottle-Neck MLP features perform as well as or even slightly better than a standard 39 PLP based front-end. This paper also explores different combination schemes (feature concatenations, cross adaptation, and hypothesis combination). Extending the feature vector by combining various feature sets led to a 9% relative word error rate reduction relative to the PLP baseline. Significant gains are also reported with both ROVER hypothesis combination and cross-model adaptation. Feature concatenation appears to be the most efficient combination method, providing the best gain with the lowest decoding cost.
This work was in parts supported under the GALE program of the Defense Advanced Research Projects Agency, Contract No. HR0011-06-C-0022 an in parts by OSEO under the Quaero program.
This is a preview of subscription content, log in via an institution to check access.
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Zhu, Q., Stolcke, A., Chen, B.Y., Morgan, N.: Using MLP features in SRI’s conversational speech recognition system. In: INTERSPEECH 2005, pp. 2141–2144 (2005)
Hermansky, H., Ellis, D., Sharma, S.: TANDEM connectionist feature extraction for conventional HMM systems. In: ICASSP 2000, Istanbul, Turkey (2000)
Hermansky, H., Sharma, S.: TRAPs - classifiers of TempoRAl Patterns. In: ICSLP 1998 (November 1998)
Grézl, F., Karafiát, M., Kontár, S., Černocký, J.: Probabilistic and bottle-neck features for LVCSR of meetings. In: ICASSP 2007, April 2007, pp. 757–760. IEEE Signal Processing Society, Hononulu (2007)
Grézl, F., Fousek, P.: Optimizing bottle-neck features for LVCSR. In: ICASSP 2008, Las Vegas, ND (2008)
Gauvain, J., Lamel, L., Adda, G.: The LIMSI Broadcast News Transcription System. Speech Communication 37(1-2), 89–108 (2002)
Lamel, L., Messaoudi, A., J.L.G.: Improved Acoustic Modeling for Transcribing Arabic Broadcast Data. In: Interspeech 2007, Antwerp, Belgium (2007)
Leggetter, C., Woodland, P.: Maximum likelihood linear regression for speaker adaptation of continuous density hidden Markov models. Computer Speech and Language 9(2), 171–185 (1995)
Fousek, P.: Extraction of Features for Automatic Recognition of Speech Based on Spectral Dynamics. PhD thesis, Czech Technical University in Prague, Faculty of Electrical Engineering, Prague (March 2007)
Athineos, M., Hermansky, H., Ellis, D.P.: LP-TRAP: Linear predictive temporal patterns. In: ICSLP 2004 (2004)
Fiscus, J.: A Post-Processing System to Yield Reduced Word Error Rates: Recogniser Output Voting Error Reduction (ROVER) (1997)
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 2008 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Fousek, P., Lamel, L., Gauvain, JL. (2008). On the Use of MLP Features for Broadcast News Transcription. In: Sojka, P., Horák, A., Kopeček, I., Pala, K. (eds) Text, Speech and Dialogue. TSD 2008. Lecture Notes in Computer Science(), vol 5246. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-87391-4_39
Download citation
DOI: https://doi.org/10.1007/978-3-540-87391-4_39
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-87390-7
Online ISBN: 978-3-540-87391-4
eBook Packages: Computer ScienceComputer Science (R0)