Different Evaluation Approaches of Confusion Network in Chinese Spoken Classification

Abstract:

As kind of multiple results in speech recognition system, lattice and confusion network are widely applied in spoken document classification and spoken document retrieval. In this paper, the relation between lattice and confusion network is analyzed firstly. Then based on the confusion network by clustering algorithm, different evaluation approaches are applied here to test the performance of confusion network, which include accuracy, complexity and distortion of confusion network. For the experiments on speech recognition system, it can be drawn that compared with 1-best result, confusion network can have higher words accuracy.

• Periodical: Advanced Materials Research (Volume 142)
• Pages: 174-179
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.142.174
• Online since: October 2010
• Authors: Lei Zhang, Jing Xin Chang, Xue Zhi Xiang
• Keywords: Accuracy, Complexity, Confusion Network, Distortion, Lattice, Spoken Document Classification

References

[1] Mangu L, Brill E, Stolcke A: Computer Speech and Language Vol. 14 (2000) No. 4, pp.373-400.

[2] Zhang L, Chang J X, Xiang X Z: Topic Indexing of Spoken Documents Based on Optimized N-Best Approach, Proceedings of ICIS, China (2009).

DOI: 10.1109/icicisys.2009.5357691

[3] Xue J, Zhao Y X: Improved confusion network algorithm and shortest path search from word lattice, Proceedings of ICASSP, Philadelphia, PA 2005, pp.353-356.

DOI: 10.1109/icassp.2005.1415248

[4] Tur G., Wright J, and Gorin A, et al: Improving spoken language understanding using word confusion networks, Proceedings of ICSLP, Denver, Colorado, 2002, pp.1137-1140.

DOI: 10.21437/icslp.2002-374

[5] Xue J and Zhao Y X: Random forests-based confidence annotation using novel feature from confusion network, Proceedings of ICASSPJ, Toulouse, France 2006, pp.1149-1152.

DOI: 10.1109/icassp.2006.1660229

[6] Bertoldi N and Federico M: A new decoder for spoken language translation based on confusion networks, IEEE ASRU Workshop, Cancun, Mexico, 2005, pp.134-140.

DOI: 10.1109/asru.2005.1566492

[7] Hillard D and Ostendorf M: Compensation forward posterior estimation bias in confusion networks, Proceedings of ICASSP Toulouse, France, 2006, pp.1153-1156.

[8] Tur G, Hakkani D and Riccardi G: Extending boosting for call classification using word confusion networks, Proceedings of IEEE International conference on acoustic, Speech, and signal Processing, 2004, pp.1520-6149.

DOI: 10.1109/icassp.2004.1326016

[9] Stolcke A, Konig Y, Weintraub M, Explicit word error minimization in N-best list rescoring, Proceedings of EUROSPEECH, 1997, pp.163-166.

DOI: 10.21437/eurospeech.1997-68

[10] Wu B, Liu G, Guo J: Research on confusion network algorithm for mandarin large vocabulary continuous speech recognition, Proceedings of ICWAPR, China, 2007: 145-155.

DOI: 10.1109/icwapr.2007.4421593

[11] Wang H L, Han J Q, Zheng T R: Quality evaluation and optimization of confusion network, Proceedings of ICMIP, (2005).