Im Sommersemester 2003 findet am Lehrstuhl für Informatik VI ein

Seminar "Speech Recognition and Language Processing"

statt.

Ablauf und Termine:

Das Seminar findet als Blockveranstaltung an folgenden Terminen, jeweils im Seminarraum des Lehrstuhls für Informatik VI (Raum 6124) statt:

    Mittwoch,     30.7.2003, 13:30 - 18:00
    Donnerstag,   31.7.2003,  8:30 - 17:00

Agenda des Seminars im einzelnen (die Zeiten können im genauen Ablauf geringfügig variieren):

      Mittwoch,    30.7.2003

            13:30    "Multi-Pass Search" Arne Theres
            15:00    "Speaker Adaptation using Eigenvoices" Daniel Schneider
            16:30    "Histogram Normalization" Haihua Luo 

      Donnerstag,  31.7.2003

                 8:30  "Discriminative Training" Lars Haferkamp
            10:00    "Confidence Measures in Speech Recognition" Daniel Peger
            11:30    "Word Error Rate Minimization" Thorsten Palm

              13:00    MITTAGSPAUSE

            14:00    "Natural Language Database Queries using Classification and Regression Trees" Samuel Senyo Okae
            15:30    "Language Modelling and Word Morphology using Maximum Entropy" Paul Tawiah

• Gliederungen:   Abgabe bis spätestens 6 Wochen vor dem Probevortragstermin im Sekretariat des Lehrstuhls Informatik VI oder bei dem Betreuer/der Betreuerin.
• Ausarbeitungen: Abgabe bis spätestens 1 Monat vor dem Probevortragstermin im Sekretariat des Lehrstuhls Informatik VI  oder bei dem Betreuer/der Betreuerin.
• Vortragsfolien: Abgabe bis spätestens 1 Woche vor dem Probevortragstermin im Sekretariat des Lehrstuhls Informatik VI  oder bei dem Betreuer/der Betreuerin.
• Probevorträge: siehe Themen, mindestens 2 Wochen vor dem Vortragstermin.
• Seminarvorträge: Der Vortragsblock findet vermutlich Ende Juli/Anfang August statt. Die genaue Einteilung wird unter den einzelnen  Themen bekanntgegeben.
• Endgültige (ggfls. korrigierte) Ausarbeitungen und Vortragsfolien: Abgabe bis spätestens 2 Wochen nach dem Vortragstermin im Sekretariat des Lehrstuhls Informatik VI oder bei dem Betreuer/der Betreuerin.
• Anwesenheitspflicht: Voraussetzung für die Vergabe eines Leistungsnachweises ist die Anwesenheit aller Seminarteilnehmer und -teilnehmerinnen zu allen Vortragsterminen.

Vortragsthemen, jeweilige Literatur und Teilnehmer:

1. A^* Search (Betreuer:Oliver Bender)

• F. Jelinek, Statistical Methods for Speech Recognition, MIT Press, Cambridge, MA, 1997, chapter 6.
• F. Jelinek, "A Fast Sequential Decoding Algorithm using a Stack", IBM Journal of Research Development, Vol. 13, 675-685, November 1969.
• D. B. Paul, "An Essential A^* Stack Decoder Algorithm for Continuous Speech Recognition with a Stochastic Language Model," Proc. Int. Conf. on Acoustics, Speech, and Signal Processing, pp. 25-28, San Francisco, CA, March 1992.

2. Multi-Pass Search (Vortragender: Arne Theres, Betreuer: Stephan Kanthak)

• R. Schwartz, L. Nguyen, J. Makhoul: "Multiple-Pass Search Strategies," in Automatic Speech and Speaker Recognition, C.-H. Lee, F. K. Soong, K. K. Paliwal (eds.), pp. 429-456, Kluwer Academic Publishers, Norwell, MA, 1996.
• X. Huang, A. Acero, F. Alleva, M. Hwang, L. Jiang, M. Mahajan: "From SPHINX-II to Whisper - Making Speech Recognition Usable," in Automatic Speech and Speaker Recognition, C.-H. Lee, F. K. Soong, K. K. Paliwal (eds.), pp. 481-508, Kluwer Academic Publishers, Norwell, MA, 1996.
• P.S. Gopalakrishnan, L.R. Bahl: "Fast Match Techniques," in Automatic Speech and Speaker Recognition, C.-H. Lee, F. K. Soong, K. K. Paliwal (eds.), pp. 413-428, Kluwer Academic Publishers, Norwell, MA, 1996.
• L.R. Bahl, S.V. De Gennaro, P.S. Gopalakrishnan, R.L. Mercer: "A Fast Approximate Acoustic Match for Large Vocabulary Speech Recognition," IEEE Trans. on Speech and Audio Processing, Vol. 1, pp. 59-67, January 1993.
• P.S. Gopalakrishnan, L.R. Bahl, R.L. Mercer: "A Tree Search Strategy for Large Vocabulary Speech Recognition," Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vol. 1, pp. 572-575, Detroit, MI, May 1995.

3. Transducer-based Search (Betreuer:Stephan Kanthak)

• G. Antoniol, F. Brugnara, M. Cettolo, M. Federico, "Language Model Representations for Beam-Search Decoding", Proc. of the Int. Conf. on Acoustic, Speech and Signal Processing (ICASSP'95), 588-591, Detroit, MI, May 1995.
• F. Brugnara, M. Cettolo, "Improvements in Tree-Based Language Model Representation", Proc. Europ. Conf. on Speech Communication and Technology, 1797-1800, Madrid, September 1995.
• K. Demuynck, J. Duchateau, D. V. Compernolle, P. Wambacq, "An Efficient Search Space Representation for Large Vocabulary Continuous Speech Recognition", Speech Communications, Vol. 30, 37-53, January 2000.
• M. Mohri, F. Pereira, M. Riley, "Weighted Finite-State Transducers in Speech Recognition", Proc. of the ISCA ITRW Automatic Speech Recognition: Challenges for the new Millenium (ASR'2000), 97-106, Paris, France, August 2000.
• F. Pereira, M. Riley, "Speech Recognition by Composition of Weighted Finite Automata", 1997.

4. Speaker Adaptation using MLLR (Betreuer:Michael Pitz)

• D. Giuliani, R. DeMori: Spoken Dialogs with Computers, Chapter 11, Academic Press, 1997.
• P. Woodland: "Speaker Adaptation for Continuous Density HMMs: A Review", Proc. ISCA Workshop on Adaptation Methods for Speech Recognition, pp. 11-19, Sofia Antinopolis, France, 2001.
• C.J. Leggetter, P.C. Woodland: "Maximum Likelihood Linear Regression for Speaker Adaptation of Continuous Density Hidden Markov Models", Computer Speech and Language, Vol. 9, pp. 171-185, 1995.
• M. J. F. Gales: "Maximum Likelihood Linear Transformations for HMM-Based Speech Recognition", Computer Speech and Language, Vol. 12, pp. 75-98, 1998.
• T. Anastaskos, J. McDonough, R. Schwartz, J. Makhoul: "A Compact Model for Speaker Adaptive Training", Proc. of the Int. Conf. on Spoken Language Processing (ICSLP'96), pp. 1137-1140, Philadelphia, PA, October 1996.

5. Speaker Adaptation using Eigenvoices (Vortragender: Daniel Schneider, Betreuer: Michael Pitz)

• R. Kuhn, J.C. Junqua, P. Nguyen, N. Niedzielski: "Rapid Speaker Adaptation in Eigenvoice Space," IEEE Transactions on Speech and Audio Processing, Vol. 8, No. 6, pp. 695-707, 2000.
• H. Botterweck: "Very Fast Adaptation for Large Vocabulary Speech Recognition Using Eigenvoices," Proc. Int. Conf. on Spoken Language Processing, Vol. IV, pp. 354-357, Bejing, October 2000.
• M.J.F. Gales: "Cluster Adaptive Training of Hidden Markov Models," IEEE Transactions on Signal and Audio Processing, Vol. 8, No. 4, pp. 417-420, 2000.
• P. Woodland: "Speaker Adaptation for Continuous Density HMMs: A Review", Proc. ISCA Workshop on Adaptation Methods for Speech Recognition, pp. 11-19, Sofia Antinopolis, France, 2001.

6. Vocal Tract Length Normalization (Betreuer: Florian Hilger)

• A. Acero, R. M. Stern: "Robust Speech Recognition by Normalization of the Acoustic Space," Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vol. II, pp. 893-896, Toronto, Canada, May 1991.
• E. Eide, H. Gish: "A Parametric Approach to Vocal Tract Length Normalization," Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vol. I, pp. 346-349, Atlanta, GA, May 1996.
• S. Wegmann, D. McAllaster, J. Orloff, B. Peskin: "Speaker Normalization on Conversational Telephone Speech," Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vol. I, pp. 339-342, Atlanta, GA, May 1996.
• L. Lee, R. C. Rose, "A frequency warping approach to speaker normalization," IEEE Trans. Speech and Audio Processing, vol. 6, no. 1, pp. 49-60, Jan. 1998.
• D. Pye, P. C. Woodland: "Experiments in Speaker Normalization and Adaptation for Large Vocabulary Speech Recognition," Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vol. II, pp. 1047-1050, Munich, Germany, April 1997.

7. Histogram Normalization (Vortragender: Haihua Luo, Betreuer: Florian Hilger)

• S. Dharanipragada, M. Padmanabhan: "A Nonlinear Unsupervised Adaptation Technique for Speech Recognition," Proc. Int. Conf. on Spoken Language Processing, pp. 556-559, Bejing, China, October 2000.
• M. Padmanabhan, S. Dharanipragada: "Maximum Likelihood Non-linear Transformation for Environtment Adaptation in Speech Recognition Systems," Proc. European Conf. on Speech Communication and Technology, pp. 2359-2362, Aalborg, Denmark, September 2001.
• J. C. Segura, M. C. Benitez, A. de la Torre, A. J. Rubio: "Feature extraction combining spectral subtraction and cepstral histogram equalization," Proc. Int. Conf. on Spoken Language Processing, vol. 1, pp. 225-228, Denver, CO, USA, September 2002.
• F. Hilger, S. Molau, H. Ney: "Quantile Based Histogram Equalization For Online Applications," Proc. Int. Conf. on Spoken Language Processing, vol. 1, pp. 237-240, Denver, CO, USA, September 2002.
• S. Molau, F. Hilger, D. Keysers, H. Ney: "Enhanced Histogram Normalization in the Acoustic Feature Space," Proc. Int. Conf. on Spoken Language Processing, Vol. 1, pp. 1421-1424, Denver, CO, USA, September 2002.

8. Discriminative Training (Vortragender: Lars Haferkamp, Betreuer: Wolfgang Macherey)

• Y. Normandin, "Maximum Mutual Information Estimation of Hidden Markov Models," in Automatic Speech and Speaker Recognition, C.-H. Lee, F. K. Soong, K. K. Paliwal (eds.), pp. 57-81, Kluwer Academic Publishers, Norwell, MA, 1996.
• V. Valtchev, J. J. Odell, P. C. Woodland, S. J. Young: "MMIE Training of Large Vocabulary Recognition Systems," Speech Communication, Vol. 22, No. 4, pp. 303-314, September 1997.
• P.C. Woodland, D. Povey: "Large Scale Discriminative Training of Hidden Markov Models for Speech Recognition," Computer Speech and Language, Vol. 16, No. 1, pp. 25-47, January 2002.

9. Discriminative Model Combination (Betreuer: Wolfgang Macherey)

• P. Beyerlein: Diskriminative Modellkombination in Spracherkennungssystemen mit großem Wortschatz, Dissertation, RWTH Aachen, Germany, October 2000.

10. Confidence Measures in Speech Recognition (Vortragender: Daniel Peger, Betreuer: Ralf Schlüter)

• F. Wessel, R. Schlüter, K. Macherey, H. Ney: "Confidence Measures for Large Vocabulary Continuous Speech Recognition," Proc. IEEE Trans. on Speech and Audio Processing, Vol. 9, No. 3, pp. 288-298, March 2001.
• F. Wessel: "Word Posterior Probabilities for Large Vocabulary Continuous Speech Recognition," Dissertation, Aachen, Germany, January 2002.

11. Word Error Rate Minimization (Vortragender: Torsten Palm, Betreuer:Ralf Schlüter)

• F. Wessel: "Word Posterior Probabilities for Large Vocabulary Continuous Speech Recognition," Dissertation, Aachen, Germany, January 2002.
• A. Stolcke, Y. König, M. Weintraub: "Explicit Word Error Rate Minimization in N-best List Rescoring," Proc. Europ. Conf. on Speech Communication and Technology, Vol. 2, pp. 163-166, Rhodes, Greece, September 1997.
• G. Evermann, P. C. Woodland: "Large Vocabulary Decoding and Confidence Estimation Using Word Posterior Probabilities," Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vol. 3, pp. 1655-1658, Istanbul, Turkey, June 2000.
• L. Mangu, E. Brill, A. Stolcke: "Finding Consensus Among Words: Lattice-Based Word Error Minimization," Proc. Europ. Conf. on Speech Communication and Technology, Vol. 1, pp. 495-498, Budapest, Hungary, September 1999.
• V. Goel, W. Byrne: "Minimum Bayes Risk Automatic Speech Recognition," Computer Speech and Language, Vol. 14, pp. 115-135, 2000.

12. Natural Language Database Queries using Classification and Regression Trees (Vortragender: Samuel Senyo Okae, Betreuer: Klaus Macherey)

• R. Kuhn, R. De Mori: "The Application of Semantic Classification Trees to Natural Language Understanding," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 17, No. 7, pp. 449-460, 1995.
• R. Kuhn, R. De Mori: "Sentence Interpretation," in Spoken Dialogues with Computers, R. De Mori (ed.), Chapter 14, pp. 485-522, Academic Press, San Diego, CA, 1998.
• R.O. Duda, P.E. Hart, D.G. Stork: Pattern Classification, John Wiley and Sons, New York, NY, 2001, pp. 395-411.

13. Error Handling in Spoken Dialogue Systems (Betreuer: Klaus Macherey)

• M. Swerts, D. Litman, J. Hirschberg: "Corrections in Spoken Dialogue Systems", Proc. Int. Conf. Spoken Language Processing (ICSLP 2000), Vol. II, pp. 615-618, Beijing, October 2000.
• K. Komatani, T. Kawahara: "Generating Effective Confirmation and Guidance Using Two-Level Confidence Measures for Dialogue Systems", Proc. Int. Conf. Spoken Language Processing (ICSLP 2000), Vol. II, pp. 648-651, Beijing, October 2000.
• E. Krahmer, M. Swerts, M. Theune, M. Weegels: "Error Detection in Spoken Human-Machine Interaction," International Journal of Speech Technology, Vol. 4, No. 1, pp. 19-30, 2001.
• D. J. Litman, M. A. Walker, M. S. Kearns: "Automatic Detection of Poor Speech Recognition at the Dialogue Level", Proc. 37th Annual Meeting of the Association for Computational Linguistics (ACL 1999), pp. 309-316, 1999.

14. Language-Independent Named Entity Recognition (Betreuer:Oliver Bender)

• E.F. Tjong Kim Sang: "Introduction to the CoNLL-2002 Shared Task: Language-Independent Named Entity Recognition." Proc. Computational Natural Language Learning Workshop, Taipei, Taiwan, 2002, pp. 155-158.
• X. Carreras, L. Márques, Lluís Padró, "Named Entity Extraction using AdaBoost," Proc. Computational Natural Language Learning Workshop, Taipei, Taiwan, 2002, pp. 167-170.
• R. Florian: "Named Entity Recognition as a House of Cards: Classifier Stacking." Proc. Computational Natural Language Learning Workshop, Taipei, Taiwan, 2002, pp. 175-178.

15. Language Modelling and Word Morphology using Maximum Entropy (Vortragender: Paul Tawiah, Betreuerin: Nicola Ueffing)

• A.L. Berger, S. Della Pietra, V. Della Pietra: "A Maximum Entropy Approach to Natural Language Processing", Computational Linguistics, Vol. 22, No. 1, pp. 39-71, March 1996.
• R. Rosenfeld: "A Maximum Entropy Approach to Adaptive Statistical Language Modelling", Computer Speech and Language, Vol. 10, No. 3, pp. 187-228, July 1996.
• S.C. Martin, H. Ney, C. Hamacher. "Maximum Entropy Language Modeling and the Smoothing Problem", IEEE Trans. on Speech and Language Processing, Vol. 8, No. 5, pp. 626-632, September 2000.

16. Parsing Algorithms for Grammar-based Language Models (Betreuer: Max Bisani)

• B. Roark: "Probabilistic top-down parsing and language modelling", Computational Linguistics, Vol. 27, No. 2, pp. 249-276, 2001.
• B. Roark: "Markov parsing: lattice rescoring with a statistical parser", Proc. 40th Annual Meeting of the Association for Computational Linguistics (ACL 2002), pp. 287-294, Philadelphia, PA, USA, 2002.
• E. Charniak: "Immediate-Head Parsing for Language Models" Proc. 39th Annual Meeting of the Association for Computational Linguistics (ACL 2001), pp. 116-123, Toulouse, France, July 2001.
• D.H. Van Uytsel, F. Van Aelten, D. Van Compernolle: "A Structured Language Model based on Context-Sensitive Probabilistic Left-Corner Parsing" Proc. Meeting of the North-American Chapter of the Association of Computaional Linguists, pp. 223-230, Pittsburgh, PA, USA, June 2001.

17. Long-range Language Models: Distant Bigrams, Link Grammars and Word Triggers (Betreuer:Max Bisani)

• M. Simons, H. Ney, S. C. Martin: "Distant Bigram Language Modelling Using Maximum Entropy," Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vol. 2, pp. 787-790, Munich, Germany, April 1997.
• S. Della Pietra, V. Della Pietra, J. Gillet, J. Lafferty, H. Prinz, L. Ures: "Inference and Estimation of a Long-Range Trigram Model," in Grammatical Inference and Applications, Second International Colloquium, Lecture notes in Artificial Intelligence, No. 862, Springer Verlag, Berling, 1994.
• R. Rosenfeld: "A whole sentence maximum entropy language model," in Proc. IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU), pp., Santa Barbara, CA, December 1997.

18. Parsing in Language Modelling (Betreuerin:Nicola Ueffing)

• E. Charniak: "Immediate-Head Parsing for Language Models", Proc. 39th Annual Meeting of the Association for Computational Linguistics (ACL 2001), pp. 116-123, Toulouse, France, July 2001.
• C. Chelba, F.Jelinek: "Exploiting Syntactic Structure for Language Modeling", Proc. 36th Annual Meeting of the Association for Computational Linguistics (ACL 2001), pp. 225-231, Montréal, Canada, August 1998.
• B. Roark: "Probabilistic top-down parsing and language modelling", Computational Linguistics, Vol. 27, No. 2, pp. 249-276, 2001.
• B. Roark, Robust Probabilistic Predicitve Syntactic Processing: Motivations, Models, and Applications, PhD thesis, Brown University, Providence, RI, May 2001.

Informationen zur Ausarbeitung und zum Vortrag:

Ich empfehle, die ca. 20-seitige Ausarbeitung als auch die Folien für den Seminarvortrag (45 Minuten reine Redezeit + 15 Minuten Diskussion) in LaTeX zu erstellen. Weiter unten finden sich Dokumentvorlagen für die Ausarbeitung und den Vortrag sowie mehrere  LaTeX Dokumentationen, die im WWW verfügbar sind. In jedem Fall sollen die Folien und die Ausarbeitung im pdf-Format elektronisch eingereicht werden.

• Online LaTeX-Dokumentationen:

• LaTeX: "A document preparation system"
• LaTeX help 1.1
• LaTeX 2e for authors

• Dokumentvorlagen:

• Vorlage für die Ausarbeitung, beinhaltet die Vorlage und alle benötigten Dateien im tar Format.
• Vorlage für den Vortrag, beinhaltet die Vorlage, alle benötigten Grafiken und unsere Lehrstuhlstilvorlage, ebenfalls im tar Format.

• Einige Regeln für Folien und Ausarbeitung:

• Beachten Sie Bezüge zu anderen Themen im Seminar und kommunizieren Sie untereinander! Z.B. finden Methoden der transformations-basierten Kompression Anwendung in der Bildkompression.
• Es wird erwartet, dass Sie sich weitere Literatur zu Ihrem Thema eigenständig besorgen. Fragen zur Literaturrecherche werden Ihnen in der Bibliothek der Fachgruppe Informatik gern beantwortet. Die Möglichkeit zur Recherche besteht natürlich auch in der Lehrstuhlbibliothek am Lehrstuhl für Informatik VI.
• Tabellen haben immer eine Überschrift.
• Grafiken haben immer eine Unterschrift.
• Falls Sie keine adäquate Übersetzung für englische Fachausdrücke finden, benutzen Sie diese unverändert.
• Ausarbeitungen und Folien können auch in Englisch angefertigt werden (gute Übung!).
  
• Zitieren Sie alle von Ihnen verwendete Literatur.
• Die Form der Zitate soll wie in der Vorlage für die Ausarbeitung vorgegeben aussehen.
• Verwenden Sie Beispiele, um das Gesagte anschaulich zu erläutern.
• Beispiele sollten so komplex wie nötig und so einfach wie möglich sein.
• Ihre Folien sollen Sie als Vortragenden nicht ersetzen, sondern:
• wesentliche Zusammenhänge aufzeigen;
• eine Gedächtnisstütze für den Zuhörer (und für Sie als Vortragenden) sein;
• dem Zuhörer die Orientierung in Ihrem Vortrag erleichtern;
• Keine ausformulierten Sätze, sondern statt dessen prägnante Stichworte enthalten;
• Illustrationen einsetzen, wo immer Sie sinnvoll sind - ein Bild kann tausend Worte ersetzen!
• Abkürzungen bei erster Nennung in der folgenden Form definieren: z.B. "[...] an der Rheinisch-Westfälischen Technischen Hochschule (RWTH) gibt es [...]"
• Prüfen Sie, dass Sie in Ihrem Thema bleiben! Dazu sollten Sie sich auch der Bezüge zu den anderen Themen im Seminar bewusst sein! Ggfls. auch Querverweise auf andere Vorträge/Ausarbeitungen dieses Seminars vornehmen.

Rückfragen in Bezug auf alle organisatorischen Punkte bitte an:

Raum 6125b
Telefon: 0241 / 80 21 612
E-Mail: schlueter@cs.rwth-aachen.de