GSLT: Speech Technology 1

Term paper, Fall semester 2002

During the course each student should prepare a term paper and present it at the closing seminar.

Examples of topics are listed below. It is possible to choose another topic after a discussion with one of the teachers in the Speech Technology course.

The term paper should be reviewed by two reviewers also participating in the course.

The paper may not exceed 10 printed pages.

Time table

November 1	Topic selection to Rolf Carlson rolf@speech.kth.se
December 12	Electronic submission of paper to two reviewers and the teacher
January 9	Electronic submission of reviews to author
January 26	Electronic submission of final paper to the reviewers, the teacher and Rolf Carlson rolf@speech.kth.se
February 6-7	Presentation of paper at the closing seminar at KTH

Possible topics

Acoustic Phonetics
David House davidh@speech.kth.se http://www.speech.kth.se/~davidh

• Phonetic cues for phrase boundaries and discourse finality
• Phonetic variation and speaking styles

Speech Synthesis
Björn Granström bjorn@speech.kth.se http://www.speech.kth.se/~bjorn

• Data-driven methods in speech synthesis
• Comparison of different speech synthesis methods.

Speech Recognition
Mats Blomberg mats@speech.kth.se http://www.speech.kth.se/~mats

• Why is speech recognition difficult?
• Evaluation of speech recognition systems

Speaker Verification
Mats Blomberg mats@speech.kth.se http://www.speech.kth.se/~mats

• Similarities and differences between speech recognition and speaker verification
• Text dependent and text independent speaker verification systems. Technology and applications.

Dialog Systems
Rolf Carlson rolf@speech.kth.se http://www.speech.kth.se/~rolf

• The need for robustness in dialog systems
• Use of multimodality in spoken dialog systems

Term Papers

Author	Subject Preliminary Title	2 Reviewers	Adviser
Markus Forsberg	Why is speech recognition difficult	Rebecca Jonson Kenneth Willhelmsson	Mats Blomberg
Rebecca Jonson	How can a dialogue system compensate for speech recognition deficiencies?	Gabriel Skantze Per Weijnitz	Rolf Carlson
Svetoslav Marinov	Text Dependent and Text Independent Speaker Verification Systems. echnology and Applications	Susanne Schötz Markus Forsberg	Mats Blomberg
Susanne Schötz	Linguistic & Paralinguistic Phonetic Variation in Speaker Recognition & Text-to-Speech Synthesis	Kenneth Willhelmsson Håkan Sundblad	David House
Gabriel Skantze	The use of speech recognition confidence scores in dialogue systems	Rebecca Jonson Gustav Öquist	Mats Blomberg
Håkan Sundblad	Disfluencies and correlations to F0	Gabriel Skantze Svetoslav Marinov	Rolf Carlson
Per Weijnitz	Monophonic Music Recognition	Susanne Schötz Markus Forsberg	Mats Blomberg
Kenneth Willhelmsson	Talking Heads and Hearing Impaired Persons	Gustav Öquist Per Weijnitz	Björn Granström
Gustav Öquist	Formative evaluation of speech components in spoken dialog systems	Håkan Sundblad Svetoslav Marinov	Rolf Carlson

Term Papers 2001

Program and term papers at the closing seminar Speech Technology Course 2001