From SpeechWiki

Jump to: navigation, search

Landmark-Based and Prosody-Dependent Speech Recognition, Summer 2009

The landmark-based speech recognition group will meet during Summer 2009 on roughly alternate Thursdays, from 10:00-11:30 AM, in Beckman 2369.

August 6, 2009
Sarah will present her work with auditory modeling.
July 23, 2009
Chi will lead a discussion of three papers on finite state transducers to model pronunciation variation:
Timothy J. Hazen, I. Lee Hetherington, Han Shu, and Karen Livescu, 2002. PRONUNCIATION MODELING USING (Priority)
Han Shu and I. Lee Hetherington, 2002. EM TRAINING OF FINITE-STATE TRANSDUCERS
I. Lee Hetherington, 2001. An Efficient Implementation of Phonological Rules
using Finite-State Transducers.
Chi will present her work with Xiaodan on word recognition from tract variables using Vikram's data(If time is enough)
July 16, 2009
Alina will discuss her current work on the frequency of vowel co-occurrence patterns in the English CELEX lexicon.
Dave will lead the discussion of Tilsen & Johnson's JASA paper. The 2008 CLS paper covers the same material but is lighter on technical detail and directed to a linguistics reader. The 2009 CogSci paper will not be discussed but is shared here.
Tilsen, S. & Johnson, K. (2008). Low-frequency Fourier analysis of speech rhythm. Journal of the Acoustical Society of America, 124:2, pp. EL34-39.
Tilsen, S. (2008). Relations between speech rhythm and segmental deletion. Paper presented at the 44th annual meeting of the Chicago Linguistic Society.
Tilsen, S. (2009). Multitimescale dynamical interactions between speech rhythm and gesture. Cognitive Science, 33, 839-879.
These articles can be found at
July 2, 2009
Alina discussed the design of her EMA study on plosive release
June 18, 2009
Discuss plans for summer
May 7-8, 2009
Multi-University Landmark-Based Speech Recognition Group Meeting
University of Maryland
April 30
Practice talks for Illinois Speech Day, ASA
Yoonsook Mo, Arthur Kantor, Chi Hu, Jui-Ting Huang, Sarah Borys
April 23
A nice intro to kernel methods is Kernel Methods and their potential use in signal processing, F. Perez-Cruz, O. Bousquet, IEEE SIGNAL PROCESSING MAGAZINE MAY 2004 --Arthur
April 16
Discussion of Interspeech Papers
April 9
April 2
March 26
Spring break
March 19
Five-minute presentations of student research; Bob McMurray will be here
March 12
Practice of the Universal Access Open House demo
Heejin Kim, Mark Hasegawa-Johnson, Sarah Borys, Sujeeth Bhardwoy
March 5, 2009
Language Processing in the Natural World, Michael T. Tanenhaus and Sarah Brown-Schmidt
February 26, 2009
Cross-Lingual Recognition and Sound Pattern Retrieval, Jui-Ting Huang and Xiaodan Zhuang
February 12, 2009
Automatic Burst Location, Alina Khasanova
February 19, 2009
Discussion of Kuperman et al. 2008 (JASA v. 124.6) and Margaret Fleck's attempts to replicate results with Buckeye
February 5, 2009
F0 Peak and Formant Values as Cues for Prominence, Yoonsook Mo
January 29, 2009
Landmark-Based Speech Recognition Using SVM/HMM Hybrids, Sarah Borys
January 22, 2009
Planning meeting

Landmark-Based and Prosody-Dependent Speech Recognition, September 12, 2008

Faculty and students from the University of Maryland, Boston University, the University of Illinois, UCLA, and USC met in Urbana on September 12, 2008 to present new results in landmark-based speech recognition.

Insights Into Pronunciation Modeling and ASR Using Mixed Unit Pronunciation Models
Arthur Kantor
Personal tools