SCARF: A Segmental CRF Toolkit for Speech Recognition
SCARF is a toolkit for doing speech recognition with segmental conditional random fields. It allows the researcher to advance state-of-the-art performance by harnessing large numbers of independent features. Novel scientific ideas can be evaluated and utilized by adding derived features to the best systems already available.
People
The 2010 Johns Hopkins University Summer Workshop focused on using SCARF. The final presentation may be found here. Broadcast News lattices derived from the workshop are now available through the Linguistic Data Consortium.
Publications
- Samuel Thomas, Patrick Nguyen, Geoffrey Zweig, and Hynek Hermansky, MLP Based Phoneme Detectors for Speech Recognition, in ICASSP, IEEE, 2011
- Justine Kao, Geoffrey Zweig, and Patrick Nguyen, Discriminative Duration Modeling for Speech Recognition with Segmental Conditional Random Fields, in ICASSP, IEEE, 2011
- Kris Demuynck, Dino Seppi, Dirk Van Compernolle, Patrick Nguyen, and Geoffrey Zweig, Integrating Meta-Information into Exemplar-Based Speech Recognition with Segmental Conditional Random Fields, in ICASSP, IEEE, 2011
- Geoffrey Zweig, Patrick Nguyen, Dirk Van Compernolle, Kris Demuynck, Les Atlas, Pascal Clark, Greg Sell, Meihong Wang, Fei Sha, Hynek Hermansky, Damianos Karakos, Aren Jansen, Samuel Thomas, G.S.V.S. Sivaram, Samuel Bowman, and Justine Kao, Speech Recognition with Segmental Conditional Random Fields: A Summary of the JHU CLSP 2010 Summer Workshop, in ICASSP 2011, IEEE, 2011
- Geoffrey Zweig, Patrick Nguyen, Dirk Van Compernolle, Kris Demuynck, Les Atlas, Pascal Clark, Greg Sell, Fei Sha, Meihong Wang, Aren Jansen, Hynek Hermansky, Damianos Karakos, Samuel Thomas, G.S.V.S. Sivaram, Keith Kintzley, Sam Bowman, and Justine Kao, Speech Recognition with Segmental Conditional Random Fields: Final Report from the 2010 JHU Summer Workshop, no. MSR-TR-2010-173, November 2010
- Patrick Nguyen and Geoffrey Zweig, Extensions to the SCARF framework, no. MSR-TR-2010-129, 22 September 2010
- Geoffrey Zweig, Patrick Nguyen, Jasha Droppo, and Alex Acero, Continuous Speech Recognition with a TF-IDF Acoustic Model, International Speech Communication Association, September 2010
- Geoffrey Zweig and Patrick Nguyen, SCARF: A Segmental Conditional Random Field Toolkit for Speech Recognition, International Speech Communication Association, 2010
- Patrick Nguyen and Geoffrey Zweig, Speech Recognition with Flat Direct Models, in IEEE Journal of Selected Topics in Signal Processing, IEEE, 2010
- Geoffrey Zweig and Patrick Nguyen, From Flat Direct Models to Segmental CRF Models, in ICASSP, IEEE, 2010