HLab is a C++ speech recognition engine developed from scratch by Chanwoo Kim under Prof. Bhiksha Raj's supervision. This engine is a Hidden Markov Model (HMM)-based speech recognition engine. It uses the segmental K-means algorithm for training and the viterbi algorithm for decoding. It can handle a trigram language model. This engine can be configured to use either word-model HMMs or phone-model HMMs. This engine has been tested for several speech recognition domains such as Aurora, AN4, etc. The engine is still under development.

This webpage will be updated to include more information about this engine.

The Current source code can be found here.
The Doxygen documentation is availabe here.

About Me

Yeah, it´s me! Chanwoo Kim

chanwcom at gmail dot com

Pittsburgh, PA

Latest News

  • I got the bronze award in the 2011 17th Samsung Humantech thesis after getting the honor prize in the 2010 16th Samsung Humantech thesis.
  • I graduated from CMU and got the Ph. D. degree in Dec. 16th, 2010
  • At ICASSP 2011, one paper was accepted as the first author, and the other was accepted as the second author
  • I did my Ph. D. defense Sept. 20th ,2010
  • One more US patent (Patent No. 7,761,294) is issued July 20th ,2010
  • Two INTERSPEECH 2010 papers were accepted. July 2nd, 2010.
  • This webpage was re-created after I deleted the previous webboard and wiki years ago due to the unsolicited usage of the board by bots July, 2010.