I am a speech scientist, working as a consultant, with large experience on speech
recognition. I have been a key collaborator in various aspects of the
Carnegie Mellon University (CMU) Sphinx project, having been a
designer/developer of Sphinx-2, Sphinx-3,
Sphinx-4, and SphinxTrain, available as open
source. I am always looking for new challenges.
My areas of interest are voice search, robust speech recognition,
lexical modeling, multilingual systems, systems with limited
resources. My resumé
has the details about my professional and academic experience. The Publications
page has my list of publications with copies of the papers. I also
have professional profiles on and
I have been affiliated with several academic and industrial
institutions. In the industry, I worked for some Fortune 500
companies, with tasks including building application packages using
the AT&T Speech API. I worked on voice search while at the Mitsubishi
Electric Research Labs (MERL), in the Boston area. I was
also one of the key developers who created the speech recognition
system at Honeywell Vocollect, in Pittsburgh.
In the academic world, I was a research staff member at Carnegie Mellon, affiliated with the Robust Speech Group as well as the
project LISTEN. More recently, I was an affiliate
member of the Machine Learning for Signal Processing Group (MLSP) of Carnegie Mellon's Language
Technologies Institute. I was also engaged in a joint project with the T.U. Darmstadt.
I have been involved in the open source CMU Sphinx
project since 2001.