[Index]

3 Acknowledgements

The initial development of flite was primarily done by awb while travelling, perhaps the name is doubly appropriate as a substantial amount of the coding was done over 30,000ft). During most of that time awb was funded by the Language Technologies Institute at Carnegie Mellon University.

Kevin A. Lenzo was involved in the design, conversion techniques and representations for the voice distributed with flite (as well as being the actual voice itself).

Other contributions are:

Nagoya Institute of Technology The MLSA, MLPG code comes directly NITECH’s hts engine code, though we have done some optimizations.
Marcela Charfuelan (DFKI) For the mixed-excitation techniques (but no direct code). These originally came from NITECH but we understood the techniques from Marcela’s Open Mary Java code and implemented them in our optimized version of MLSA.
David Huggins-Daines: much of the very early clunits code, porting to multiple platforms, substantial code tidy up and configure/autoconf guidance (up to 2001).
Cepstral, LLC (http://cepstral.com): For supporting DHD to spend time (in early 2001) on flite and passing back the important fixes and enhancements while on a project funded by the Portuguese Foundation for Science and Technology (FCT) Praxis XXI program specifically to produce an open source synthesizer.
Willie Walker <william.walker@sun.com> and the Sun Speech Group: lots of low level bugs (and fixes).
Henry Spencer: For the regex code
University of Edinburgh: for releasing Festival for free, making a companion runtime synthesizer a practical project, much of the design of flite relies on the architecture decisions made in the Festival Speech Synthesis Systems and the Edinburgh Speech Tools.
The duration cart tree and intonation (accent and F0) models for the US English voice were derived from the models in the Festival distribution. which in turn were trained from the Boston University FM Radio Data Corpus.
Carnegie Mellon University The included lexicon is derived from CMULEX and the letter to sound rules are constructed using the Lenzo and Black techniques for building LTS decision graphs.
Craig Reese: IDA/Supercomputing Research Center and Joe Campbell: Department of Defense who wrote the ulaw conversion routines in src/speech/cst_wave_utils.c