1 2The initial development of flite was primarily done by awb while 3travelling, perhaps the name is doubly appropriate as a substantial 4amount of the coding was done over 30,000ft). During most of that 5time awb was funded by the Language Technonologies Institute at 6Carnegie Mellon University. 7 8Kevin A. Lenzo was involved in the design, conversion techniques and 9representions for the voice distributed with flite (as well as being 10the actual kal voice itself). 11 12Other contributions are: 13 14Henry Spencer 15 For the regex code 16University of Edinburgh 17 for releasing Festival for free, making a companion runtime synthesizer 18 a practical project, much of the design of flite relies on the 19 architecture decisions made in the Festival Speech Synthesis Systems and 20 the Edinburgh Speech Tools. 21 The duration cart tree and intonation (accent and F0) models were 22 derived from the models in the Festival distribution. which in turn 23 were trained from the Boston University FM Radio Data Corpus. 24Carnegie Mellon University 25 The included lexicon is derived from CMULEX and the letter to sound 26 rules are constructed using the Lenzo and Black techniques for 27 building LTS decision graphs. 28Nagoya Institute of Technology 29 The mlsa code derives from HTS (following a long chain) 30Tomoki Toda 31 The mlsa and mlpg support came view Tomoki's support for voice convertion 32 in FestVox which in turn (some of which) comes from NITECH's HTS. 33Marcela Charfuelan (DFKI) 34 For the mixed-excitation techniques. These originally came from NITECH 35 but we understood the technqiues from Marcela's Open Mary Java code and 36 implemented them in our optimized version of MLSA. 37David Huggins-Daines (dhd@cepstral.com) 38 much of the clunits code, porting to multiple platforms, substantial 39 code tidy up and configure/autoconf guidance. 40Cepstral, LLC (http://cepstral.com) 41 For supporting DHD to spend time (in 2001) on flite and passing 42 back the important early fixes and enhancements including SAPI 43 support (funded by Portuguese FCT to produce an open source 44 synthesis solution). 45Willie Walker <william.walker@sun.com> and the rest of the Sun Speech Group 46 lots of low level bugs (and fixes). 47Portuguese Foundation for Science and Technology (FCT) Praxis XXI program 48 The SAPI interface provided by Cepstral, LLC was partially funded by 49 the above program. 50Craig Reese: IDA/Supercomputing Research Center 51Joe Campbell: Department of Defense 52 who wrote the ulaw conversion routines in src/speech/cst_wave_utils.c 53Mario Lang: 54 causing the support of shared libraries to happen 55Eric House (fixin@peak.org) 56 who provided examples of how to do 68K Call Backs for system functions 57Greg Parker gparker@sealiesoftware.com 58 peal, the binding glue and shared library foo for getting the arm 59 version doing something reasonable under PalmOS 60Lukas Loehrer <loehrerl@gmx.net> Feb 2006 61 alsa support (default if available) 62Udhyakumar N 63 For making the mixed excitation code work, and show its value 64Brian Langner 65 redid the Visual Studio support 66Alok Parlikar 67 Android support, and cg voice dumping (and loading), indic support 68Gopala Anumanchipalli 69 spamf0 support, unitran integration 70Richard Sproat and Kyoung-young Kim (UIUC) 71 Unitran: unicode to sampa grapheme mapping tables 72Sun Microsystems 73 g72x code 74Larry McCourry 75 Windows Visual Student support for 2.0.0 76Cobalt Speech and Language Inc 77 Updates to Visual Studio Support and Support for Clustergen Voices under SAPI 78Suresh Bazaj and Shyam Krishna 79 Indian Language support 80