Version 1.08:
 * support 64-bit machine.
 * re-write excitation function.
 * add some functions to get generated parameters and number of frames.
 * add some functions for separating synthesis step.
 * bug fixes.

Version 1.07:
 * support new HTS voice format.
 * add LSP postfilter.
 * change volume unit to DB.
 * add function to get generated speech.
 * bug fixes.

Version 1.06:
 * modify state duration calculation algorithm.
 * change many function from void to boolean.
 * change source format.
 * bug fixes.

Version 1.05:
 * support PortAudio for audio output.

Version 1.04:
 * support C++ compiler.
 * add low-pass filter stream.
 * change definition of GV weights.
 * add stop switch.
 * add volume controller.
 * bug fixes.

Version 1.03:
 * add '-z' option to control audio buffer size.
 * change PDF file format.
 * bug fixes.

Version 1.02:
 * context-dependent GV without silent and pause phoneme.
 * buffer size control of audio output for Windows.
 * makefiles for nmake of VC.
 * save detail information.
 * support singing voice synthesis.
 * bug fixes.

Version 1.01:
 * bug fixes.
 * transfer site to SourceForge.

Version 1.00:
 * bug fixes and performance improvements.
 * support linear gain in addition to log gain for LSP-type parameters.
 * first stable release.

Version 0.99:
 * bug fixes.
 * switch license to the New and Simplified BSD license.

Version 0.96:
 * bug fixes.
 * support flexible model structure.
 * support LSP-type parameters.

Version 0.95:
 * support GV.

Version 0.9:
 * implement API-style hts_engine.
 * support speaker-interpolation.