# Copyright 2001,2002,2003,2004,2005 Sun Microsystems, Inc. # All Rights Reserved. Use is subject to license terms. # # See the file "license.terms" for information on usage and # redistribution of this file, and for a DISCLAIMER OF ALL # WARRANTIES. # Change history for FreeTTS. -------------------------- February 06, 2005 Release 1.2 (final) - Hang with JDK1.5 (bug 1019337) - Added support to build w/o requiring JSAPI - Only build tests on demand (more freedom from JUnit) - Added build targets to create standalone emacspeak app - Added ability for emacspeak demo to strip DECTalk commands from text - Added speaking rate change support to emacspeak demo - Allowed emacspeak text to omit '{' and '}' if text has no spaces - Changed state name expansion logic (bug 1101229) - Added pronunciations for British Sterling and "not" sign (bug 1100888) - Modified missing speech.properties message (rfe 1041437 and bug 1041296) - Added pronunciations for Linux terms (bug 1006630) - Added JUnit to prerequisites list (bug 975557) - Fixed -dump{Multi}Audio problem in freetts app (bug 986503). - Added importing tools specifically for CMU ARCTIC Voices. - Fixed problem in mbrola phonem remapping (bug 1013781) - In the Voice class, removed the call to AudioPlayer.drain() if the audio is cancelled, since it will block until the next audio is available. - Fixed problem in FreeTTSSynthesizer where in some cases waiting threads are not notified when a queue item is cancelled. - Made the JavaStreamingAudioPlayer.cancel() method unsynchronized so that audio output can be stopped immediately. - Added rudimentary internationalization to Festvox import. This imports international (non-US-English) cluster unit voices, but obviously still lacks a lexicon and other NLP components for new languages. June 15, 2004 Release 1.2beta2 - Added experimental CMUArcticVoice that may (or may not) work with the CMU ARCTIC databases. - Added CMUTimeVoice, which has noop post lexical processing to support the FestVox limited domain time voice scheme. Modified FestVoxToFreeTTS to use this voice in the template. - Restructured cluster unit voice for simple extensions. - Added com.sun.speech.freetts.lexicon.userAddenda property - Added FreeTTS pronunciation to addenda - Fix problem where the space character in the path would cause issues with finding voice jars (bug 938225) - Added backwards compatibility for Gnome Speech API - Fix to kill audio output thread on deallocate from JSAPI (bug 922579) - Added "@" and "?" to list of known characters for German. - Fixed problem where some jar files in the FreeTTS directory would cause issues with discovering voices. - Added defaultAudioPlayer property to voice (RFE 922580). December 22, 2003 Release 1.2beta - Many documentation updates to prep for beta release. November 22, 2003 Release 1.2beta - Abandon GNUMake in favor of Ant. - Many documentation updates. - Add capability to set the AudioPlayer on a JSAPI Voice. September 3, 2003 Release 1.2beta - Modified MBROLA output to work on Solaris and Mac August 15, 2003 Release 1.2beta - Added support to import voices from FestVox into FreeTTS - Extended voice packaging scheme to simplify management and addition of new voices. - VoiceManager class provices API to access voices. - Dynamic detection of voice jarfiles is supported. Jun 5, 2003 Release 1.2beta - Updated FreeTTS version message Jun 4, 2003 Release 1.2beta - Imported changes from Flite 1.2: - Regenerated cmulex_addenda, cart_dump - Reads year "1200" as "twelve hundred" rather than "twelve oh zero" - TimeUtils reads 00:01 as "just after midnight" rather than "just after midnight, in the morning". - Added "daughter1" to be an alias for "daughter" - Minor bug fixes in TokenToWords and ClusterUnitSelector - Regenerated test data to compare FreeTTS against Flite 1.2 - Wrote script gen_flite_data.sh to help automate updates to test data. May 13, 2003 Release 1.1.2 - Fixed bug in TokenToWords that causes abbreviations like "OK" and "HI" to be spoken as a US state rather than just "ok" and "hi". - Fixed race condition in JavaStreamingAudioPlayer that would cause output to not be totally cancelled. - Refactored all System.getProperty calls to call a wrapper utility called Utilities.getPropery. This method catches a SecurityException and deals with it properly, so that it can be called from an applet. - Fixed problem in the JSAPI implementation where the audio player was not being closed when the synthesizer was deallocated. This would cause problems for certain types of AudioPlayers such as SingleFileAudioPlayer that require a 'close' before they flush their accumulated wave data. September 6, 2002 Release 1.1.1 - Fixed index.html so that it looks for the RELEASE_NOTES in the proper spot. - Added RELEASE_NOTES to the deploy_doc Make target. - Fixed all run.bat files to run properly under Win32 systems. - Fixed race condition in com.sun.speech.engine.EngineMonitor that would cause a NullPointerException in the Player demo on occasion. - Added emacspeak deploy options - Added non-JSAPI version of Emacspeak speech server - Fixed non-JSAPI Emacspeak demo crashing problem by using the JavaStreamingAudioPlayer (instead of ClipPlayer) as the default. - Added RawFileAudioPlayer to write raw audio bytes to a binary file. - Changed all JDK_DIR references to JAVA_HOME - Fixed problem in parsing certain dates (MonthRange) - Fixed problem with some number parsing - Removed hardcoded array size of LPCResult.residualToFloatMap - Fixed problem where output thread was not being shut down when a voice was closed. - Added TokenGame test which exercises some tricky tokenizing - Screen shots of Player in demo/JSAPI/Player/readme.html now of current version. - parameterized utterance tests to allow easy addition of more tests - demo/freetts/ClientServer batch files specifies host and port - Added contributors agreement: ContributorsAgreement.pdf - Added handout: FreeTTS_handout.pdf FreeTTS_handout.sxw - Added 'How to contribute' section to the web page - Refactored identical demo code in the JSAPI and freetts branches so that they share the same code (put in demo/util directory). - Added "exit" command to Emacspeak. This will close the socket connection. - Added JSAPI/Web Start clock demo to show how to use FreeTTS/JSAPI with Java Web Start. - Added support for MBROLA Voices, as well as JSAPI support for MBROLA Voices. Specifically, the latter refers to check for the MBROLA binaries and databse for an MBROLA voice. If they are found, such a voice is consider 'valid'. - Fixed the following bug: A word followed by a breaking punctuation (:?!), followed by space and a single word, e.g., "RE: Frank", causes the word following the breaking punctuation (i.e., "Frank") not to be synthesized. - Fixed the following bug: "From:" or "From" causes the system to hang. - Added MBROLA support to all appropriate demos. January 30, 2002 Release 1.1 - Created index.php to automatically put news on freetts front page - Added a very small, time-domain specific lexicon - Updated voice to allow easier switching betwen different lexicons - Adapted cluster unit database layout to conform to flite 1.1 format - Adapted cluster unit selection to conform to new database layout - Upgraded 16khz diphone database with Flite 1.1 data. - Added support for duration shrinking/stretching at the segment level. - Added support for pitch shift adjustments at the segment level. - More sophisticated phrasing CARTS. - PostLexicalAnalyzer now deals with "'ve" and "'ll" - Some refactoring of the FeatureProcessors - A number of new Feature processors have been added. - Added ASCII file support for Pronounceable finite state machines - Expanded tokenizer. New features include: 1) processing abbreviations (e.g. U.S.A.) 2) 7-digit phone numbers, 3) partial phone numbers 4) time (e.g. 12:35) 5) digits w/ dashes (e.g. 999-999-999) 6) Roman numerals (e.g. IV) 7) St Andrew's St. 8) US Money (e.g. $14.53) 9) "80s" or "70s" 10) fractions (e.g. 3/4) 11) US states (e.g. NH, MA) - Improved rules for generating binary forms of the databases - Specifies 'Locale.US' (instead of null) when constructing a SynthesizerModeDesc in the demo code - Output of regression tests are more consistent - Workaround for the Linux sound bug reduces JVM crashes. January 8, 2002 Release V1.0.5 - Fixed Locale.US problem in JSAPI demos. January 8, 2002 Release V1.0.4 - Fixed speech.properties path message in demos. - Added SourceForge news to docs front page. December 21, 2001 Release V1.0.3 - Added .bat files for win32 demos December 20, 2001 Release V1.0.2 - Added jsapi.exe - win32 jsapi installer December 20, 2001 Release V1.0.1 - Added jsapi.sh - unix jsapi installer December 14, 2001 Release V1.0 - First public release November 7, 2001 Release V0.9 - Internal Test Release