CMU Sphinx 4 1.0 Beta 5 - Changelog

What's new in CMU Sphinx 4 1.0 Beta 5:

September 1st, 2010

· Large arbitrary-order language models
· Simplified and reworked model loading code
· Raw configuration and and demos
· HTK model loader
· Many code optimizations
· JSAPI-independent JSGF parser
· Noise filtering components
· Lattice rescoring
· Server-based language model

What's new in CMU Sphinx 4 1.0 Beta 3:

October 5th, 2009

New Features and Improvements:
· BatchAGC frontend component
· Completed transition to defaults in annotations
· ConcatFeatureExtrator to cooperate with cepwin models
· End of stream signals are passed to the decoder to fix cancellation
· Timer API improvement
· Threading policy is changed to TAS

Bug fixes:
· Fixes reading UTF-8 from language model dump.
· Huge memory optimization of the lattice compression
· More stable fronend work with DataStart and DataEnd and optional SpeechStart/SpeechEnd

What's new in CMU Sphinx 4 1.0 Beta 2:

February 17th, 2009

· new much cleaner and more robust configuration system
· refactored to make use of java5 language features
· new API to setup speech recognizers without any xml
· improved feature extraction (better voice activity detection, many bug-fixes)
· Cleaned up some of the core APIs
· 'include' and 'extends' for configuration files
· better JavaSound support
· fully qualified grammar names in JSGF (Roger Toenz)
· support for dictionary addenda in the FastDictionary (Gregg Liming)
· added batch tools for measuring performance on NIST corpus with CTL files
· many performance and stability improvements