Julius is an open-source, high-performance large vocabulary continuous speech recognition (LVCSR) engine for speech-related researchs and developments. With HMM acoustic model and language model, you can construct your own speech recognition system.
Moved to github: https://github.com/julius-speech/julius
[free version: does not contain grammar-based engine Julian]
In addition to the fixes in normal version of 3.3p2,
This multi-path version newly supports handling of model-skip
transition in any HMM. So, from this version, you can
use "any" type of state transition in HTK format
for acoustic model.
[free version: does not contain grammar-based engine Julian]
Changes in 3.3p1-3.3p2 (multipath):
- newly supports handling of model-skip transition in any HMM.
- add new feature: "-record dir" records speech inputs sucessively into the specified directory.
- fix segfault on Solaris with "-input mfcfile".
- fix adin-cut bug when using module mode and adinnet together.
- add output flush after last recognition output.
Changes in 3.3-3.3p1 (multipath):
- (BIG!) fix misconfigure of default language model weights.
- fix output of wrong value in "-separatescore".
- fix dictionary reading (allow double space)