Software from the Speech Processing Group

HMM toolkit STK

This distribution includes SERest - a tool for embedded training of HMM's with supporting scripts. Key features of SERest include re-estimation of linear transformations (MLLT, LDA, HLDA) within the training process, and use of recognition networks for the training. More info here.

Phoneme recognizer based on long temporal context

This distribution contains read-to-use phoneme recognizer featuring the unique LCRC (Left Context - Right Context) architecture from Speech@FIT. This recognizer was included for example in the winning language identification system. It can be used also for indexing and search in audio archives and for on-line keyword spotting. Source code of the recognizer is available under GPL license. The distribution includes neural nets for Czech, Russian, Hungarian and US-English -- these can be used for research and educational purposes. More info here.

On-line keyword spotting system with a microphone input - DLL library

This distribution contain dynamic linked library with a keyword spotting system. The system is able to run on-line with a microphone input. An application interface is very easy and it is consist only from four functions. This DLL is able to make your programs very quickly and easy understand speech commands. More info here.

Sound recording class for Windows

This class encapsulates Windows API functions for recording waveform, a callback mechanism and a buffering and enables easy access to raw waveform. Data is read from the class instance as from a file. More info here.


Quite old ANSI-C software for basic LPC analysis and synthesis.
Author: Jan Cernocky
Documentation: README
Distribution: unianal.tar.gz

More to come...


Software produced by speech processing group.