SciSpeech is an open sourced speech analysis, synthesis and recognition toolbox especially designed for Scilab. The SciSpeech toolbox can be easily used or extended by scholars who is doing research in the realm of speech analysis, synthesize and recognition.
SciSpeech provides various analysis functions including time analysis, frequency analysis, censtral analysis and WRLS-VFF analysis, which provides great convenience for the both developers and users, while the famous software MATLAB voicebox toolbox only offers few speech analysis functions.
SciSpeech uses the PSOLA algorithm to supply the prosodics adjusting function in the module of speech synthesis. And this complex function is open sourced so that it can be easily applied and modified in a speech synthesis system.
SciSpeech also includes a practical speech recognition system, which can recognize speech command within a short time and the telephone number recognition ratio is very good. Our system implement the function of speech dialing, it can also shut down, restart and logoff the operating system and open the help of scilab. In fact it contains our open source versions of “speech command” and “speech dialing” functions provided by the "IBM Viavoice" that developed by IBM, which is the fundamental functions of IBM Viavoice.