SPIRIT Announces Speaker Verification Software
RUSSIA - SPIRIT, the DSP software developer for TI platform, announces a speaker verification software solution designed for PC-based systems. The software can be used at home and offices integrated with secure telephony, voice-based access systems, voice locks, childproof electronics, and other applications where it is necessary to identify a person by analyzing his/her voice. SPIRIT's speaker verification solution functions in two major operating modes. In static mode, the user has to speak the password for the SV system makes its decision on whether the user is legal. Dynamic mode is more complicated and more trustworthy. In this case the user is required to speak a set of standard system keywords (e.g. numerals) in random order. Dynamic mode is considered significantly more reliable though lengthier. "One of the significant advantages of the system is high accuracy and robustness of provided results regardless of different acoustic conditions," - says Andrew Sviridenko, SPIRIT's founding CEO. "The system based on SPIRIT speaker verification is very robust and the probability of wrong acceptance is less that 0.01%". Speaker Verification system supports both direct and remote speaker verification - with live voice or via phone line. User enrollment and verification can be performed under different acoustic conditions (quiet room, phone verification, etc.). Enrollment and verification thresholds are selectable, enabling an adjustable security level. To improve verification reliability, the algorithm also implements signal/noise ratio and non-linear distortion analysis. SPIRIT's speaker verification software is available in the form of DLL libraries for Microsoft Windows. The company provides services for custom porting to DSP or RISC platforms on request.