Simple Speech Recognition

I need a speech recognition module that takes two speech samples as inputs, the first speech sample is a pronunciation of a known (predefined) English alphabet, syllable or phoneme and the second speech sample is an unknown English alphabet, syllable or phoneme, the module perform a match of the two speech samples and return the probability score that the two samples are same. The module inputs are restricted to pronunciation of alphabets or phonemes, not complete words. The module is to be written in C or C++ language.