I need some code to compare a sound sample sentence (raw audio) from a native speaker with that of a foreign language student and return a percentage score representing correctness of pronunciation. No user interface needed. Must work with both 16khz 16bit and 8khz 8bit floating point samples. Must work with minimal training data. Must recognize incorrect sentence.