Conclusions

Wavelets prove ineffective in our process since we found we lacked the necessary frequency resolution to determing vowel sounds or pitch. Perhaps sampling at higher than 8khz would give better results, but the processing cost needed to pursue higher sampling rates led us other methods...

Formants analysis is an effective way of analyzing vowel sounds regardless of speaker. If we were able to build an extraction program, we would have been well on our way to processing whole words and not just vowel sound.

Pitch determination is an semi-effective way of implementing speaker recognition. Our pitch determination process could be refined or could run in parallel with other speaker recognition methods to produce better results.