The people that study this kind of thing spend a lot of time listening to and transcribing human speech. Plus, you can see the vowel formants in a spectrogram of the audio.