Vowels:

The first thing I noticed when considering these graphs is that anytime there is a significantly dominant portion to the waveform of the word, it almost always corresponds to the vowel sound in the word. Also, vowel sounds seem to typically create a horizontal striping pattern in the spectrogram. This would seem intuitive considering the quasi-periodic nature of the vocal activity in vowel production, and the harmonics that may be produced within the tract. Vowels tend to have a strong energy response, centered at a clearly noticeable band of frequencies on the spectrogram. For example, "beat" has a very clearly defined frequency band centering about 3500 Hz. This may most likely be near to the fundamental frequency of the vowel phoneme /i/.

Consonants:

A noteworthy observation about the unvoiced plosives concerns their energy response as displayed by the spectrographic analysis. They generally have an almost flat energy response over a loosely restricted part of the range. The voiced plosives were more diverse in their energy responses, but had more noticeable uniformity in the frequency spectrum. They generally peaked at a notch band of frequencies, and then maintained an almost flat, relatively low-magnitude frequency response after that peak band. The peak in frequency most likely is a result of the voicing. The energy spectrum differences were probably a result of the differences in velocity required to produce a particular type of plosive.

Amongst fricatives, one interesting characteristic is a random pattern that is produced, similar to noise, with a relatively evenly distributed energy spectrum. Unvoiced fricatives in particular were noticeable for this behavior. Voiced fricatives tend to act in this manner as well, however this noise-like behavior is bandlimited, seemingly centered on the pitch frequency of the phoneme. On the frequency spectrum, fricatives in general seem to have the majority of the energy of the signal focused in on a particular frequency band, which may or may not be noticeably divided into separate formants within that band.

Nasals however behave differently. They seem to emulate the same horizontal striping patterns in the spectrographic analysis of the vowel sounds. This could be due in part to the resonance of the nasal cavity and in part to the fact that a nasal sound is produced by rerouting what otherwise would have been a vowel sound - if it had been allowed to escape through the mouth.

The Whispers produced an interesting effect on the spectrogram, displaying a behavior almost indicative of white noise. The 'h' sound in the sample word, "how," has a relatively evenly distributed energy spectrum with time. This observation seems logical considering the nature of the sound as simply a puff of air with no vocal excitation behind it.

Diphthongs:

Perhaps my favorite spectrogram pattern is the result from "boy." You can clearly see the frequency transition between vowels that characterizes the diphthong (see glossary). The other words also display this characteristic, but not as dramatically. There is a 'z'-shaped pattern which indicates to me that there is a transition in time from an energy concentration in one frequency band to an energy concentration in another band.

Affricates:

The affricates seem to have a rather similar energy spectrum, condensed over a small band of frequencies and consisting of rather uniform energy within that band. Their frequency response is also rather similar, peaking in the higher frequencies. This is most likely a result of the higher velocity at which the air is being forced out of the oral cavity constriction, creating a higher frequency of vibration.

Semi-Vowels:

The glides tend to have a lower energy level than the liquids, but neither type displayed high energy levels in general. Both also had frequency responses which decayed as the frequency increased - almost exponentially. This might be due to the transitional nature of semi-vowels. By existing more to compliment and transition into another vowel sound, semivowels may not necessarily need to be very high in energy.