SPECTROGRAMS

Visual representations of vocal sounds, called spectrograms, are made with an instrument called a sound spectrograph. The examples shown here were made in the phonetics lab of the Linguistics Department of McGill University. Vocal sounds were recorded in an acoustically dead room with a microphone. With the help of an acoustic engineer, I recorded my voice into a half-track tape recorded in the adjoining room and the recorded sound was then fed into the spectrograph. The sound is converted into heat energy by a bank of electro-acoustic filters. The output from the filters goes to a heat stylus which burns varying degrees of darkness around a revolving drum, and the resulting pattern that appears on paper is called a spectrogram. The spectral content of sound is displayed on an axis of frequency against time. The relative intensity of component frequencies is shown by the relative darkness of the impression. Formants appear as dark, horizontal bands on the spectrogram, in the low end of the frequency spectrum between 85 and 3500 Hz. (Fig.5). The high end of the frequency spectrum, from aprox. 3500 to 8000 Hz. shows in general an uneven distribution of sound over a greater bandwidth. This is the result of consonantal noises produced during speech by lips, teeth, and the back of the tongue. (Fig.6) Spectrograms of speech sounds reveal organized patterns of formants and noise sounds, separated by short intervals of silence. Silences re the result of the spacing of world boundaries, as well as interruptions in the flow of sound caused by stop consonants.

Explanation of the Formant Charts

In formant charts, each symbol represents an area, rather than a point in the vowel space. The frequency of the first formant is shown on the vertical axis, plotted against the frequencies of the first and second formant on the horizontal axis. The axes have been placed so that zero frequency is in the upper right corner, arranged so that formant charts will correspond to the charts used by linguists to indicate the auditory qualities of vowels. Vowel symbols are arranged from left to right to correspond with the front to back dimensions of the oral cavity. Secondary articulations are placed in columns, following the primary place of articulation. The primary organization is by decreasing degrees of stricture. The diagram roughly corresponds to the mouth, with the left side representing the front of the mouth, the right side the back of the mouth.
Frequencies are arranged according to the “mel” scale, in which perceptually equal intervals are represented as equal distances along the scale.

Advertisements