ABSTRACT
Ever since the invention of the sound spectrograph some thirty years ago (Koenig, Dunn, & Lacey, 1946), the spectrogram has been the single most widely used form of display for speech. The popularity of the spectrogram is at least partly due to the fact that it is relatively easy to produce, and that it provides a visual display of the relevant temporal and spectral characteristics of speech sounds. To be sure, a speech spectrogram some times introduces distortions to the acoustic structure of speech and often does not provide adequate information on certain linguistically relevant cues, such as stress and intonation. Nevertheless, a speech spectrogram gives a good description of the segmental acoustic cues of speech, and it has been an invaluable tool in the development of our understanding of speech production and perception.
