Spectrogram
| CAT | This article needs proper categorization. You can help Glottopedia by categorizing it Please do not remove this block until the problem is fixed. |
A graphic representation of the components (harmonics or formants) of a sound as they vary in frequency and intensity over time. Frequency is shown on the vertical axis; time is shown on the horizontal axis, and intensity as relative darkness of the image.
Subtypes
Narrowband spectrograms are marked by the more or less narrow horizontal bands which represent the harmonics of the glottal source. The darker bands represent the harmonics that are closest to peaks of resonance in the vocal tract. The lighter bands represent harmonics whose frequencies are further away from the resonance peaks. The bandwidth of the filter used to generate narrowband spectrograms is usually between 30 and 50 Hz. As the fundamental frequency is unlikely to be lower than 50 Hz, a filter with that bandwidth will respond to and capture each harmonic separately as it scans through the frequencies in the speech signal. Narrowband spectrograms have traditionally been used for making measurements of fundamental frequency and intonation.
Wideband spectrograms are marked by the relatively broad bands of energy that depict the formants. The centre of each band of energy is taken to be the frequency of the formant, and the range of frequencies occupied by the band is taken to be the bandwidth of the formant. The relative degree of darkness of a band of energy can be used as a rough estimate of the intensity of the signal, and the relatively large horizontal blank spaces between the formants represent troughs (antiformants) in the resonance curve of the vocal tract. Information about the timing of changes in vocal tract resonance is more reliably obtained from wideband spectrograms. Unlike a narrowband spectrogram, a wideband display will effectively represent an aperiodic source that is being resonanted in the vocal tract. The bandwidth of a filter used to generate a wideband spectrogram is generally between 300 and 500 Hz. A filter with such a relatively wide bandwidth will respond in the same way to one, two, three or even more harmonics that fall within its range: the filter will not resolve the energy within its bandwidth into individual harmonics.
