A spectrogram is a plot a spectrum's amplitude (as a color scale) as a function of time (x-axis) and frequency (y-axis). A spectrogram reveals the underlying frequencies that compose the audio signal. The audio signal is decomposed into the component frequencies with a Fast Fourier Transform (FFT). A moving window is passed across the time series data, and the FFT is calculated on each window.
Note that there is a trade-off between frequency and time resolution. A large temporal window results in high frequency resolution and low temporal resolution. The opposite is also true: a small temporal window results in low frequency resolution and high temporal resolution.
Note that the units of amplitude are meaningless, as the colors have been scaled for improved visualization.