The spectrogram shows how the frequency profile of a signal evolves with time. Recently, the rich information encoded by its zeros has been brought to the foreground: the zero set of a spectrogram exhibits a rather rigid pattern with predictable statistics, as soon as the signal is impacted by even a moderate amount of noise, and this information can be exploited in practice.

I will discuss the statistics of zeros of time-frequency representations and their phase windings, and also the performance of algorithms for the computation of zeros from finite data. Joint work with Luis Alberto Escudero, Naomi Feldheim, Antti Haimi and Guenther Koliander.