The wavelet transform is of interest for analysing non-stationary sign
als. The squared modulus of the wavelet transform leads to the wavelet
spectrogram or scalogram. When signals are embedded in additive noise
, it is important to study the estimation accuracy in terms of bias an
d variance. The mean and variance statistical properties of the wavele
t spectrogram of a signal embedded in additive gaussian white noise ar
e derived in this paper. Examples and simulation results are also pres
ented.