Processing procedures based on the continuous complex wavelet transform are applied to the analysis of velocity signals obtained in a developing turbulent flow. It is shown that the frequencies contributing to the energy and to the Reynolds stresses may be characterized better than with Fourier analysis. Indeed, the time variation of the contribution of the different frequencies to the energy of the fluctuations and to the Reynolds stresses may be singled out, so that it may be ascertained if these contributions are simultaneous or alternative, continuous or intermittent. In particular, it is shown that in certain positions at the end of the potential cores of a coaxial jet high energy fluctuations, probably connected to the passage of rolled-up vortical structures, may produce low values of the Reynolds stress because their contribution to correlation oscillates in time between positive and negative values. Examples of the application of different types of wavelet dynamical filters are also given, and a wavelet local correlation coefficient is defined.