Time-frequency SSIM (WSSIM): Hines et al. [8] applied the SSIM to short-time Fourier transforms of acoustic signals,where the image ‘intensity’ was the modulus of the Fourier coefficients. In this work,a continuous wavelet transform(CWT)was used to take advantage of the higher temporal resolution for higher frequency signal components.The wavelet used here was a complex-valued Morlet wave let with ω0 =6 as it shares the characteristic shape of a modulated wave packet,
leading to efficient collapse of the investigated signals on to a few wavelet coefficients. The wavelet scales can be converted to pseudo-frequencies [10], and the resulting time–frequency ‘image’ is then p˜ (t,f pseudoÞ) = |jCWT [p(t)]