Figure 3From: Biomimetic multi-resolution analysis for robust speaker recognitionSpeech signal spectral analysis. (a) Average spectral modulation profile Y ¯ (Ω)= Y ( Ω ; t 0 ) T Ψ ; (b) Top panel: MI between feature representation and speech message as a function of scale. Bottom panel: MI between feature representation and speaker identity as a function of scale.Back to article page