An auditory feedback for Brain Computer Interface (BCI) applications is proposed. This is achieved based on the so-called sonification of the mental states of humans, captured by Electro-Encephalogram (EEG) recordings. Two time-frequency signal decomposition techniques, the Bump Modelling and Empirical Mode Decomposition (EMD), are used to map the EEG recordings onto musical scores. This auditory feedback proves to have extremely high potential in the development of on-line BCI interfaces. Examples based on the responses from visual stimuli support the analysis.