Electroencephalographic (EEG) signals are generally non-stationary, however, nearly stationary brain responses, such as steady-state visually evoked potentials (SSVEP), can be recorded in response to repetitive stimuli. Although Fourier transform has precise resolution with long time windows (5 or 10 s for instance) to extract SSVEP response (1-100 Hz ranges), its resolution with shorter windows decreases due to the Heisenberg-Gabor uncertainty principle. Therefore, it is not easy to extract evoked responses such as SSVEP within short EEG epochs. This limits the information transfer rate of SSVEP-based brain-computer interfaces. In order to circumvent this limitation, we concatenate EEG signals recorded simultaneously from different channels, and we Fourier analyze the resulting sequence. From this constructed signal, high frequency resolution can be obtained with time epochs as small as only 1 s, which improves SSVEPs classification. This method may be effective for high-speed brain computer interfaces (BCI).