Hadamard-transform (HT) multiplexing has recently been applied to increasingly complex spectroscopic techniques. It had been shown that the data acquisition time for fluorescence excitation emission matrix (EEM) spectroscopy can be reduced by 1 or 2 orders of magnitude using HT multiplexing of the excitation light using a programmable light source. In these previous studies, the data acquisition rate had been limited by the time it took to record an EEM, that is, to complete one cycle of multiplexed excitation spectra. The extraction of chemical information, such as concentration and chemical identity, is then obtained from parallel factor (PARAFAC) analysis of the sequence of EEMs. In this contribution, we increase the data acquisition rate by another order of magnitude, limited ultimately by the time it takes to record a single excitation spectrum. Our algorithm is entirely based on improved data processing, that is, it can be applied to previously recorded HT multiplexed data sets. The algorithm is based on three previously unexplored approaches: (1) we perform a PARAFAC multivariate analysis on the raw (multiplexed) data set, (2) the time-independent PARAFAC loading vectors are obtained prior to obtaining the time-dependent score vectors, and (3) when loading vectors are difficult to obtain from the EEMs, we instead use a rolling-average approach to considerably increase the stability of the fit. Analysis of experimental data shows that the scores of fluorescence EEMs with seven excitation wavelengths and over 1000 emission wavelengths can be obtained in less than 20 ms.