Extracting Tempo from Audio Files

I extract temporal data from a set of audio signals using Peeters' method in his paper "Time Variable Temp Detection and Beat Marking." This procedure involves creating a reassigned spectrogram and spectral energy flux function, then utilizing the DFT and Frequency Mapped Autocorrelation Function to estimate the periodicity of a given signal. In Peeters' paper, these periodicity amplitudes are then used as observations in the Viterbi decoding algorithm, in which the most likely temporal and metric path of the audio signal is calculated. However, I employ my own approach of tempo estimation where the likelihood of each individual tempo is estimated from the periodicity amplitudes, and the estimated tempo is taken as the weighted average of the three most dominant tempos.