Two feature extraction algorithms (FEA) are proposed to be implemented for the first beta of HOLO.
One FEA should be wave-based, i.e. song waveform should be analyzed in time-amplitude domain.
The second FEA should be FFT-based, i.e. waveform should be converted to time-frequency domain and all calculations should be based upon the resulting sonogram.
Two feature extraction algorithms (FEA) are proposed to be implemented for the first beta of HOLO. One FEA should be wave-based, i.e. song waveform should be analyzed in time-amplitude domain. The second FEA should be FFT-based, i.e. waveform should be converted to time-frequency domain and all calculations should be based upon the resulting sonogram.