Traditionally,　diffuse　correlation　spectroscopy　(DCS)　derives　blood　flow　(BF)　by　measuring　of　the　temporal　intensity　fluctuations　of　multiply　scattered　light　from　a　single　source-detector　pair.　In　this　paper,　a　multi-wavelength　DCS　approach　was　proposed　to　quantify　tissue　blood　flow　and　scattering　coefficient　based　on　long　short-term　memory　(LSTM)　architecture.　Phantom　experiments　were　established　to　measure　normalized　intensity　autocorrelation　function　data　by　multi-wavelength　DCS　system　at　different　velocities　and　scattering　coefficients.　The　results　support　the　notion　of　using　proposed　LSTM　architecture　for　quantification　of　blood　flow　and　scattering　coefficient　in　DCS.