Subthreshold　depression(SD)　is　a　prodromal　state　of　depression　that　both　affects　quality　of　life　and　increases　the　risk　of　developing　major　depression.　Early　intervention　is　challenging　due　to　the　difficulty　of　accurate　identification　by　conventional　diagnostic　methods.　Speech,　as　a　natural,　non-invasive　signal　rich　in　emotional　information,　has　the　potential　to　recognise　SD.　There　is　a　lack　of　data　related　to　SD　in　existing　publicly　available　datasets,　so　this　study　established　a　specialised　voice　database　by　self-mining　data.　The　theory　of　pentatonic　can　comprehensively　reflect　the　physiological　and　psychological　state　of　an　individual,　and　has　unique　advantages　that　cannot　be　covered　by　traditional　speech　features.　Based　on　this　theory,　this　paper　extracted　the　pentatonic　features,　i.e.,　the　sum　of　the　energies　of　the　corresponding　frequency　bands　of　the　five　phonemes:　Gong,　Shang,　Jue,　Zhi,　and　Yu.　Comparing　the　distribution　of　pentatonic　energy　between　SD　and　healthy　people,　we　found　that　the　pentatonic　energy　of　healthy　people　is　higher　than　that　of　SD,　and　the　trend　of　pentatonic　energy　is　the　same　in　both　groups.　The　results　provide　a　new　idea　and　basis　for　the　early　detection　and　intervention　of　SD.