In　order　to　improve　the　accuracy　of　bird　sound　recognition　under　non-stationary　noise,　a　rapid　bird　sound　recognition　method　using　anti-noise　texture　features　was　proposed.　This　method　also　solved　the　problem　of　big　occupation　space,　as　well　as　large　computation,　time-consuming　of　gray　level　co-occurrence　matrix　(GLCM)　in　extracting　texture　features.　The　method　contained　three　steps.　Firstly,　short-time　spectrum　estimation　was　conducted　on　noise　spectrum　of　bird　sound　to　get　the　enhanced　spectrum.　Then,　sum　and　difference　histograms　(SDH)　were　applied　to　calculate　texture　features　rapidly.　Finally,　random　forest　(RF)　was　used　to　make　classification.　In　the　experiment　part,　two　contrast　experiments　were　designed.　The　results　show　that　this　method　has　good　recognition　performance,　lesser　time　consumption;　meanwhile,　it　is　robust　to　noise.　©,　2015,　Chinese　Institute　of　Electronics.　All　right　reserved.