{"cells":[{"metadata":{},"cell_type":"markdown","source":"I reported that [discovering missing labels and retraining](https://www.kaggle.com/c/rfcx-species-audio-detection/discussion/208830) did not worked.  \nFor your inference, I show missing labels example.\n\nexample: *0ab6aa734.flac*"},{"metadata":{"_kg_hide-input":true,"trusted":true,"scrolled":false},"cell_type":"code","source":"import librosa\nfrom pathlib import Path\nimport IPython.display as ipd\n\nROOT = Path(\"..\")\nTRAIN_AUDIO_ROOT = Path(\"../input/rfcx-species-audio-detection/train\")\n\nSR = 48000\nname = \"0ab6aa734.flac\"\nprint(name)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4704212%2F9ca088bb386abf7114543c019c1d8a5f%2Ffig.png?generation=1609892974092435&alt=media)\n\n+ Tp label(species id is \"9\") is given at around 49~52sec.\n+ But trained model discovered many pseudo-labels in the whole clip.\n\nIs that true?  \nLet's listen."},{"metadata":{"trusted":true,"scrolled":true},"cell_type":"code","source":"clip, _ = librosa.load(TRAIN_AUDIO_ROOT / name,\n                       sr=SR,\n                       mono=True,\n                       res_type=\"kaiser_fast\")\nipd.display(ipd.Audio(clip, rate=SR))      ","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"I repeat, discovering missing labels and retraining **did not improve score** for me."}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}