/audio_classifier

Primary LanguagePython

Audio Classifier

This repo is based on paper ICASSP 2018 https://arxiv.org/pdf/1711.01369.pdf "Knowledge Transfer From Weakly Labeled Audio Using Convolutional Neural Network For Sound Events And Scenes" Anurag Kumar, Maksim Khadkevich, Christian Fügen -> forked from: https://github.com/anuragkr90/weak_feature_extractor

Installation

install ffmpeg: https://pythonbasics.org/convert-mp3-to-wav/ install requirements

RUN

  1. convert MP3 to WAV python convert_mp3_to_wav.py "../COVID-19-train-audio/not-covid19-coughs/PMID-16436200" cough python convert_mp3_to_wav.py "../COVID-19-train-audio/not-covid19-coughs/PMID-32095775" cough

  2. [OPTIONAL] prepare sample data (other class)

wget https://mmadlbs.blob.core.windows.net/share/ambient.mp3 python convert_wav_to_slices_by_time.py ./tmp 2

  1. extract features python feature_extract.py "./data" "./features/out.csv"

  2. run classification ... so far manually