Problem Statement - Analytics and alerts on women safety using mobile microphone and public area cameras.
-Bhrigu kansra
-Ambika
-Jatin Katyal
-Data Cleaning + Training on ML models.ipynb
Contains code of cleaning audio data and merging cleaned audio data chunks into one audio file.
-scrappedData
Contains audio dataset scrapped from Youtube.
-Conversion of audios in MP3 format to wav format.ipynb
Contains code for conversion of mp3 format to wav format of audios.
-scrap.py
Contains code for scrapping audios from Youtube.
-Audio cleaning 6 labels.ipynb
Contains code for cleaning audio files in data folder consiting of 6 labels - Conversations, Stress, Human-Gathering,
Multimedia, Outdoors, Sobb and Cry.
-Model Training Deep Learning 6 labels.ipynb
Contains code for model training on clean dataset available in DataClean folder.
Two models are trained:
- RNN
- CNN
-Spectrogram + KNN + SVM.ipynb
Contains code for plotting the Spectrogram and mel Spectrogram.
Feature extraction using pyAudioAnalysis and training on Two models:
- SVM
- KNN
-audio.csv
Contains the wavefiles along with their labels.
-All5Models.ipynb Contains Feature extraction using pyAudioAnalysis and training on Five models:
- SVM
- KNN
- Random Forest
- Gradient Boosting
- Extra Trees