Train AI model to Detect AC

[andrewg23]

Use pyAudioAnalysis to Train an AI model to detect an AC in an audio clip.

[andrewg23]

Simply running the following import statement causes my kernel to crash.

from pyAudioAnalysis import audioTrainTest as aT

I'm not sure why this is happening. I will review the pyAudioAnalysis docs to make sure I didn't miss any steps in the installation process.

[amalaquias]

This works for me. Could you try installing opencv-python==4.5.3.56? It should let you as described here. If you're able to revert back to this version and you still receive kernel crashes, try following step 9 of setup/headful/virtualEnvironmentSetup.md

[andrewg23]

Solved the kernel crashing problem by downgrading to numpy==1.18.1 as stated in #13.

[amalaquias]

Uploaded an initial notebook that outlines some of the functionality of pyAudioAnalysis. I will continue to update this notebook to include more on that library and for Librosa

[andrewg23]

I trained an AI model to detect ACs based on audio samples. I need to test it to see how accurate it is. About 30 audio samples were used to train it so it may require more.
It can be found in projects/AC AI.ipynb

[andrewg23]

I am trying to train the model using all the files from the UrbanSounds dataset to improve it's accuracy at classifying the sounds. It doesn't support 24 bit files, so I will have to identify and filter out the 24 bit files. I will do this by modifying the /projects/urbanSoundsSorter.py script that I wrote to sort the files. It will check the bit depth of each file and only sort it if its not a 24 bit file.