Bird or Human Classifier

• 

  Data that got the model to work

• 

  Audio visualization of one of our files

Inspiration

After taking a beginner-level course on Al, we wanted to code it on our own, and see if we could make a simple classifier with audio files.

What it does

This model classifies whether humans or birds made the sounds in the audio files it tests on.

How we built it

We used many python libraries such as scikit, pandas, numpy, librosa, glob, and other tools for classifying based on nearest neighbors.

Challenges we ran into

The file size for training and testing data was very large, the model got angry when you gave it arrays, it took very long to run the code and download our data, saving our code as a csv file, not having enough ram to run the program sometimes

Accomplishments that we're proud of

We were able to get around 70% accuracy on the model, which is decent for our first machine learning project, and our first project that works with audio data.

What we learned

We learned about useful python libraries, how to make a machine learning model on our own, and how to turn audio files into numerical data that Al can train off of.

What's next for Bird or Human Classifier

We can have the program classify what type of bird is making the noise, if it detects that the noise made is from a bird and not a human.

Built With

• glob
• machine-learning
• numpy
• pandas
• python
• scikit
• seaborn