Inspiration
I chose this project because I enjoy birding and being in nature. I wanted to implement a classifier that would help improve my birding experiences, as well as keep track of the different species of birds that live near me.
How I built it
I used Sci-kit Learn's PCA method to keep the most important components of the spectrograms and passed them into Sci-kit's SVM method to classify the bird call.
Challenges I ran into
Converting the audio files to usable data was challenging and took a very long time. This left me with little time to try different hyperparameters for my model.
Accomplishments that we're proud of
I am proud of completing the web app and having it output a bird species based on the input bird call. As well as my model being a lot more accurate than random guessing.
What we learned
I learned how to successfully implement a machine learning model as well as make a functioning web app.
What's next for Bird Call Classifier
I plan on implementing a CNN to classify the bird calls to increase the accuracy as the accuracy of my current model isn't as high as I'd hoped for.
Built With
- librosa
- python
- scikit-learn
Log in or sign up for Devpost to join the conversation.