There were two algorithms investigated by this project. The first used a tokenization routine to express the time variance of the dominant frequency of a given signal as a token string that could then be searched. The second used a variety of neural networks for pattern recognition of the audio recordings.

Tokenization Algorithm

Neural Network