Skip to content

dkohlsdorf/wdp-ds

Repository files navigation

wdp-ds

Cancel changesData Science For The Wild Dolphin Project

This is a collection of scripts to build and analyze dolphin communication models and communicate the results with domain experts.

History

  • The models from IJCNN20, December 2019 are here.
  • The version with DTW + HMMs is here, June 2020
  • The HTK version of the Conv LSTM HMM Hybrid is here, Aug 2021
  • The version that also includes a lot of evaluastion and training here Sep 2021

Contents

    1. lib_dolphin: Training and Evaluation of Machine Learning Models
    1. pipeline.py: Run this to build a model
    1. decoder_worker.py: Implement everything abotu decoding audio files
    1. app.py: Web server visualizing the results
    1. notebooks: Some experimental notebooks
    1. templates: HTML templates for webapp
    1. dtw_index: dtw index server in c++
    1. dolphin_mnist: java program to label audio data

Dependencies for python