Hainan gibbon (Nomascus hainanus) bioacoustics dataset for machine learning

Data accompanying the paper: "Empowering Deep Learning Acoustic Classifiers with Human-like Ability to Utilize Contextual Information for Wildlife Monitoring"

We provide the audio data (.wav) used to test our neural network classifier along with the corresponding labelled text files (.svl). The audio and labelled files can easily be viewed in Sonic Visualiser. Drag and drop the audio file. Create the spectrogram layer. Drag and drop the corresponding .svl file.

The dataset provided here is a subset of the full dataset provided here: 10.5281/zenodo.3991714. This dataset has additional files that were manually annotated, which were not manually verified in the original version (10.5281/zenodo.3991714).

Files provided

• Audio_x.zip -- we provide x number of .zip files containing audio files, numbered 1 to 4. These were created in batches to simplify downloads.
• Annotations.zip -- .svl files which contain the manually verified labels. These files can be read in Sonic Visualiser, or as .XML files in a programming language. We took care to annotate the start and stop time of each gibbon call. The height of each bounding box is not important as the frequency range of Hainan gibbons is already known.
• model_weights_tensorflow.hdf5 -- the Tensorflow model. Load the model using: model = tf.keras.models.load_model(model_filepath) note that the model expects a three channel input as explained in the research article.

ED is supported by a research chairship from the African Institute for Mathematical Sciences South Africa. This work was carried out with the aid of a grant from the International Development Research Centre, Ottawa, Canada, www.idrc.ca, and with financial support from the Government of Canada, provided through Global Affairs Canada (GAC), www.international.gc.ca.