David Nicholson - VocalPy: a core Python package for acoustic comms research | PyData Global 2023

The Need for a Core Package

• VocalPy, a Python package for acoustic communication research, aims to provide a standardized framework for studying animal vocalizations.
• The package is designed to facilitate research by providing a single point of entry for data preparation, analysis, and visualization.

Annotation Formats

• VocalPy supports various annotation formats, including CSV, JSON, and SQLite.
• The package allows researchers to work with different formats, reducing the need to convert data.

Metrics and Evaluation

• VocalPy includes a range of evaluation metrics, such as precision, recall, and F1-score, to assess the accuracy of segmentation and annotation.
• The package also includes tools for computing similarity between annotation formats.

Neural Networks and Deep Learning

• VocalPy utilizes neural networks and deep learning techniques, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), for audio processing and analysis.
• The package can be used to train and predict acoustic features, such as spectrograms and boundary detection.

Interoperability and Scalability

• VocalPy is designed to be scalable, allowing researchers to work with large datasets and distribute their computations.
• The package provides tools for parallelizing computations using Dask.

Community Efforts

• VocalPy has an active community, with contributors from various research fields, including biology, ecology, and computer science.
• The package is hosted on GitHub and has a growing forum for discussion and collaboration.

Case Study

• A case study is presented, demonstrating the application of VocalPy in the analysis of animal vocalizations.
• The case study shows how VocalPy can segment and annotate animal sounds, and how it can be used to study the evolution of speech in animals.

Future Work

• Future work includes the development of more features and tools within VocalPy, as well as further evaluation and testing of the package.
• The package will continue to be developed and maintained to support the research community in acoustic communication research.