soundevent 0.1.1

soundevent is an open-source Python package for the computational biocoustic community, providing standardized tools for sound event analysis and data management.

soundevent

Warning This package is under active development use with caution. However, most of the data definitions are not expected to change. Will be adding data loading/exporting and prediction evaluation functions soon.

soundevent is an open-source Python package that aims to support the computational biocoustic community in developing well-tested, coherent, and standardized software tools for bioacoustic analysis. The main goal of the package is to provide a flexible yet consistent definition of what a sound event is in a computational sense, along with a set of tools to easily work with this definition. The package comprises three key components:

Main features

1. Data Classes for Bioacoustic Analysis

The soundevent package defines several data classes that conceptualize and standardize different recurrent objects in bioacoustic analysis. These data classes establish the relationships between various concepts and specify the attributes each object possesses. They are designed to be flexible enough to cover a wide range of use cases in bioacoustic analysis. The package also includes data validation mechanisms to ensure that the information stored is valid and meaningful. Specifically, it defines objects related to sound events, such as user annotations and model predictions.

2. Serialization, Storage, and Reading Functions

To promote standardized data formats for storing annotated datasets and other information about sounds in recordings, the soundevent package provides several functions for serialization, storage, and reading of the different data classes offered. These functions enable easy sharing of information about common objects in bioacoustic research. By employing a consistent data format, researchers can exchange data more efficiently and collaborate seamlessly.

3. Handling Functions for Sound Events

The soundevent package also includes a variety of functions that facilitate the handling of sound event objects. These functions serve multiple purposes, such as matching sound events for model prediction evaluation, transforming sound events, and managing metadata and labels. By offering a comprehensive set of handling functions, the package aims to streamline the analysis workflow for bioacoustic researchers, providing them with powerful tools to manipulate and extract insights from their data.

Installation

You can install soundevent using pip:

pip install soundevent

Documentation

For detailed information on how to use the package, please refer to the documentation.

Example Usage

Here's a brief example demonstrating the usage of soundevent package:

import soundevent

dataset = soundevent.load_dataset("example_dataset.json", audio_dir="audio")
print(f"Dataset {dataset.name} has {len(dataset.recordings)} Recordings")

recording = dataset.recordings[0]
print(f"Recording {recording.id}:  duration={recording.duration}, samplerate={recording.samplerate}")

wav = soundevent.load_recording_audio(recording)

Contributing

We welcome contributions from the community to make soundevent even better. If you would like to contribute, please refer to the contribution guidelines.

License

soundevent is released under the MIT License.