SeisBench is a toolbox for applying machine learning methods to seismological data. It provides a common interface for accessing benchmark seismic datasets, loading pretrained models and constructing data processing pipelines for training and evaluating neural networks. The framework reduces the amount of custom integration code needed when comparing methods across different datasets or applying existing models to new observations.
The software is organized around three principal components: data management, machine learning models and data generation. Its dataset interface provides access to seismic recordings and associated metadata, while its model interface standardizes how pretrained networks are loaded and applied. The data generation component supports preprocessing, sampling and augmentation for training workflows. SeisBench includes models for tasks such as seismic phase picking, earthquake detection, waveform denoising and depth-phase identification. It also supports models designed for distributed acoustic sensing data, allowing researchers to work with measurements that differ substantially from conventional three-component seismograms. By presenting datasets and models through consistent interfaces, SeisBench makes it easier to reproduce experiments and compare alternative methods.
This is free and open source software.
Key Features
- Unified interface for accessing and managing benchmark seismic datasets.
- Load pretrained machine learning models for seismological applications.
- Train supported neural network architectures using custom seismic datasets.
- Seismic P-wave and S-wave phase picking.
- Earthquake detection and classification using waveform models.
- Seismic waveform denoising and signal reconstruction.
- Depth-phase identification and earthquake depth estimation.
- Support for low-frequency earthquake detection and phase picking models.
- Model interfaces for conventional waveform data and distributed acoustic sensing measurements.
- Preprocessing and data augmentation pipelines for training neural networks.
- Configurable sampling, filtering, normalization and waveform transformations.
- Load and apply models through standardized annotation and classification methods.
- Integration with ObsPy waveform streams and supported distributed acoustic sensing data structures.
- GPU acceleration for compatible machine learning models.
- Support for reproducible training, evaluation and comparison of seismic models.
Website: github.com/seisbench/seisbench
Support:
Developer: SeisBench developers
License: GNU General Public License v3.0
SeisBench is written in Python. Learn Python with our recommended free books and free tutorials.
Related Software
| Earth Science - Geophysics, seismology and earthquake engineering | |
|---|---|
| SimPEG | Simulation and parameter estimation for geophysics |
| pyGIMLi | Modelling and inversion framework for geophysics |
| OpenQuake Engine | Seismic hazard and risk calculation engine |
| ObsPy | Framework to process, analyse, and visualise seismological data |
| SeisBench | Machine learning toolbox for seismology |
| OpenSees | Open System for Earthquake Engineering Simulation |
| OpendTect | Seismic interpretation and subsurface analysis platform |
Read our verdict in the software roundup.
LinuxLinks helps you discover the best free and open source software for Linux. Explore our extensive collection of recommended applications, including alternatives to popular proprietary software.We also publish software reviews, Linux tutorials, hardware reviews, free programming books and articles covering the wider open source community. Browse our Linux resources to discover more. Know a useful free and open source Linux application we haven't covered? Suggest a program for inclusion. Has your software been featured on LinuxLinks? Display one of our Featured on LinuxLinks badges on your website or GitHub README. |


Please read our Comment Policy before commenting.