SPWB 1.1.1

Full Changelog: v1.1.0...v1.1.1

SPWB 1.1.0

Full Changelog: v1.0.2...v1.1.0

SPWB 1.0.2

Documentation only — no code changed.

Every docs/ link in the PyPI project description pointed nowhere. PyPI
renders the README standalone, so a relative link like docs/manuals/
resolved against pypi.org/project/spwb/ instead of the repository. On
GitHub the same links worked, which is why it went unnoticed.

They are absolute now, so the manuals, the HDF5 format reference and the
example notebooks are reachable from the project page as well as from the
repository.

SPWB 1.0.1

Metadata only — no code changed, and no reason to upgrade if 1.0.0 works
for you.

PyPI serves whatever the latest release declares, so the project page still
described SPWB as a port of a LabVIEW application. It now says what the
software is:

A DSP library and desktop application for spectra, transfer functions,
spectrograms and adaptive filtering

The keyword list was refreshed to match, picking up adaptive filtering,
frequency response, hdf5 and measurement — all of which the package
gained without the metadata following — and dropping labview. The
heritage is not hidden; it is recorded where it means something, under
Numerical fidelity,
alongside the reference data the numbers are pinned to.

SPWB 1.0.0

The first release of pySPWB, the Python port of Charette AI Group's
Signal Processing Work Bench — originally a LabVIEW application, now
open source and running anywhere Python does.

pip install spwb          # the library: numpy + scipy, no GUI stack
pip install spwb[gui]     # the full application, then run:  spwb

What it is

Two things, and either is usable without the other:

• a signal-processing library — spectra, transfer functions, coherence,
  spectrograms, adaptive filtering, plus HDF5/TDMS/WAV/CSV IO and read-only
  RPC-III, Nastran punch and HEAD acoustics — that runs happily in a notebook;
• a desktop application built on it, with the multi-window signal
  sharing the original was known for.

spwb.processing is Qt-free by contract and a test suite enforces it, so a
notebook user never installs Qt.

The numbers are pinned, not re-derived

Reference data generated by driving LabVIEW 2022 itself over COM is
committed in tests/fixtures/, so results carried over from the original
application still match.

Documentation

A user manual
for every analysis window — Time Processing, FFT Analysis, Transfer
Function, Time-Frequency and Adaptive Filtering — each working through
demonstration datasets whose expected values are checked automatically, so
every number quoted is one the application demonstrably produces. Each has a
companion notebook that computes the same numbers in a few lines.

The datasets are built into the package: File > Create Demo Data ..., or
from spwb.demo import write_demo_data. The manuals are reachable from
inside the application under Help, where F1 opens the manual for the
window you are in.

Platforms

One py3-none-any wheel for Windows, macOS and Linux; Python 3.10 to 3.13.
The GUI is tested on all three operating systems.