Real-time noise power-spectral-density estimation from streaming data blocks. More...
#include "../dsp_global.h"#include <fiff/fiff_info.h>#include <QThread>#include <QMutex>#include <QSharedPointer>#include <Eigen/Core>

Go to the source code of this file.
Classes | |
| class | RTPROCESSINGLIB::RtNoiseWorker |
| Background worker that computes a noise power spectral density estimate from accumulated data blocks. More... | |
| class | RTPROCESSINGLIB::RtNoise |
| Controller that manages RtNoiseWorker for real-time noise spectrum estimation. More... | |
Namespaces | |
| namespace | RTPROCESSINGLIB |
Real-time noise power-spectral-density estimation from streaming data blocks.
SPDX-License-Identifier: BSD-3-Clause Copyright (c) 2026 MNE-CPP Authors
RtNoiseWorker accumulates a configurable number of contiguous data blocks, splits them into complete FFT-length segments and averages their Hann-windowed periodograms into a one-sided PSD in dB re unit^2/Hz, the same estimate as scipy.signal.welch with no overlap and no detrending. Longer accumulation windows trade temporal responsiveness for spectral variance reduction. RtNoise is the QObject front-end that owns the worker, forwards configuration changes, and re- emits the PSD result back on the GUI thread.
This estimate is what feeds the live noise-spectrum display and provides a sanity-check companion to RTPROCESSINGLIB::RtCov whenever a quick frequency-domain view of the noise floor is more informative than the full channel–channel covariance matrix.
Definition in file rt_noise.h.