Dew.Signal
Dew.Signal is a high-performance digital signal processing library built on top of Dew.Math, providing
a comprehensive suite of optimized algorithms for real-time signal analysis, filtering, spectral
estimation, modeling, and streaming signal workflows. The library is designed for scientific,
engineering, audio, RF, vibration, instrumentation, control, and monitoring applications requiring
both numerical accuracy and hardware-level performance on multi-core CPUs with AVX2/AVX512 support.
Filter Design and Processing:
- IIR filter design (Butterworth, Chebyshev I/II, Elliptic, Bessel), analog and digital domains
- Order estimation, frequency transformations, bilinear and matched-Z transforms
- State-space and zero-pole-numerator-domain modeling with group delay and stability analysis
- FIR filter design using window methods and Remez exchange algorithm
- FIR/Hilbert/differentiator/integrator design, Savitzky–Golay filtering, envelope detectors
- Multi-rate filtering: half-band polyphase filters, decimation, interpolation, zoom-spectrum analysis
Nonlinear and Adaptive Filters:
- Sample-and-hold, sample-and-decay, and median filtering
- High-quality rate conversion with 160dB stopband attenuation at high speed
Spectral and Frequency-Domain Analysis:
- FFT-based spectrum analyzer components with real-time UI integration support
- Parametric spectral estimation: Yule–Walker, Burg, Covariance, Modified Covariance
- Chirp-Z transform, bispectrum, bicoherence, transfer function, coherence estimation
- Peak interpolation and peak-tracking enhancements, phase unwrapping
- Real and complex cepstrum, inverse cepstrum
- Spectral statistics: noise floor, SFDR, THD, THDN, SINAD, RMS, SNR measurements
Signal Synthesis, Streaming, and Measurement:
- Signal generators with stack-based vectorized function evaluation
- Audio capture/playback with monitoring and triggering support
- Data streaming and file format components for continuous acquisition and logging
Forecasting and Time-Series Modeling:
- Spectral forecasting based on controlled peak selection, enabling clear component-based prediction
Integration and Extensions:
- Works seamlessly with Dew.Math numeric structures (vectors/matrices)
- Optional high-performance charting available via a separate Dew.Signal.Tee package (Windows visualization)
- Part of the Dew Lab Studio ecosystem for unified math, DSP, and statistical analysis workflows
Designed for reproducibility, determinism, and stable numerical behavior in long-running or real-time environments.
v6.3.10↙ 5.7K