Releases
v1.4.8
Compare
Sorry, something went wrong.
No results found
dspx
1.4.3
Patch Changes
1.4.2
Patch Changes
1.4.1
Patch Changes
1.4.0
Minor Changes
1.3.7
Patch Changes
1.3.6
Patch Changes
926117e Thanks @A-KGeorge ! - Corrected Typescript types to accept Float32Array for process methods
1.3.5
Patch Changes
1.3.4
Patch Changes
1.3.3
Patch Changes
1.3.2
Patch Changes
1.3.1
Patch Changes
1.3.0
Minor Changes
1.2.4
Patch Changes
00bab51 Thanks @A-KGeorge ! - pipelines of various configs can load state into each other (plug-n-play)
1.2.3
Patch Changes
1.1.6
Patch Changes
1.1.5
Patch Changes
1.1.4
Patch Changes
1.1.3
Patch Changes
1.1.1
Patch Changes
1.1.0
Minor Changes
1.0.1
Patch Changes
Fixed parks-mcclellan example file
Updated licensing documentation (FFTPACK public domain, Eigen MPL-2.0)
1.0.0
Major Changes
First stable release! 🎉
Parallel FFT batch processing with thread pool and LRU caching
Comprehensive DSP pipeline with 40+ processing stages
Native C++ acceleration with SIMD optimizations (AVX2, NEON)
Redis-backed state persistence for distributed processing
TypeScript/JavaScript API with full type safety
Cross-platform support: Windows, macOS (x64/ARM64), Linux
Features Included
Filters : FIR, IIR, Notch, Bandpass, Highpass, Lowpass, Bessel, Butterworth, Chebyshev
FFT : Forward/inverse FFT with batch processing and caching (100x speedup for repeated signals)
Time-Series : Irregular timestamp support, interpolation, resampling
Matrix Analysis : Covariance, eigendecomposition, whitening
Spatial Filters : Beamforming, CSP (Common Spatial Patterns), channel selection
Advanced : Wavelet transforms, Hilbert envelope, adaptive filters (LMS, NLMS, RLS)
Utilities : Moving average, variance, rectification, peak detection, envelope detection
1.0.0-alpha.18
Major Changes
0.2.0-alpha.17
Minor Changes
0.2.0-alpha.16
Minor Changes
0.2.0-alpha.15
Patch Changes
0.2.0-alpha.14
Patch Changes
0.2.0-alpha.13
Patch Changes
0.2.0-alpha.12
Patch Changes
0.2.0-alpha.11
Minor Changes
0.2.0-alpha.10
Minor Changes
08a1f50 Thanks @A-KGeorge ! - fixed performance for convolution, add differential filter/linear regression/dot product for ts, optimized for arm neon as well
0.2.0-alpha.9
Minor Changes
0.2.0-alpha.8
Minor Changes
0.2.0-alpha.7
Minor Changes
0.2.0-alpha.6
Patch Changes
0.2.0-alpha.5
Patch Changes
0.2.0-alpha.4
Patch Changes
0.2.0-alpha.3
Patch Changes
0.2.0-alpha.2
Patch Changes
0.2.0-alpha.1
Patch Changes
You can’t perform that action at this time.