dsp
A lightweight digital signal processing and spectrum analysis library for MoonBit
dsp
fft
filter
signal-processing
embedded
Author
Version
0.1.0
License
Apache-2.0
Repository
Last updated
2 months ago
Downloads
45
#MoonBit DSP
MoonBit-native digital signal processing for sensor, audio, control, and embedded/Wasm-oriented workloads. The library provides reusable numerical building blocks, streaming filters, spectral analysis, and domain-focused diagnostics without requiring a foreign runtime.
#What it provides
- Complex numbers, matrices, polynomials, Q15/Q31 fixed-point helpers, and special functions.
- Radix-2/Radix-4 FFT, Bluestein FFT, RFFT, DCT, CZT, STFT, 2D FFT, and spectrum metrics.
- Window functions with coherent-gain and ENBW helpers.
- IIR and FIR design/runtime filters, including Butterworth, Chebyshev, Bessel, biquad, Hilbert, and Remez components.
- Adaptive filters and trackers: LMS/NLMS, RLS, Kalman, EKF, UKF, particle filtering, and alpha-beta variants.
- Signal utilities such as correlation, resampling, wavelets, pitch, DTMF, peak detection, envelope extraction, Welch PSD, and zero-crossing analysis.
- Domain packages for audio, speech, image, radar, sonar, spatial audio, telecom, power, biomedical, vibration, and control processing.
- Fixed-capacity streaming primitives and a frame analyzer that reports RMS, peak, dominant frequency, and spectral centroid.
The implementation is intentionally split into focused packages so applications can depend on only the algorithms they use.
#Repository layout
moonbit-dsp/
├── moon.mod
├── LICENSE
├── README.md / README_CN.md
├── .github/workflows/test.yml
├── docs/benchmarks.md
├── examples/
└── src/
├── math/ window/ fft/
├── filter/{iir,iir_advanced,fir,adaptive,statistical}/
├── signal/ signal_ext/ spectral_estimation/
├── dsp_core/ dsp_utils/ streaming/
├── matrix_ext/ modal_analysis/ quantization/
├── audio_dsp/ bio_dsp/ control_dsp/ image_dsp/
├── power_dsp/ radar_dsp/ sonar_dsp/ spatial_audio/
├── speech_dsp/ telecom_dsp/ vibration_dsp/
└── dsp_bench/ dsp_benchmark/#Installation
The module name is hwlxmm/dsp. Add the packages you need to your consumer package's moon.pkg:
import {
"hwlxmm/dsp/src/filter/iir" @iir,
"hwlxmm/dsp/src/fft" @fft,
"hwlxmm/dsp/src/window" @window,
}moon.mod declares module metadata and external dependencies; package imports belong in moon.pkg.
#Minimal example
fn main {
let sample_rate = 1000.0
let cutoff = 50.0
let coeffs = @iir.butterworth_biquad_lowpass(cutoff, sample_rate)
let filter = @iir.BiquadCascade::new([coeffs])
let filtered = filter.process_sample(1.25)
println("filtered sample = \{filtered}")
}For spectral analysis, use @fft.rfft with a power-of-two input and a window generated by @window.WindowType.
#Examples
Run examples with the native backend:
moon run --target native examples/sensor_filter_demo
moon run --target native examples/fft_spectrum_demo
moon run --target native examples/kalman_tracker_demo
moon run --target native examples/spectrogram_demo
moon run --target native examples/ukf_tracker_demo
moon run --target native examples/vibration_monitor_demo
moon run --target native --release examples/benchmark_demoThe vibration demo combines time-domain indicators, an envelope spectrum, and the streaming frame analyzer. Benchmark methodology and measured local data are documented in docs/benchmarks.md.
#Quality checks
The repository CI checks all supported MoonBit targets on Linux, macOS, and Windows, and runs a native test job. The same checks can be run locally:
moon version --all
moon update
moon fmt --check
moon check --deny-warn
moon check --target all
moon test --deny-warn
moon test --deny-warn --target all
moon test --deny-warn --target native
moon infomoon info generates the package interface summaries (pkg.generated.mbti). Review those files when public APIs change; do not edit them by hand.
#Compatibility and scope
The primary development and verification backend is native. The library is also checked for MoonBit's wasm, wasm-gc, and js targets where the selected APIs support them. Native-only runtime behavior should be kept in executable examples or explicitly target-restricted packages.
Streaming APIs use fixed-capacity buffers for sample ingestion. Frame-level analysis creates owned arrays when a complete frame is emitted; this is distinct from claiming that every frame calculation is allocation-free.
#License
MoonBit DSP is distributed under the Apache License 2.0.
Install
Download zipA lightweight digital signal processing and spectrum analysis library for MoonBit
dspfftfiltersignal-processingembedded
Apache-2.0
Updated 2 months ago
45 downloads