eDSP (easy Digital Signal Processing) is a digital signal processing framework written in modern C++ that implements some of the common functions and algorithms frequently used in digital signal processing, audio engineering & telecommunications systems.

It aims to provide efficient and streamlined base calculations while at the same time having a straightforward and easy-to-use interface similar to another numeric language like MATLAB. The library integrates a C++ and Python interfaces and supports Windows, Linux, Mac OS, iOS and Android.

Yet another DSP framework?

Got it! Excellent question! There are plenty of established DSP frameworks for C++. You may be wondering, What's the point of implementing other DSP framework? What does eDSP bring to the party that differentiates it from these?

eDSP is a modern framework written for modern C++ developers. It is an alternative for those who want a powerful template DSP library that keeps a simple STL-like API. eDSP is a framework that:

• Is implemented as a Headers only library, just copy and use.
• Is Template library: supports single and double floating point, and fixed-point operations.
• Integrates an easy to use interface, STL-like API.
• Provides fast implementations: check the available benchmarks.
• Aims to be cross-platform
• Supports modern C++11/14/17 features
• Integrates bindings for different languages: Python and JavaScript

What platforms are supported?

eDSP relies in a modern C++ compiler and some standard third-party dependencies. The idea is to build a framework compatible with most of the modern platforms with the minimum modifications.

Those are the supported platforms:

• Linux, FreeBSD, OpenBSD, Solaris, AIX
• Windows (msvc 2013+, cygwin)
• macOS (clang 3.5+), iOS
• Android

Features

eDSP is organized in different modules. Each module integrates different related features. This is the list of the available modules:

• Audio Coder: encoder/decoder for different audio formats.
• Digital filters: Biquad, FIR, IIR, moving average, moving median etc.
• Spectral analysis: Fourier transform, Hilbert transform, Hartley transform, cepstrum, periodogram etc.
• Oscillators: sinusoidal, square, triangular and sawtooth signals.
• Random: binary sequence, uniform distribution, Bernoulli, Weibull, Student, Fisher etc.
• Math: extra functionalities not available in the C++ standard.
• Noising: white noise, brown noise, pink noise, Perlin noise etc.
• Algorithms: clipper, amplifier, lineal-space, logarithmic-space , padder, normalizer etc.
• Quantizer: u-law, A-law etc.
• Statistics: moment, median, skewness, kurtosis etc.
• Windowing: Bartlett, Hamming, Kaiser, Welch, Blackman etc.
• Modulation: AM, FM, M-PSK, OFDM etc.
• MIR: feature engineering for MIR (Music Information Retrieval) applications.
• Equalization: LMS, RLS etc.

Note: eDSP is a framework under construction. Some features are scheduled for future development.

Please check the official docs or the list of available features to find a detailed information of each module.

Quick start

• How to install it?
• What's about the external dependencies?
• Benchmark: How fast is eDSP?
• MIR (Music Information Retrieval): feature engineering

Documentation

Check the wiki for more information. The latest development documentation is available online.

TODO

• [ ] Test the different features [WIP]
• [ ] Benchmark the library [WIP]
• [ ] Write Documentation [WIP]
• [ ] Python Wrapper
• [ ] Open-Frameworks Wrapper
• [x] Compile time optimizations: constexpr & traits
• [x] Modern C++: const method, noexcept construction, contract, assertions...

Check the Kanban of this project to get more information about the current state.

License

Copyright (c) 2018, Mohammed Boujemaoui Boulaghmoudi, All rights reserved.

Distributed under the GPLv3 license.