CMSIS-DSP Python API
This is a Python wrapper for the Arm open source CMSIS-DSP and it is compatible with
The idea is to follow as closely as possible the C CMSIS-DSP API to ease the migration to the final implementation on a board.
The signal processing chain can thus be tested and developed in a Python environment and then easily converted to a C implementation running on a Cortex-M or Cortex-A board.
A tutorial is also available but with less details than this README: https://developer.arm.com/documentation/102463/latest/
This wrapper is also containing the scripts for the new CMSIS-DSP compute graph framework (CG).
CG is also including some nodes to communicate with Modelica using the VHT Modelica blocks developed as part of our VHT-SystemModeling demos.
An history of the changes to this wrapper is available at the end of the README.
How to build and install
The building of this package has been tested on Windows with the Python install from python.org and Microsoft Visual 2017.
It has also been tested with
cygwin. In that case,
python-devel must be installed too. On Mac, it was tested with standard XCode installation.
To run the examples,
matplotlib must also be installed.
Other configurations should work but the
setup.py file would have to be improved.
Python 3 must be used.
Installing and Building
It is advised to do it in a Python virtual environment. Then, in the virtual environment you can just do:
pip install cmsisdsp
You must have a recent
pip (to automatically install the dependencies like
NumPy) and you should have a compiler which can be found by Python when building the package.
DSP examples are available in the CMSIS-DSP PythonWrapper examples folder.
You can also install and run it from Google colab:
It it is not working (because it is not possible for us to test all configurations), you can then try to build and install the wrapper manually.
It is advised to do this it into a virtualenv
Since the CMSIS-DSP wrapper is using
NumPy, you must first install it in the virtual environment.
> pip install numpy
NumPy is installed, you can build the CMSIS-DSP python wrapper. Go to folder
Now, you can install the cmsisdsp package in editable mode:
> pip install -e .
Before using this command, you need to rebuild the CMSIS-DSP library which is no more built by the
There is a
CMakeLists.txt in the
PythonWrapper folder for this. The
build folders in
PythonWrapper are giving some examples of the options to use with the
cmake command to generate the
Makefile and build the library.
This library is then used by the
setup.py script to build the Python extension.
Running the examples
Install some packages to be able to run the examples
> pip install numpy > pip install scipy > pip install matplotlib
Depending on the example, you may have to install more packages.
The examples are in the CMSIS-DSP PythonWrapper examples folder.
You can test the scripts
example.py and try to run them from this virtual environment.
example.py is requiring a data file to be downloaded from the web. See below in this document for the link.
Note that due to the great number of possible configurations (OS, Compiler, Python), we can't give any support if you have problems compiling the
PythonWrapper on your specific configuration. But, generally people manage to do it and solve all the problems.
The idea is to follow as closely as possible the CMSIS-DSP API to ease the migration to the final implementation on a board.
First you need to import the module
> import cmsisdsp as dsp
If you use numpy:
> import numpy as np
If you use scipy signal processing functions:
> from scipy import signal
Functions with no instance arguments
You can use a CMSIS-DSP function with numpy arrays:
> r = dsp.arm_add_f32(np.array([1.,2,3]),np.array([4.,5,7]))
The function can also be called more simply with
> r = dsp.arm_add_f32([1.,2,3],[4.,5,7])
The result of a CMSIS-DSP function will always be a numpy array whatever the arguments were (numpy array or list).
Functions with instance arguments
When the CMSIS-DSP function is requiring an instance data structure, it is just a bit more complex to use it:
First you need to create this instance:
> firf32 = dsp.arm_fir_instance_f32()
Then, you need to call an init function:
The third argument in this function is the state. Since all arguments (except the instance ones) are read-only in this Python API, this state will never be changed ! It is just used to communicate the length of the state array which must be allocated by the init function. This argument is required because it is present in the CMSIS-DSP API and in the final C implementation you'll need to allocate a state array with the right dimension.
Since the goal is to be as close as possible to the C API, the API is forcing the use of this argument.
The only change compared to the C API is that the size variables (like blockSize for filter) are computed automatically from the other arguments. This choice was made to make it a bit easier the use of numpy array with the API.
Now, you can check that the instance was initialized correctly.
Then, you can filter with CMSIS-DSP:
The size of this signal should be
blockSize was inferred from the size of the state array :
numTaps + blockSize - 1 according to CMSIS-DSP. So here the signal must have 5 samples.
If you want to filter more than 5 samples, then you can just call the function again. The state variable inside firf32 will ensure that it works like in the CMSIS-DSP C code.
If you want to compare with scipy it is easy but warning : coefficients for the filter are in opposite order in scipy :
> filtered_x = signal.lfilter([3,2,1.], 1.0, [1,2,3,4,5,6,7,8,9,10]) > print(filtered_x)
The principles are the same for all other APIs.
Here is an example for using FFT from the Python interface:
Let's define a signal you will use for the FFT.
> nb = 16 > signal = np.cos(2 * np.pi * np.arange(nb) / nb)
The CMSIS-DSP cfft is requiring complex signals with a specific layout in memory.
To remain as close as possible to the C API, we are not using complex numbers in the wrapper. So a complex signal must be converted into a real one. The function imToReal1D is defined in testdsp.py
> signalR = imToReal1D(signal)
Then, you create the FFT instance with:
You initialize the instance with the init function provided by the wrapper:
> status=dsp.arm_cfft_init_f32(cfftf32, nb) > print(status)
You compute the FFT of the signal with:
> resultR = dsp.arm_cfft_f32(cfftf32,signalR,0,1)
You convert back to a complex format to compare with scipy:
> resultI = realToIm1D(resultR) > print(resultI)
For matrix, the instance variables are masked by the Python API. We decided that for matrix only there was no use for having the CMSIS-DSP instance visibles since they contain the same information as the numpy array (samples and dimension).
So to use a CMSIS-DSP matrix function, it is very simple:
> a=np.array([[1.,2,3,4],[5,6,7,8],[9,10,11,12]]) > b=np.array([[1.,2,3],[5.1,6,7],[9.1,10,11],[5,8,4]])
NumPy result as reference:
> print(np.dot(a , b))
> v=dsp.arm_mat_mult_f32(a,b) > print(v)
In a real C code, a pointer to a data structure for the result
v would have to be passed as argument of the function.
This example depends on a data file which can be downloaded here:
This signal was created for a master thesis:
Lugovaya T.S. Biometric human identification based on electrocardiogram. [Master's thesis] Faculty of Computing Technologies and Informatics, Electrotechnical University "LETI", Saint-Petersburg, Russian Federation; June 2005.
and it is part of the PhysioNet database
Goldberger AL, Amaral LAN, Glass L, Hausdorff JM, Ivanov PCh, Mark RG, Mietus JE, Moody GB, Peng C-K, Stanley HE. PhysioBank, PhysioToolkit, and PhysioNet: Components of a New Research Resource for Complex Physiologic Signals. Circulation 101(23):e215-e220 [Circulation Electronic Pages; http://circ.ahajournals.org/cgi/content/full/101/23/e215]; 2000 (June 13).
The Python wrapper is containing three submodules :
fixedpoint is proving some tools to help generating the fixedpoint values expected
mfcc is generating some tools to generate the MEL filters, DCT and window coefficients
expected by the CMSIS-DSP MFCC implementation.
MEL filters are represented as 3 arrays to encode a sparse array.
datatype is an API on top of
fixedpoint to provide more reuse when converting between data formats.
The wrapper is now containing the compute graph Python scripts and you should refer the the documentation in
DSP/ComputeGraph folder to know how to use those tools.
- Corrections to the RFFTs APIs
- More flexibility in the compute graph to specify the additional arguments of the scheduler and nodes
- Possibility to set the FIFO scaling factor at FIFO level (in asynchronous mode)
Same as 1.9.4 but will work in Google Colab.
- Dynamic Time Warping API
- Window functions for FFT
- New asynchronous mode for the compute graph (see compute graph documentation for more details.
- Corrected real FFTs in the wrapper
- Corrected arm_fir_decimate and arm_fir_interpolate
- Possibility to customize the FIFO class on a connection for the Python wrapper
- New customization options for the compute graph:
- Small fix to the compute graph generator. The
#ifdefat beginning of the custom header should be different for different scheduler names
addVariableArgin compute graph to use variable number of arguments
- New scheduling mode, in the compute graph generator, giving priority to sinks in the scheduling. The idea is to try to decrease the latency between sinks and sources.
- More customization options (Macros to be defined) in the C++ code generated by the compute graph generator
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Hashes for cmsisdsp-1.9.7-cp311-cp311-win_amd64.whl
Hashes for cmsisdsp-1.9.7-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Hashes for cmsisdsp-1.9.7-cp311-cp311-macosx_10_9_universal2.whl
Hashes for cmsisdsp-1.9.7-cp310-cp310-win_amd64.whl
Hashes for cmsisdsp-1.9.7-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Hashes for cmsisdsp-1.9.7-cp310-cp310-macosx_10_9_universal2.whl
Hashes for cmsisdsp-1.9.7-cp39-cp39-win_amd64.whl
Hashes for cmsisdsp-1.9.7-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Hashes for cmsisdsp-1.9.7-cp38-cp38-win_amd64.whl
Hashes for cmsisdsp-1.9.7-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Hashes for cmsisdsp-1.9.7-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl