Digital Filters for Embedded Systems

// is this course for me?

Who this course is for

🎯

Target Audience

[Dan to fill in] — e.g. working engineers, embedded developers, graduate students who have completed a signals and systems course and want hands-on DSP experience.

📋

Prerequisites

[Dan to fill in] — e.g. undergraduate signals and systems course, basic familiarity with C, no RF license needed.

💻

What You'll Need

▸Computer: [Dan to fill in] — OS versions (Windows 10+, macOS 12+, Ubuntu 20.04+?)
▸Python 3.9+ with Jupyter Notebook
▸Audio source: [Dan to fill in] — e.g. line-in source, headphones, or use onboard mics
▸RadioSonic board (included in course bundle)

// what you get

Where filter theory meets
the hardware it runs on

Every concept in this course is grounded in math, implemented in Python, and then deployed to the RadioSonic board so you can hear it work.

⚡

Real-time on ESP32-S3

Push filter coefficients to the board over Wi-Fi and hear your design running at 48 kHz through the TLV320AIC3204 I²S audio codec in real time.

📓

Jupyter Notebook Labs

All filter design and analysis lives in interactive Jupyter notebooks using scipy, numpy, and matplotlib, with built-in frequency response and pole-zero plots.

🎯

Two Learning Paths

Choose Guided for weekly instructor sessions and graded feedback, or Self-Paced for lifetime access on your own schedule. Same content, same hardware labs.

// python + live hardware

Design a filter in Python.
Run it on real hardware.

Labs use Jupyter notebooks for all filter design, analysis, and visualization, then you push the same coefficients to the RadioSonic board over Wi-Fi and hear the result through the audio codec.

▸ scipy.signal : FIR/IIR design, frequency response, bode plots
▸ numpy : coefficient math, convolution, fixed-point rounding
▸ matplotlib : magnitude / phase / pole-zero plots
▸ Board SDK : push coefficients via Wi-Fi, capture mic audio
▸ C deployment : fixed-point biquad chains in the ESP-IDF firmware

lab_02_fir_design.ipynb

In [1]:
import numpy as np
from scipy import signal
import matplotlib.pyplot as plt
from radiosonic import Board, push_coeffs   # board SDK
In [2]:
# FIR lowpass, Hamming window method
fs  = 48_000   # sample rate Hz, TLV320AIC3204
fc  = 1_200    # cutoff Hz
N   = 63       # filter length (odd → linear phase)

h = signal.firwin(N, fc, fs=fs, window='hamming')
f, H = signal.freqz(h, fs=fs, worN=8192)
In [3]:
fig, ax = plt.subplots(figsize=(7, 3))
ax.plot(f, 20*np.log10(np.abs(H)+1e-12), '#D42B2B', lw=1.5)
ax.axvline(fc, color='#61afef', lw=1, ls='--')
ax.set(xlabel='Hz', ylabel='dB', xlim=[0,8000], ylim=[-80,5])
plt.tight_layout()
Out:

                
                  
                  
                  
                  
                  
                  
                  
                  
                  0
                  1.2k
                  4k
                  8k Hz
                  5
                  -20
                  -40
                  -60
                  -72
                  fc=1.2k
                
              
In [4]:
# Push to RadioSonic board over Wi-Fi
board = Board('radiosonic.local')
push_coeffs(board, h, channel='left')
print(f"✓ {len(h)}-tap FIR loaded, listen now")
Out:
✓ 63-tap FIR loaded, listen now

// course curriculum

A structured path from fundamentals
to real-time deployment

Each module pairs video lessons and Jupyter notebook labs with a hands-on board assignment. 5 modules, 15 lessons, 3 hardware lab projects.

Module 01

Hardware Bring-Up

Configure the RadioSonic ESP32-S3 board, bring up the TLV320AIC3204 I²S audio codec, verify the dual-mic array, and connect the Python SDK over Wi-Fi. By the end you're capturing and playing back audio in code.

ESP32-S3I²STLV320AIC3204Wi-FiESP-IDF

Module 02

DSP Foundations

Sampling theory, DFT, DTFT, convolution, and linear time-invariant systems. Build intuition for how digital filters shape signals in both the time and frequency domains before any design work begins.

DFT / FFTconvolutionLTI systemsz-transformDTFT

Module 03

FIR Filter Design

Window method, frequency sampling, and Parks-McClellan (equiripple) design. Design lowpass, bandpass, and highpass FIR filters with controlled ripple and transition width, then deploy coefficients to the board.

firwinParks-McClellanlinear phasewindowingequiripple

Module 04

IIR Filter Design

Butterworth, Chebyshev I/II, and elliptic prototypes with bilinear transform. Understand stability, pole-zero placement, fixed-point quantization effects, and second-order section (SOS) cascade implementations.

ButterworthChebyshevbilinearpole-zeroSOS

Module 05

Practical Design & Capstone

End-to-end projects: design filters to spec, optimize for fixed-point arithmetic, and deploy to the board for real-time audio processing. Includes a beamforming lab using the 6-mic expansion array for direction-of-arrival estimation.

fixed-pointreal-time audiobeamformingAoAC deployment

// your instructors

Meet the instructors

👤

[Dan to fill in: Instructor name]

[Title / Role]

[Dan to fill in: instructor bio — background, industry experience, why they teach this material.]

👤

[Dan to fill in: Instructor name]

[Title / Role]

[Dan to fill in: instructor bio — background, industry experience, why they teach this material.]

// enrollment

Choose your learning path

Same five modules, same Jupyter notebooks, same live hardware labs, two formats to fit how you learn.

GUIDED

Weekly cohort

Digital Filters: Guided

Live weekly sessions with the instructor, graded lab feedback, and a cohort of peers working through the material together. Ideal for structured accountability and direct Q&A.

▸ Weekly live instructor sessions
▸ Graded lab assignment feedback
▸ Cohort discussion forums
▸ All 5 modules + 3 lab projects
▸ Certificate of completion

EARLY ENROLLMENT

$249

REGULAR

$299

Enroll in Guided Course →

// enrollment key delivered by email after checkout

SELF-PACED

Lifetime access

Digital Filters: Self-Paced

All the same content, labs, and notebooks, on your own schedule. Work through each module whenever it suits you with no deadlines and no cohort timing.

▸ Flexible, learn at your own pace
▸ Community discussion forums
▸ All 5 modules + 3 lab projects
▸ Downloadable Jupyter notebooks
▸ Lifetime access to updates

EARLY ENROLLMENT

$99

REGULAR

$149

Enroll Self-Paced →

// enrollment key delivered by email after checkout

🔑

Already have an enrollment key?

Buy the RadioSonic board bundle on the hardware site to receive your key by email. Enter it when you enroll below.

Student Login →

Digital Filters forEmbedded Systems

Who this course is for

Target Audience

Prerequisites

What You'll Need

Where filter theory meetsthe hardware it runs on

Design a filter in Python.Run it on real hardware.

A structured path from fundamentalsto real-time deployment

Meet the instructors

Choose your learning path

Hear your filteron real hardware.

Digital Filters for
Embedded Systems

Where filter theory meets
the hardware it runs on

Design a filter in Python.
Run it on real hardware.

A structured path from fundamentals
to real-time deployment

Hear your filter
on real hardware.