DSP Filtering

Band-pass filters isolate a range of frequencies. Notch filters remove specific frequencies.
Together, they are fundamental tools in digital signal processing.

Scenario

A voice transmission has been corrupted. Broadband noise and unknown interference are making it impossible to understand. Use band-pass and notch filters to recover the voice signal.

Press Play, raise the volume, and study the frequency spectrum. What do you notice? Which parts of the spectrum contain useful information, and which parts are interference?

Frequency Spectrum Input (before filters)

What You Are Hearing

Press Play to begin. You will hear a corrupted signal. Your job is to recover the voice using filters.

Try this: Start by studying the frequency spectrum. Look for patterns. What looks like random noise? What looks like something deliberate?
Volume
Level 1%
Display
Show output
Spectrogram
Band-Pass Filter
Enabled
Notch Filters
No notch filters added
Click Play to begin

Vocabulary

Digital Signal Processing (DSP)

The use of computers or digital circuits to analyze, modify, and improve signals. DSP converts real-world analog signals (sound, light, sensor data) into numbers, processes those numbers using mathematical operations like filtering, and converts them back. It is used in audio engineering, telecommunications, medical imaging, radar, and countless other fields.

Signal

The information you want to keep. In this simulation, the signal is a voice transmission. In the real world, a signal could be a radio broadcast, a sensor reading, or any data you are trying to capture and process.

Noise

Unwanted energy spread across many frequencies that obscures the signal. Noise is random and has no pattern. Think of static on a radio or the hiss you hear between stations. Filtering helps separate noise from the signal you care about.

Frequency Spectrum

A graph showing how much energy exists at each frequency at a single moment in time. Tall bars mean more energy at that frequency. The spectrum lets you see what you cannot easily hear: exactly where the signal, noise, and interference live in the frequency range.

Spectrogram

A visual representation of how the frequency spectrum changes over time. Time moves along one axis, frequency along the other, and color or brightness indicates the energy level. A spectrogram lets you see patterns that a single spectrum snapshot cannot reveal, like how interference tones persist while noise fluctuates.

Band-Pass Filter

A filter that allows only a specific range of frequencies to pass through, rejecting everything above and below. You set the low cutoff and the high cutoff. Everything between those two points gets through; everything outside is attenuated. The center frequency is the midpoint of the passband, and the bandwidth is how wide that passband is.

Notch Filter

A filter that removes a very narrow slice of the frequency spectrum. It targets one specific frequency and suppresses it while leaving neighboring frequencies untouched. Notch filters are ideal for eliminating pure-tone interference.

Frequency (Hz)

The number of times a wave repeats per second, measured in Hertz. A 1,000 Hz tone completes 1,000 cycles each second. Low frequencies sound deep (bass), and high frequencies sound bright (treble). Human hearing ranges from roughly 20 Hz to 20,000 Hz.

Attenuation

The reduction in strength of a signal. When a filter attenuates a frequency, it makes that frequency quieter. A good filter provides strong attenuation outside its passband and minimal attenuation inside it.

Q Factor

A number that describes how sharp a filter's transition is. For a band-pass filter, low Q (like 0.707) gives a smooth, gentle rolloff. Higher Q creates a steeper, more resonant transition with a peak near the cutoff frequency. For a notch filter, high Q means a very narrow notch that precisely targets one frequency.

Based on Teaching DSP by Connor McKay, Christopher Light, and Ayoade Adewole (Public Domain).