Build an Audio Spectrum Analyzer with Arduino and LEDs

Build an Audio Spectrum Analyzer with Arduino and LEDs

An audio spectrum analyzer with Arduino is one of the most visually impressive electronics projects you can build. By analyzing audio frequencies in real-time and displaying them as LED bar graphs, you create a professional-looking VU meter that responds to music, speech, or any sound source. This guide walks through building a complete spectrum analyzer using Arduino, an electret microphone, and an LED matrix — all readily available in India for under ₹800.

How FFT Audio Analysis Works

A spectrum analyzer works by decomposing an audio signal into its constituent frequencies using Fast Fourier Transform (FFT). The human hearing range spans 20Hz–20kHz, and a spectrum analyzer displays the amplitude (loudness) at different frequency bands simultaneously — typically split into 8, 16, or 32 bands.

On Arduino Uno (16MHz, 10-bit ADC), we can sample audio at approximately 9,600 samples/second, giving a Nyquist frequency of 4,800Hz. This covers the most musically important range (bass 20-300Hz, midrange 300Hz-4kHz, presence 4-8kHz). Using the ArduinoFFT library, a 64-point FFT executes in approximately 2ms — fast enough for responsive LED updates at 30 fps.

Components Required

• Arduino Uno or Nano (₹250-350)
• Electret microphone module with MAX4466 or LM393 (₹80-150)
• LED bar graph (10-segment) × 4 pieces, or individual LEDs × 40 (₹120-200)
• 74HC595 shift registers × 5 (for 8-column display) (₹40-80)
• Resistors: 220Ω × 40 (for LED current limiting) (₹20)
• 100µF and 100nF capacitors (₹10)
• 3.5mm audio jack + cable (₹30-50)
• Breadboard or PCB (₹50-100)
• Total estimated cost: ₹600-950

Circuit Schematic

The microphone module connects to Arduino’s analog input. Power the module from 3.3V (not 5V) for lower noise floor. The 74HC595 shift registers chain together to drive multiple LED columns with only 3 Arduino pins.

/* Circuit Connections:
 *
 * MICROPHONE MODULE:
 * VCC → Arduino 3.3V
 * GND → Arduino GND
 * OUT → Arduino A0 (via 10K pull-down to GND)
 *
 * 74HC595 CHAIN (for 8 LED columns):
 * Arduino Pin 11 → DS (Data in of first 595)
 * Arduino Pin 12 → SHCP (Clock) — all 595s in parallel
 * Arduino Pin 8  → STCP (Latch) — all 595s in parallel
 * QA-QH → LED anodes (via 220Ω resistors)
 * Next 595: DS connected to Q7S (serial output) of previous
 *
 * LED ROWS (common cathodes):
 * Arduino Pins 2-9 → 10 LED row cathodes (via 470Ω)
 */

Arduino FFT Code

// Audio Spectrum Analyzer with Arduino FFT
// Library: arduinoFFT (install via Library Manager)
// Install: Sketch → Include Library → Manage Libraries → Search "arduinoFFT"

#include "arduinoFFT.h"

#define SAMPLES 64          // Must be power of 2
#define SAMPLING_FREQ 9615  // Max for Arduino Uno (depends on ADC speed)
#define MIC_PIN A0
#define NUM_BANDS 8         // Number of frequency bands to display

double vReal[SAMPLES];
double vImag[SAMPLES];
ArduinoFFT FFT = ArduinoFFT(vReal, vImag, SAMPLES, SAMPLING_FREQ);

// Shift register pins
#define DATA_PIN  11
#define CLOCK_PIN 12
#define LATCH_PIN 8

// LED row pins (columns 1-8 = frequency bands)
int rowPins[8] = {2, 3, 4, 5, 6, 7, 9, 10};
int bandValues[8] = {0};  // Current amplitude for each band
int peakValues[8] = {0};  // Peak hold values

void setup() {
  Serial.begin(115200);
  
  // Setup shift register pins
  pinMode(DATA_PIN, OUTPUT);
  pinMode(CLOCK_PIN, OUTPUT);
  pinMode(LATCH_PIN, OUTPUT);
  
  // Setup LED row pins
  for (int i = 0; i < 8; i++) {
    pinMode(rowPins[i], OUTPUT);
    digitalWrite(rowPins[i], HIGH); // Common cathode: HIGH = OFF
  }
  
  Serial.println("Audio Spectrum Analyzer Ready");
}

void loop() {
  // Sample audio at precise intervals
  unsigned int sampling_period_us = round(1000000.0 / SAMPLING_FREQ);
  unsigned long microseconds = micros();
  
  for (int i = 0; i < SAMPLES; i++) {
    vReal[i] = analogRead(MIC_PIN) - 512; // Center around zero
    vImag[i] = 0;
    while (micros() - microseconds < sampling_period_us) {} // Wait
    microseconds += sampling_period_us;
  }
  
  // Apply FFT
  FFT.Windowing(FFT_WIN_TYP_HAMMING, FFT_FORWARD);
  FFT.Compute(FFT_FORWARD);
  FFT.ComplexToMagnitude();
  
  // Map FFT bins to 8 frequency bands
  // Bins 0-1: DC and very low (skip)
  // Bands: Sub-bass, Bass, Low-Mid, Mid, Upper-Mid, Presence, Brilliance, Air
  int bandBins[9] = {2, 3, 5, 8, 13, 21, 28, 37, 48}; // Logarithmic spacing
  
  for (int band = 0; band < NUM_BANDS; band++) {
    double bandAmp = 0;
    int binCount = bandBins[band+1] - bandBins[band];
    
    for (int bin = bandBins[band]; bin = peakValues[band]) {
      peakValues[band] = bandValues[band];
    } else if (peakValues[band] > 0) {
      peakValues[band]--; // Slow decay
    }
  }
  
  // Update LED display
  updateDisplay();
}

void updateDisplay() {
  // Multiplex through 8 rows
  for (int row = 0; row < 8; row++) {
    // Build column data for this row
    byte colData = 0;
    for (int band = 0; band  row) {
        colData |= (1 << band);
      }
      // Peak indicator
      if (peakValues[band] == row + 1) {
        colData |= (1 << band);
      }
    }
    
    // Activate row (common cathode - LOW = active)
    for (int r = 0; r < 8; r++) {
      digitalWrite(rowPins[r], (r == row) ? LOW : HIGH);
    }
    
    // Shift out column data
    digitalWrite(LATCH_PIN, LOW);
    shiftOut(DATA_PIN, CLOCK_PIN, MSBFIRST, colData);
    digitalWrite(LATCH_PIN, HIGH);
    
    delayMicroseconds(500); // Display each row briefly
  }
}

LED Matrix Display

For a cleaner build, use a MAX7219-based 8×8 LED matrix module (₹120-180) instead of individual LEDs and shift registers. The MAX7219 handles all multiplexing internally and connects to Arduino via SPI (3 wires). The LedControl library makes programming straightforward.

// MAX7219 LED Matrix version (simpler wiring)
#include <LedControl.h>
// LedControl(DIN, CLK, CS, numDevices)
LedControl lc = LedControl(11, 13, 10, 1);

void setupMatrix() {
  lc.shutdown(0, false);  // Wake up MAX7219
  lc.setIntensity(0, 8);  // Brightness 0-15
  lc.clearDisplay(0);
}

void displayBand(int band, int value) {
  // band: 0-7 (column), value: 0-8 (height)
  for (int row = 0; row < 8; row++) {
    lc.setLed(0, 7-row, band, row < value); // Bottom-up filling
  }
}

Advanced: MSGEQ7 for Better Performance

The MSGEQ7 is a dedicated 7-band graphic equalizer IC that performs analog spectrum analysis in hardware. It outputs 7 frequency bands (63Hz, 160Hz, 400Hz, 1kHz, 2.5kHz, 6.25kHz, 16kHz) via a multiplexed analog output. This is much faster and more accurate than Arduino software FFT, allowing 60fps updates and freeing Arduino resources for display work.

// MSGEQ7 Spectrum Analyzer (faster, dedicated IC)
#define STROBE_PIN 4
#define RESET_PIN  5
#define AUDIO_PIN  A0

int spectrum[7]; // 7-band output

void readMSGEQ7() {
  digitalWrite(RESET_PIN, HIGH);
  delayMicroseconds(1);
  digitalWrite(RESET_PIN, LOW);
  
  for (int band = 0; band < 7; band++) {
    digitalWrite(STROBE_PIN, LOW);
    delayMicroseconds(36); // MSGEQ7 settling time
    spectrum[band] = analogRead(AUDIO_PIN); // 0-1023
    digitalWrite(STROBE_PIN, HIGH);
  }
  // spectrum[0] = 63Hz (sub-bass)
  // spectrum[3] = 1kHz (midrange)
  // spectrum[6] = 16kHz (air frequencies)
}

Enclosure and Finishing

For a professional look popular in Indian electronics fairs and school projects:

• Use a transparent acrylic sheet (2-3mm) for the LED face panel — available at plastic shops in metro cities for ₹50-100
• Colored transparent acrylic for color-coded frequency bands (red=bass, green=mid, blue=treble)
• Alternatively, use frosted film over clear LEDs for diffused glow effect
• 3D-printed enclosure: design in Tinkercad (free), print at local makerspaces (₹100-200)
• Add a 3.5mm audio jack input alongside the microphone for direct line-in from phone/computer

Frequently Asked Questions