Real-Time Audio Dynamic Range Compressor

This project implements a digital dynamic range compressor to reduce audio signal dynamic range by attenuating amplitudes above a user-defined threshold. Leveraging MATLAB, Pro Tools, and DSP algorithms, the system features a Butterworth pre-filter, logarithmic dB conversion, gain computation/smoothing, and makeup gain stages. Designed for music production and broadcasting applications, it addresses clipping prevention, volume normalization, and transient preservation while maintaining <1 dBFS output compliance.

Objectives

• Implement threshold-based compression with adjustable ratio (1:1 to 10:1)
• Design attack/release time controls (1ms–500ms) for smooth gain transitions
• Integrate 4th-order Butterworth filter to limit input to human-audible frequencies (20Hz–20kHz)
• Validate dBFS conversion accuracy and prevent output clipping via makeup gain
• Test using synthetic sine waves (520Hz pulsed at 140BPM) and live drum recordings

Project Process

1. Algorithm Design:
   • Developed core DSP blocks:
     • Butterworth Filter: 4th-order IIR for anti-aliasing
     • dB Converter: \( x_{dB}[n] = 20\log_{10}(x[n]) \) with dBFS scaling
     • Gain Computer: Threshold/Ratio logic (\( g_c = x_{sc} - x_{dB} \))
     • Gain Smoothing: Exponential averaging with α_A/α_R coefficients
   • Derived time constants: \( \alpha = \exp(-\ln(9)/(F_s \cdot T)) \)
2. Component Testing:
   • dB Block: Verified 0.5 → -6 dB conversion (Figure 1)
   • Gain Computer: Confirmed -6dB input → -12dB output at 10:1 ratio (Figure 2)
   • Gain Smoothing: Validated attack/release response using 520Hz pulsed sine (Figures 3-4)
   • Makeup Gain: Identified clipping issue during long attack/release (Figures 5-6)
3. System Integration:
   • Processed drum recordings (48kHz) with varying snare dynamics
   • Achieved 6dB LUFS loudness increase while preserving transient details
   • Reduced peak-to-average ratio by 63% in amplitude tests (Figure 8)

Conclusion and Future Improvements

The compressor successfully normalized audio levels with configurable thresholds and smooth gain adjustments, though makeup gain limitations required constrained attack/release settings. Future enhancements could implement soft-knee compression for gradual ratio transitions, sidechain triggering for multi-track mixing, and adaptive sampling rate modulation for broader compatibility. The modular algorithm structure provides a foundation for embedded DSP deployment in DAWs or hardware audio processors.