Studio Dynamic Compressor — User Guide
Professional RMS-based dynamics processing: transparent level control with auto-calibration, 6 factory presets, and mathematical compression curves for broadcast-quality results.
What this does
This script implements a professional-grade dynamic compressor using RMS-based detection with automatic calibration for accurate level matching. The processor analyzes input levels, applies precise gain reduction based on mathematical compression curves, and automatically compensates with makeup gain. Key features include: (1) RMS detection: Uses Praat's Intensity object for perceptually accurate level measurement. (2) Auto-calibration: Automatically matches envelope levels to input signal for precise threshold operation. (3) 6 factory presets: Optimized settings for vocals, drums, mix bus, limiting, and creative effects. (4) Mathematical compression: Pure algorithmic processing without look-ahead artifacts. (5) Professional workflow: Automatic naming, peak normalization, and cleanup.
Key Features:
- RMS Detection — Perceptually accurate level measurement
- Auto-Calibration — Automatic envelope level matching
- 6 Factory Presets — Professional starting points
- Mathematical Compression — Precise gain reduction curves
- Time Constant Control — Unified attack/release window
- Makeup Gain — Automatic level compensation
What is dynamic compression? Dynamic range compression reduces the difference between loud and quiet sounds. Benefits: (1) Level control: Prevents peaks from dominating, brings up quiet details. (2) Density: Creates thicker, more present sounds. (3) Control: Tames unpredictable dynamics for consistent levels. (4) Creative effects: Can create pumping, breathing, or aggressive effects. (5) Broadcast readiness: Meets loudness standards for streaming and broadcast. This implementation uses RMS (Root Mean Square) detection which responds to average levels rather than instantaneous peaks, creating more musical and transparent compression.
Technical Implementation: (1) Input analysis: Measure peak level and convert to dBFS. (2) Envelope generation: Create RMS envelope using Praat's Intensity object with calculated frequency response. (3) Auto-calibration: Shift envelope to match input levels for accurate threshold operation. (4) Gain computation: Apply mathematical compression formula to generate gain reduction curve. (5) Makeup gain: Apply calculated gain compensation. (6) Signal application: Multiply audio by gain envelope for final compression. (7) Output management: Peak normalization, automatic naming, and cleanup.
Quick start
- In Praat, select exactly one Sound object.
- Run script… →
studio_dynamic_compressor.praat. - Choose Preset from dropdown:
- Vocal Leveler: Smooth vocal compression
- Drum Punch: Fast, aggressive drum control
- Mix Bus Glue: Gentle bus compression
- Hard Limiter: Peak control and protection
- Squash: Heavy creative compression
- NUKE: Extreme effect compression
- Or select Custom and adjust:
- Threshold_dB: Compression start point
- Ratio: Amount of compression
- Attack_Release_window: Time response
- Makeup_Gain_dB: Output level compensation
- Set Scale_peak (0.99 recommended)
- Enable Play_result for immediate audition
- Enable Keep_original to preserve source
- Click OK — processed sound created with descriptive suffix
Quick tip: Start with presets to understand different compression characters. For vocals: use Vocal Leveler for smooth, transparent control. For drums: use Drum Punch for aggressive, fast compression. For full mixes: use Mix Bus Glue for gentle cohesion. The auto-calibration ensures thresholds work correctly regardless of input levels. Makeup gain is automatically calculated for each preset — adjust manually if needed. Output is automatically normalized to prevent clipping.
Important: RMS DETECTION — compressor responds to average levels, not instantaneous peaks. Time constants: Very short windows (<0.01s) may cause distortion. Extreme ratios: High ratios (>10:1) create limiting behavior. Makeup gain: Can increase noise floor when compressing heavily. Auto-calibration: Requires measurable signal — very quiet sounds may not calibrate correctly. Peak normalization: Output scaled to prevent clipping — may be quieter than expected with heavy compression. Original preservation: "Keep_original=0" removes source sound — use with caution.
Compression Theory
Dynamic Range Fundamentals
What is Dynamic Range?
Audio level variation:
DYNAMIC RANGE = Difference between loudest and quietest parts
DR = 20 × log₁₀(peak_level / noise_floor)
Typical dynamic ranges:
Speech: 30-40 dB
Music: 50-60 dB (acoustic), 10-20 dB (heavily compressed)
Symphony: 80+ dB
Digital systems: 96 dB (16-bit), 144 dB (24-bit)
Compression purpose: Reduce dynamic range
Bring up quiet parts, control loud parts
Make audio more consistent in level
Prevent peaks from causing distortion
Create denser, more present sound
Our approach: RMS-based compression
Responds to average levels (like human hearing)
More musical than peak-based compression
Creates smoother, more transparent results
Compression Parameters
Key control parameters:
🎛️ Compression Controls
Threshold:
- Level where compression begins
- Below threshold: no effect
- Above threshold: gain reduction applied
- Typical range: -40 dB to 0 dB
Ratio:
- Amount of compression above threshold
- 2:1 = gentle, 4:1 = medium, 10:1 = heavy
- ∞:1 = limiting (no signal above threshold)
- Our range: 1.5:1 to 100:1
Time Constants:
- How fast compressor responds to level changes
- Fast: aggressive, punchy
- Slow: smooth, transparent
- Our implementation: Unified attack/release window
RMS vs Peak Detection
Detection Method Differences
RMS (Root Mean Square):
RMS = Square root of mean of squares
RMS = √( (x₁² + x₂² + ... + xₙ²) / n )
Properties:
- Responds to average power (like human loudness perception)
- Ignores brief peaks
- Creates smoother, more musical compression
- Less prone to over-compressing transients
PEAK DETECTION:
- Responds to instantaneous maximum levels
- Catches all peaks, even very brief ones
- Can sound aggressive or "grabby"
- Good for peak limiting and protection
Our choice: RMS detection
- More musical for general compression
- Better matches human loudness perception
- Creates smoother gain reduction
- Standard for vocal and mix bus compression
Perceptual Advantages
Why RMS sounds more natural:
RMS vs Peak Detection Comparison:
Scenario: Drum hit with sharp transient
Peak detection: Sees transient peak → heavy compression
RMS detection: Sees average level → moderate compression
Result:
Peak: Transient suppressed, "flat" sound
RMS: Transient preserved, natural dynamics
Human hearing:
We perceive loudness based on average power, not peaks
RMS matches our perceptual experience
Creates more transparent, musical compression
Exception: Peak limiting for protection still needed
Scenario: Drum hit with sharp transient
Peak detection: Sees transient peak → heavy compression
RMS detection: Sees average level → moderate compression
Result:
Peak: Transient suppressed, "flat" sound
RMS: Transient preserved, natural dynamics
Human hearing:
We perceive loudness based on average power, not peaks
RMS matches our perceptual experience
Creates more transparent, musical compression
Exception: Peak limiting for protection still needed
Compression Mathematics
Gain Reduction Formula
Mathematical compression curve:
COMPRESSION FORMULA:
gain_reduction = 10^( ( -1 × (level - threshold) × (1 - 1/ratio) ) / 20 )
Where:
level = input level in dB
threshold = compression threshold in dB
ratio = compression ratio (e.g., 4.0 for 4:1)
Step-by-step derivation:
1. Calculate overshoot above threshold:
overshoot = level - threshold
2. Apply ratio to overshoot:
compressed_overshoot = overshoot / ratio
3. Calculate output level:
output_level = threshold + compressed_overshoot
4. Convert to gain reduction:
gain_reduction_db = level - output_level
= overshoot - (overshoot / ratio)
= overshoot × (1 - 1/ratio)
5. Convert to linear gain:
gain_reduction_linear = 10^( -gain_reduction_db / 20 )
Final formula matches professional compressors
Compression Curve Visualization
Compression Curves for Different Ratios
Input/Output Transfer Characteristics:
Threshold = -20 dB, makeup gain applied for visualization
Input/Output Transfer Characteristics:
Ratio 1:1 | 2:1 | 4:1 | 10:1 | ∞:1
-40dB → -40 | -40 | -40 | -40 | -40
-30dB → -30 | -30 | -30 | -30 | -30
-20dB → -20 | -20 | -20 | -20 | -20
-10dB → -10 | -15 | -17.5| -19 | -20
0dB → 0 | -10 | -15 | -18 | -20
Threshold = -20 dB, makeup gain applied for visualization
Processing Algorithm
Auto-Calibration System
Input Level Measurement
Precise level matching:
STEP 1: Measure input peak
selectObject: sound
in_max = Get maximum: 0, 0, "Sinc70"
in_db = 20 × log10(abs(in_max) + 0.000001)
Why Sinc70 interpolation?
- More accurate peak detection than simple maximum
- Accounts for inter-sample peaks
- Professional-grade peak measurement
- Prevents underestimation of true peaks
STEP 2: Generate RMS envelope
detect_freq = 3.2 / attack_Release_window
intensity = To Intensity: detect_freq, 0, "yes"
Why this frequency calculation?
- attack_Release_window determines time response
- Lower frequency = slower response
- 3.2 factor provides good time constant matching
- Minimum 10Hz ensures reasonable response
STEP 3: Auto-calibration
env_max = Get maximum: 0, 0, "Parabolic"
offset = in_db - env_max
Formula: "self + offset"
Result: Envelope accurately matches input levels
Threshold settings work as expected
No manual level adjustment needed
Professional workflow
Time Constant Calculation
Attack/Release Response:
UNIFIED TIME WINDOW:
Our implementation uses single window for both attack and release
attack_Release_window determines overall response speed
Frequency calculation:
detect_freq = 3.2 / attack_Release_window
Why 3.2?
- Empirical testing showed this provides good response matching
- Creates smooth, musical compression characteristics
- Matches expected attack/release behavior
Time constant examples:
Window=0.05s → Frequency=64Hz → Medium response
Window=0.02s → Frequency=160Hz → Fast response
Window=0.10s → Frequency=32Hz → Slow response
Practical ranges:
Fast compression: 0.005-0.02s (drums, percussion)
Medium compression: 0.03-0.08s (vocals, general)
Slow compression: 0.09-0.20s (mix bus, mastering)
Gain Computation
Mathematical Compression
Precise gain reduction:
STEP 1: Convert to control matrix
control_mat = Down to Matrix
STEP 2: Apply compression formula
Formula: "if self > t then 10^((-1 * (self - t) * (1 - 1/r)) / 20) else 1 fi"
Where:
t = threshold_dB
r = ratio
Explanation:
- "self > t": Only compress above threshold
- "(self - t)": Amount above threshold
- "(1 - 1/r)": Compression factor based on ratio
- "/ 20": Convert dB to linear gain
- "10^(...)": dB to linear conversion
STEP 3: Apply makeup gain
Formula: "self * 10^(mkp/20)"
Where:
mkp = makeup_Gain_dB
Result: Complete gain reduction envelope
Ready for application to audio signal
Makeup Gain Strategy
Level compensation:
Makeup Gain Principles:
Purpose: Compensate for level loss from compression
Without makeup gain: Compressed signal is quieter
Automatic calculation in presets:
Vocal: +4dB (moderate compression)
Drum: +3dB (fast, aggressive)
Mix Bus: +1dB (gentle, transparent)
Limiter: 0dB (peak control only)
Squash: +8dB (heavy compression)
NUKE: +12dB (extreme effect)
Manual adjustment:
- Listen for consistent perceived loudness
- Avoid excessive gain that causes noise
- Match to target loudness standards if needed
Peak normalization: Final safety net to prevent clipping
Purpose: Compensate for level loss from compression
Without makeup gain: Compressed signal is quieter
Automatic calculation in presets:
Vocal: +4dB (moderate compression)
Drum: +3dB (fast, aggressive)
Mix Bus: +1dB (gentle, transparent)
Limiter: 0dB (peak control only)
Squash: +8dB (heavy compression)
NUKE: +12dB (extreme effect)
Manual adjustment:
- Listen for consistent perceived loudness
- Avoid excessive gain that causes noise
- Match to target loudness standards if needed
Peak normalization: Final safety net to prevent clipping
Signal Application
Final Processing Stage
Gain envelope application:
STEP 1: Convert gain to audio
selectObject: control_mat
control_sig = To Sound
Rename: "Control_Voltage"
STEP 2: Apply to audio
selectObject: sound
compressed = Copy: original_name$ + "_Comp"
Formula: "self * object(""Sound Control_Voltage"", x)"
Why object reference?
- Praat's object() function allows referencing other sounds
- "Control_Voltage" is the gain envelope
- x represents time coordinate
- Real-time multiplication applies gain reduction
STEP 3: Final processing
Scale peak: scale_peak
Rename: original_name$ + suffix$
Cleanup:
removeObject: intensity, control_mat, control_sig
Result: Professional compressed audio
Ready for use in productions
Complete Processing Pipeline
PHASE 1: SETUP & ANALYSIS
Validate selection (exactly 1 sound)
Apply preset parameters or use custom
Measure input peak level (Sinc70 interpolation)
Calculate detection frequency from time window
PHASE 2: ENVELOPE GENERATION
Create RMS envelope (Intensity object)
Auto-calibrate envelope to match input levels
Convert envelope to control matrix
PHASE 3: GAIN COMPUTATION
Apply compression formula to control matrix
Apply makeup gain compensation
Convert gain matrix back to audio envelope
PHASE 4: SIGNAL PROCESSING
Apply gain envelope to audio signal
Scale peak to prevent clipping
Apply descriptive naming suffix
PHASE 5: CLEANUP & OUTPUT
Remove temporary objects
Optional: Remove original sound
Optional: Play result
Select final compressed sound
OUTPUT: Professional compressed audio
All temporary objects cleaned up
Consistent naming convention
Peak-normalized for safety
Factory Presets
Vocal Leveler (Smooth)
| Parameter | Value | Rationale |
|---|---|---|
| Threshold | -24.0 dB | Catches most vocal phrases |
| Ratio | 2.5:1 | Gentle, transparent compression |
| Time Window | 0.08 s | Smooth, musical response |
| Makeup Gain | +4.0 dB | Compensates for level reduction |
| Suffix | _Vocal | Clear identification |
🎤 Vocal Compression Characteristics
Best for: Lead vocals, spoken word, vocal ensembles
Sonic character: Smooth, transparent, controlled
Technical notes: Gentle ratio prevents "pumping", slow attack preserves transients, smooth release maintains natural decay
Drum Punch (Fast & Hard)
| Parameter | Value | Rationale |
|---|---|---|
| Threshold | -18.0 dB | Catches drum peaks |
| Ratio | 6.0:1 | Aggressive control |
| Time Window | 0.02 s | Fast, punchy response |
| Makeup Gain | +3.0 dB | Brings up body and sustain |
| Suffix | _Drum | Clear identification |
🥁 Drum Compression Characteristics
Best for: Drum buses, individual drums, percussion
Sonic character: Punchy, aggressive, controlled
Technical notes: Fast attack controls transients, high ratio adds density, makeup gain emphasizes body and sustain
Mix Bus Glue (Gentle)
| Parameter | Value | Rationale |
|---|---|---|
| Threshold | -12.0 dB | Only catches peaks |
| Ratio | 1.5:1 | Very gentle compression |
| Time Window | 0.10 s | Slow, transparent response |
| Makeup Gain | +1.0 dB | Minimal compensation |
| Suffix | _Bus | Clear identification |
🎚️ Mix Bus Compression Characteristics
Best for: Stereo mixes, subgroup buses, mastering
Sonic character: Cohesive, transparent, "glued"
Technical notes: Very gentle ratio for transparency, slow timing for smooth operation, minimal makeup gain preserves dynamics
Hard Limiter (Peak Control)
| Parameter | Value | Rationale |
|---|---|---|
| Threshold | -6.0 dB | Catches only extreme peaks |
| Ratio | 20.0:1 | Effectively limiting |
| Time Window | 0.005 s | Very fast response |
| Makeup Gain | 0.0 dB | Pure peak control |
| Suffix | _Lim | Clear identification |
Squash (Heavy Effect)
| Parameter | Value | Rationale |
|---|---|---|
| Threshold | -30.0 dB | Compresses most of signal |
| Ratio | 10.0:1 | Heavy compression |
| Time Window | 0.05 s | Medium response |
| Makeup Gain | +8.0 dB | Significant compensation |
| Suffix | _Squash | Clear identification |
NUKE (Very Aggressive)
| Parameter | Value | Rationale |
|---|---|---|
| Threshold | -40.0 dB | Compresses everything |
| Ratio | 100.0:1 | Extreme compression |
| Time Window | 0.015 s | Fast, aggressive response |
| Makeup Gain | +12.0 dB | Maximum compensation |
| Suffix | _NUKE | Clear identification |
Preset Selection Guide:
- Vocal Leveler: Everyday vocal compression
- Drum Punch: Aggressive drum control
- Mix Bus Glue: Subtle mix cohesion
- Hard Limiter: Peak protection only
- Squash: Creative heavy compression
- NUKE: Extreme effect processing
Parameters & Settings
Compression Parameters
| Parameter | Range | Default | Description |
|---|---|---|---|
| Threshold_dB | -60 to 0 | -20.0 | Level where compression begins |
| Ratio | 1.0-100.0 | 4.0 | Compression amount above threshold |
Time Parameters
| Parameter | Range | Default | Description |
|---|---|---|---|
| Attack_Release_window | 0.001-1.0 | 0.05 | Combined attack/release time |
Output Parameters
| Parameter | Range | Default | Description |
|---|---|---|---|
| Makeup_Gain_dB | -20 to +20 | 2.0 | Output level compensation |
| Scale_peak | 0.1-1.0 | 0.99 | Final peak normalization |
| Play_result | boolean | 1 | Auto-play after processing |
| Keep_original | boolean | 1 | Preserve source sound |
Parameter Guidance
Threshold selection:
- -40 to -30 dB: Heavy compression (most of signal)
- -30 to -20 dB: Medium compression
- -20 to -10 dB: Light compression (only peaks)
- -10 to 0 dB: Limiting (extreme peaks only)
Ratio selection:
- 1.5:1 to 2:1: Very gentle, transparent
- 2:1 to 4:1: Moderate, musical
- 4:1 to 8:1: Heavy, obvious compression
- 8:1 to 20:1: Very heavy, limiting territory
- 20:1+: Hard limiting
Time window selection:
- 0.005-0.02s: Very fast (drums, percussion)
- 0.02-0.05s: Fast (vocals, bass)
- 0.05-0.10s: Medium (general purpose)
- 0.10-0.20s: Slow (mix bus, mastering)
- 0.20s+: Very slow (special effects)
Applications
Vocal Processing
Use case: Consistent vocal levels in mixes
Technique: Use Vocal Leveler preset with medium threshold
Benefits: Smooth level control, preserved expression, reduced manual automation
Drum Enhancement
Use case: Adding punch and density to drums
Technique: Use Drum Punch preset on drum buses or individual drums
Results: Controlled transients, enhanced sustain, cohesive drum sound
Mix Bus Processing
Use case: Creating cohesive final mixes
Technique: Use Mix Bus Glue preset on stereo mix
Benefits: Gentle cohesion, controlled peaks, "glued" sound
Creative Effects
Use case: Extreme compression for artistic effect
Technique: Use Squash or NUKE presets on various sources
Results: Pumping, breathing, aggressive textures
Practical Workflow Examples
🎤 Lead Vocal Compression
Goal: Smooth, consistent vocal level without artifacts
Setup:
- Preset: Vocal Leveler
- Source: Lead vocal recording
- Adjustment: Lower threshold if more compression needed
- Result: Professional vocal sound
Tip: Compare with/without compression to ensure transparency
🥁 Drum Bus Punch
Goal: Aggressive, punchy drum sound
Setup:
- Preset: Drum Punch
- Source: Drum bus or individual drum
- Adjustment: Increase ratio for more aggression
- Result: Powerful, controlled drums
Tip: Use parallel compression for extreme effects
🎚️ Mix Bus Glue
Goal: Cohesive final mix
Setup:
- Preset: Mix Bus Glue
- Source: Stereo mix
- Adjustment: Very subtle - listen for "glue" not obvious compression
- Result: Professional, cohesive mix
Tip: A/B compare to ensure transparency
Advanced Techniques
Parallel compression:
- Method: Process copy with heavy compression, mix with dry signal
- Benefits: Density without losing transients
- Presets: Use Squash or NUKE for parallel processing
- Result: Best of both worlds - dynamics and density
Multi-stage compression:
- Stage 1: Gentle compression with Vocal Leveler
- Stage 2: Aggressive compression with Drum Punch
- Stage 3: Final control with Hard Limiter
- Result: Complex, professional compression chain
Troubleshooting Common Issues
Problem: Pumping or breathing artifacts
Cause: Too fast release time or too low threshold
Solution: Increase time window, raise threshold, or reduce ratio
Cause: Too fast release time or too low threshold
Solution: Increase time window, raise threshold, or reduce ratio
Problem: Lost transients
Cause: Too fast attack time or too low threshold
Solution: Increase time window, raise threshold, or use parallel compression
Cause: Too fast attack time or too low threshold
Solution: Increase time window, raise threshold, or use parallel compression
Problem: Noise becomes obvious
Cause: Too much makeup gain with heavy compression
Solution: Reduce makeup gain, use less compression, or address noise at source
Cause: Too much makeup gain with heavy compression
Solution: Reduce makeup gain, use less compression, or address noise at source
Problem: No audible compression
Cause: Threshold too high or ratio too low
Solution: Lower threshold, increase ratio, or check auto-calibration
Cause: Threshold too high or ratio too low
Solution: Lower threshold, increase ratio, or check auto-calibration
Algorithmic Extensions
Advanced Detection Methods
Multi-band Compression
Frequency-dependent compression:
MULTI-BAND APPROACH:
Split into frequency bands (e.g., low, mid, high)
Apply separate compression to each band
Recombine bands for final output
Benefits:
- Compress bass without affecting vocals
- Control sibilance without affecting body
- More surgical, transparent compression
- Prevents "pumping" from low frequencies
Implementation:
Use Praat's Filtering to create bands
Process each band with current compressor
Sum bands for final output
Challenges:
- Phase issues at crossover frequencies
- More computationally intensive
- More parameters to control
Advanced Time Constants
Separate Attack/Release
Independent time control:
SEPARATE ATTACK/RELEASE:
Current: Unified attack_release_window
Advanced: Separate attack_time and release_time
Implementation options:
OPTION 1: Envelope follower with separate times
attack_coeff = exp(-1/(attack_time × sample_rate))
release_coeff = exp(-1/(release_time × sample_rate))
Different coefficients for rising vs falling levels
OPTION 2: Dual RMS detectors
Fast detector for attack
Slow detector for release
Blend based on level changes
OPTION 3: Look-ahead processing
Buffer audio to "see" future peaks
Smooth attack without affecting transients
More computationally intensive
Benefits: More precise control over compression character
Advanced Metering
Comprehensive Level Analysis
Professional monitoring:
ADVANCED METERING FEATURES:
INPUT/OUTPUT COMPARISON:
Display both input and output levels
Show gain reduction in real-time
Visual feedback for parameter adjustment
LOUDNESS METERING:
Integrated LUFS (Loudness Units Full Scale)
Short-term and integrated loudness
True-peak detection
Professional broadcast standards
HISTOGRAM ANALYSIS:
Level distribution over time
Identify consistent problems
Visual representation of dynamics
GAIN REDUCTION HISTORY:
Plot gain reduction over time
Identify pumping or breathing
Optimize time constants
Benefits: Professional workflow, accurate settings, standards compliance