True Peak Limiter
Broadcast-compliant brickwall limiting with inter-sample peak detection
What It Does
The true peak limiter prevents your audio from EVER exceeding the ceiling, no matter what. It's called a "brickwall" limiter because output simply cannot go above the ceiling—it hits a brick wall.
Unlike regular compression, the limiter uses lookahead to see peaks coming BEFORE they happen, then applies instant gain reduction with no overshoot. This is the last line of defense before your audio hits the DAC, ensuring it won't clip on consumer devices.
Use compliance presets for streaming/broadcast (Spotify, Apple Music, EBU R128). They automatically set ceiling and LUFS targets for each platform's requirements.
True Peak vs. Sample Peak
The Inter-Sample Peak Problem
Your audio is stored as discrete samples (e.g., 48,000 samples per second at 48kHz). When a DAC converts digital audio back to analog, it interpolates between samples to reconstruct the waveform. This reconstruction can create peaks BETWEEN samples that were never in the digital file.
Example:
- Sample 1: -0.5 dBFS
- Sample 2: -0.5 dBFS
- Reconstructed peak between them: +0.3 dBFS (clips!)
This is called an inter-sample peak. Your file shows -0.5 dBFS, but the DAC clips at +0.3 dBFS during playback. This causes distortion on consumer devices (phones, laptops, streaming boxes).
True Peak Detection Solves This
TabDSP oversamples your audio 4x (192kHz for 48kHz input), creating "virtual samples" between real samples. The limiter scans these virtual samples to find inter-sample peaks BEFORE they happen, then prevents them.
Why -1 dBTP is standard for streaming:
- Leaves headroom for inter-sample peaks
- Prevents clipping on consumer DACs
- Required by Spotify, Apple Music, YouTube, broadcast standards
▶ Gear Head Details: 4x Polyphase FIR Oversampling
TabDSP uses 4x polyphase FIR oversampling for ITU-R BS.1770-4 compliant true peak detection.
Why polyphase:
- Efficient: Processes only necessary sub-samples, not full 4x upsampled stream
- Each "phase" handles one sub-sample position (0, 0.25, 0.5, 0.75 between samples)
- 4 phases total, each with its own FIR filter coefficients
ITU Mode (Compliance):
- 48 total taps (12 taps per phase)
- Uses fixed ITU-R BS.1770-4 coefficients
- Optimized for broadcast standards compliance
- Guaranteed to match reference implementations
Kaiser Mode (Production):
- 128 total taps (32 taps per phase)
- Kaiser window with β=8 for ~90dB stopband attenuation
- More accurate interpolation, slightly higher CPU usage
- Better for mastering/production work
What true-peak detection catches: At the digital sample rate, a waveform's actual peak often lives between samples — the stored values can sit comfortably below 0 dBFS while the reconstructed analog waveform briefly overshoots and clips the converter downstream. Oversampling by 4× reconstructs those in-between values and measures the largest magnitude across them, giving a reading that matches what a DAC will actually produce. This is the behavior specified by ITU-R BS.1770-4, and the ITU vs. Kaiser filter choice trades a small amount of CPU for tighter stopband rejection when mastering for delivery.
Comparison to plugins:
- FabFilter Pro-L 2: Uses similar 4x oversampling, offers up to 16x
- Waves L2: Uses proprietary oversampling (not ITU-compliant)
- iZotope Ozone Maximizer: IRC algorithm with true peak detection
Limiter Controls
Ceiling
What it does: Maximum output level in dBTP (decibels True Peak). Audio cannot exceed this level.
Typical values:
- -1.0 dBTP: Standard for streaming (Spotify, Apple Music, YouTube)
- -0.2 dBTP: Safety ceiling for mastering (leaves tiny headroom)
- -2.0 dBTP: Broadcast (ATSC A/85, very conservative)
- 0.0 dBTP: Absolute maximum (risky, can still clip on some DACs)
Lookahead
What it does: Delay (in milliseconds) that lets the limiter "see into the future" and react BEFORE peaks arrive at the output.
How it works:
- Audio enters limiter and goes into a delay buffer (e.g., 5ms)
- Limiter scans ahead in buffer to find upcoming peaks
- Applies gain reduction NOW for peaks that will happen 5ms from now
- Result: Instant attack with zero overshoot
Typical values:
- 0.5-2ms: Minimal latency, fast transients may overshoot slightly
- 5ms (default): Balance of safety and latency
- 10ms: Maximum safety, catches everything, but adds noticeable delay
Release
What it does: How fast limiting gain reduction decreases after a peak passes.
Typical values:
- Fast (10-50ms): Responsive, can sound pumpy on heavy limiting
- Medium (80ms default): Natural, transparent for most material
- Slow (200-500ms): Smooth, glue-like, good for mix bus limiting
▶ Gear Head Details: Lookahead Buffer Implementation
TabDSP's limiter uses a ring buffer for efficient lookahead processing.
How lookahead shapes the response: The limiter delays its audio output by a few milliseconds so that, by the time any given sample reaches the output, the detector has already scanned slightly into its future. When a peak is spotted that would exceed the ceiling, gain reduction is in place before the peak arrives rather than chasing it afterward — there is no overshoot and no audible click from a late attack. Once the peak has passed, the applied reduction recovers smoothly back toward unity over the chosen release time, so sustained program material rides near the ceiling without pumping. The trade-off is the fixed output delay set by the lookahead time.
Why this works:
- Buffer acts as a time machine—we're processing "past" audio while scanning "future" audio
- Instant attack possible because we know peaks are coming
- Zero overshoot guaranteed—we literally cannot output a peak we haven't seen yet
Latency: Total latency = lookahead time + processing delay (~11ms at 5ms lookahead)
Compliance Presets
TabDSP includes 15 presets matching streaming, broadcast, and podcast standards. Select from the "Compliance" dropdown in the parameter panel. These presets set the ceiling so the limiter enforces the right true peak level—use the LUFS meter to monitor how your audio measures against the target.
Compliance presets set the ceiling but don't automatically adjust to target LUFS. Use Input Gain or Master Gain to adjust overall loudness, then watch the LUFS meter to hit your target.
▶ Show All Compliance Presets (Streaming, Broadcast, Podcast)
Streaming Presets
| Platform | Ceiling | Target LUFS | Notes |
|---|---|---|---|
| Spotify/YouTube/Tidal | -1.0 dBTP | -14 LUFS | Standard streaming loudness |
| Apple Music | -1.0 dBTP | -16 LUFS | Slightly quieter normalization |
| Loud Masters | -2.0 dBTP | -14 LUFS | For louder masters with extra headroom |
Video Presets
| Platform | Ceiling | Target LUFS | Notes |
|---|---|---|---|
| Netflix / Streaming Video | -2.0 dBTP | -27 LUFS | Netflix, Prime Video, Disney+ |
| HBO Max | -2.0 dBTP | -24 LUFS | HBO Max delivery specs |
Broadcast Presets
| Standard | Ceiling | Target LUFS | Region |
|---|---|---|---|
| EBU R128 | -1.0 dBTP | -23 LUFS | Europe (broadcast TV/radio) |
| ATSC A/85 | -2.0 dBTP | -24 LUFS | USA/Canada (broadcast TV) |
| ARIB TR-B32 | -1.0 dBTP | -24 LUFS | Japan (broadcast TV) |
| OP-59 | -2.0 dBTP | -24 LUFS | Australia (broadcast TV) |
| BBC Radio | -1.0 dBTP | -23 LUFS | BBC (EBU R128 compliant) |
Podcast Presets
| Platform | Ceiling | Target LUFS | Notes |
|---|---|---|---|
| Apple Podcasts | -1.0 dBTP | -16 LUFS | Recommended by Apple |
| Spotify Podcasts | -2.0 dBTP | -14 LUFS | Spotify podcast target |
| NPR/PRSS | -2.0 dBTP | -24 LUFS | US Public Radio standard |
Cinema Presets
| Format | Ceiling | Target LUFS | Notes |
|---|---|---|---|
| Theatrical / Cinema | 0.0 dBTP | N/A (full range) | No ceiling limit |
Using the Limiter Canvas
When the LIMIT module is selected, the canvas shows real-time stereo waveforms and gain reduction. The orange dashed ceiling line is draggable—grab it to adjust ceiling level while watching how it affects limiting in real time.
▶ Canvas Element Guide (Waveforms, Meters, Peak Toggle, Range Selector)
Canvas Elements
- 0 dBFS reference line: Red line at top with glow (digital maximum)
- Ceiling line: Orange dashed line (draggable) showing limiter ceiling
- Input waveforms: Blue (L) and indigo (R) showing pre-limiter levels
- Output waveforms: Green (semi-transparent) overlaid on input, showing post-limiter
- GR envelope: Red gradient fill from 0dBFS downward showing gain reduction
Stereo Meters (Right Side)
PRE section (before limiter):
- L/R level bars showing input levels
- Peak hold markers (1-second hold)
POST section (after limiter):
- L/R level bars showing output levels
- Orange ceiling marker shows where limiter kicks in
- Levels should never exceed ceiling
True Peak / Sample Peak Mode
Click the "True Peak" / "Sample Peak" button to cycle between detection modes:
- True Peak mode (default): Uses 4x oversampled inter-sample peak detection. Both sample peak (cyan bar) and true peak (orange bar) are visible on the limiter canvas.
- Sample Peak mode: Uses standard sample-level peak detection only.
In True Peak mode, if the orange bar is higher than the cyan bar, inter-sample peaks are present. This is normal—that's why true peak limiting exists.
dB Range Selector
Choose canvas zoom level:
- 3 dB: Detailed view, good for checking final headroom
- 6/9/12 dB: Moderate zoom
- 18 dB (default): Wide view showing full dynamics
- 24 dB: Maximum range, see everything
Parameter Reference
| Parameter | Range | Default | Description |
|---|---|---|---|
| Enabled | On/Off | On | Enable limiting |
| Ceiling | -12 to 0 dBTP | -1.0 dBTP | Maximum output level |
| Lookahead | 0.5 to 10 ms | 5.0 ms | Peak prediction time |
| Release | 5 to 1000 ms | 80 ms | Gain reduction recovery speed |
| True Peak | On/Off | On | Enable 4x oversampling |
| Filter Mode | Kaiser, ITU | Kaiser | Oversampling filter type |
| Peak Mode | True Peak / Sample Peak | True Peak | Peak detection mode (cycle button) |
| Range | 3/6/9/12/18/24 dB | 18 dB | Canvas display range |