True Peak

In-Depth Explanation

True peak is the maximum level of an audio signal measured using oversampling to detect inter-sample peaks that occur between digital samples. Standard sample-peak meters only measure the value of individual digital samples and can miss peaks that exist in the reconstructed analog waveform between those samples. True peak measurement, defined by the ITU-R BS.1770 standard, catches these hidden peaks and is measured in dBTP (decibels true peak).

How True Peak Works

Digital audio is stored as a series of discrete samples, each representing the amplitude of the waveform at a single moment in time. When a digital-to-analog converter reconstructs the continuous waveform from these samples, the resulting analog signal can momentarily exceed the value of any individual sample. These are called inter-sample peaks.

A standard sample-peak meter reads the highest sample value and reports it as the peak. If the highest sample is -0.8 dBFS, the meter says your peak is -0.8 dBFS. But the reconstructed analog waveform between that sample and its neighbors might actually reach +0.2 dB. A sample-peak meter would tell you everything is fine. The analog output would clip.

Why Inter-Sample Peaks Exist

When a digital signal is converted to analog, a reconstruction filter interpolates the waveform between samples. If two adjacent samples are both near maximum amplitude, the interpolated curve between them can rise above both sample values. This is especially common with dense, limited material where the waveform is being driven hard against a ceiling. High-frequency content like cymbals, sibilance, and sharp transients are the most common sources of inter-sample peaks.

True Peak Measurement

True peak meters use oversampling to estimate the analog waveform between samples. The ITU-R BS.1770 specification requires at least 4x oversampling, which means the meter inserts three interpolated samples between each real sample and measures the highest value. More accurate meters use 8x or 16x oversampling.

The measurement is reported in dBTP instead of dBFS to distinguish it from sample-peak measurements. A true peak reading of -1.0 dBTP guarantees that the reconstructed analog waveform will not exceed -1.0 dB relative to full scale.

True Peak and Lossy Encoding

Inter-sample peaks become more dangerous after lossy encoding. When a mastered WAV file is encoded to AAC (Apple Music), Ogg Vorbis (Spotify), or Opus (YouTube), the codec transforms the audio data. This transformation can shift the waveform and create new peaks that were not present in the original file. A master with a sample-peak ceiling of -0.1 dBFS might produce peaks above 0 dBFS after AAC encoding.

This is why every major streaming platform recommends a true peak ceiling of -1.0 dBTP. Some engineers working with especially loud or dense material use -2.0 dBTP for extra safety.

Real-World Example

You master a dance track and set your limiter ceiling to -0.1 dBFS using a standard sample-peak meter. The meter shows no clipping. You upload the WAV to your distributor.

Apple Music encodes the file to 256 kbps AAC. During encoding, the codec introduces a small phase shift that moves the waveform slightly. An inter-sample peak that was at -0.05 dBFS in the WAV now reads +0.3 dBFS in the AAC file. The encoder clips the signal at 0 dBFS, introducing audible distortion on the kick drum and bass.

You remaster the track with a true peak limiter set to -1.0 dBTP. The limiter uses 8x oversampling to detect inter-sample peaks and prevents the output from exceeding -1.0 dB in the reconstructed waveform. After AAC encoding, the highest peak in the file is -0.7 dBFS. No clipping occurs on any platform.

Why It Matters for Independent Artists

If you master your own music, you need a true peak meter. Relying on the sample-peak meter built into your DAW is not sufficient for delivery to streaming platforms. Every major platform will encode your master to a lossy format, and that encoding can push sample-peak-safe masters into clipping.

Practical Steps

1. Use a true peak limiter. Modern limiter plugins like FabFilter Pro-L 3, iZotope Ozone Maximizer, and Waves L3 include true peak detection. Enable it and set the ceiling to -1.0 dBTP. This is the single most important setting in a streaming master.

2. Verify with a separate meter. Even if your limiter has true peak detection, verify the output with a dedicated true peak meter. iZotope Insight 2, Youlean Loudness Meter, and Voxengo SPAN all offer true peak readouts. Check both the integrated LUFS and the maximum true peak before exporting.

3. Test after encoding. If you want to be thorough, encode your master to AAC using Apple Music's encoder or iTunes Plus settings. Load the encoded file back into your DAW and check for clipping. This reveals whether your -1.0 dBTP ceiling is sufficient or if you need more headroom.

4. Watch out for loud, dense material. Heavily limited masters at -7 to -8 LUFS have waveforms packed tightly against the ceiling. These are the most likely to produce inter-sample peaks after encoding. If you master loud, consider -1.5 or -2.0 dBTP instead of -1.0.

In 2026, true peak detection is standard in every professional limiter and loudness meter. The ITU-R BS.1770-5 specification, updated in 2023, remains the current reference. AES streaming recommendations (AES TD1008.1) also specify -1.0 dBTP as the delivery ceiling. There is no reason to deliver a master that clips on streaming platforms.

Read our mastering for streaming platforms guide for platform-specific ceiling and loudness targets. Our mixing vs. mastering article explains where true peak measurement fits in the production chain.

Related Terms

• Limiter - The processor used to control true peak levels
• LUFS - The loudness standard that accompanies true peak in streaming delivery specifications
• Mastering - The stage where true peak ceilings are set
• Compression - Heavy compression increases the likelihood of inter-sample peaks
• Bit Depth - Higher bit depths provide more headroom before quantization noise affects peak measurements