What peak level should I aim for?
Many delivery specs suggest a true peak around −1 dBTP, sometimes lower for lossy formats. Check the platform’s own guidance and leave a little more room when you convert to MP3 or AAC.
What is Peak Level? Sample Peak and True Peak
Peak level is the highest absolute sample value in a digital file, usually shown in dBFS (decibels relative to full scale). 0 dBFS is the maximum a standard PCM file can store. Going past that in the digital domain means clipping unless you use floating-point processing.
Sample peak vs true peak
Sample peak only looks at stored samples. True peak estimates the reconstructed waveform between samples after conversion or lossy encoding. That reconstructed peak can be slightly higher than the sample peak, which is why streaming specs often limit true peak (for example around −1 dBTP) even when sample peak looks safe.
Headroom in plain terms
- Leaving a few dB below 0 dBFS gives room for processing and for true-peak overshoot.
- A file that “touches” 0 dBFS on every snare is not automatically “louder” in LUFS terms — it may just be limited hard.
- A quiet file can have a low peak and still need gain before it is useful on phones.
Example
An interview peaks at −12 dBFS: plenty of headroom, but it may sound soft next to other content. A club track peaks at −0.1 dBFS with high LUFS: the peak meter is maxed, and platforms may turn playback down. Peak alone does not tell you which will feel louder in a playlist — pair it with LUFS or RMS.
Check peaks on your file
Audio Quality Check reports sample peak, true-peak style estimates where available, and clipping risk. Use Volume Booster when the peak is low and the whole file needs a lift, and avoid stacking huge boosts on a file that already kisses 0 dBFS.
Why peak readings are negative numbers
dBFS counts down from the top of the scale. Zero is the loudest value a digital file can store, so everything quieter is written as a negative number. A peak at −6 dBFS means the loudest sample is half the maximum amplitude, and a peak at −20 dBFS leaves a lot of headroom. The smaller the negative number, the closer to the limit.
Inter-sample peaks after lossy encoding
MP3 and AAC encoders rebuild the waveform from compressed data, and the rebuilt wave can swing slightly higher than the original samples suggested. A master that peaks at exactly 0 dBFS may therefore clip after conversion. Leaving a little space, for instance −1 dBTP or a bit lower for lossy delivery, protects against that.
Reading a peak meter in practice
Most meters have a peak-hold line that stays at the highest value until you reset it. Play the loudest part of the file, note the held value, and reset before the next test. Loud moments such as laughter, a drum fill or a shout often set the peak, not the average speech. If a meter shows red on one syllable only, a short peak limiter or a small gain reduction is usually enough.
Adjust
Volume Booster
Raise under-level files without guessing from the waveform alone.
Open Volume Booster0 dBFS is a ceiling, not a target
New producers often raise a track until the peak meter kisses zero, assuming that equals “done.” In a floating-point DAW you can go past that while processing, but a 16-bit or 24-bit file you bounce still has a hard stop. Leaving a little headroom (for example peaking near −1 dBFS or lower when true-peak matters) reduces the chance that the next encode introduces overs.
Peaks on speech versus peaks on music
Plosives (“p”, “b”) and laughs create sharp peaks on voice tracks even when the average level is modest. Music with sustained synths can show a lower peak-to-average gap. That is why a voice memo and a finished song can share a similar peak reading and still need different treatment before they sit in the same feed.
When you inspect a file, ask two questions: is the highest peak safe for the format, and is the rest of the program loud enough to hear? Peak answers the first; LUFS or RMS help with the second.
Frequently asked questions
Does lowering the peak make the file quieter?
Applying negative gain lowers every sample, so the peak and the loudness both drop. Peak-limiting only the top spikes can lower the peak while keeping most of the programme at the same level.
What does dBTP mean?
It is decibels of true peak, relative to full scale. It estimates the highest level of the reconstructed waveform between samples, not just the stored sample values.
Can a file go above 0 dBFS?
Fixed-point files such as 16-bit WAV cannot store it, and the signal clips. Floating-point files used inside editors can exceed 0 temporarily, but they must be brought back down before export.