True Peak Mastering Standards for Streaming
For streaming masters, set a true-peak limiter ceiling around -1 dBTP for normal releases and consider -2 dBTP when the master is very loud, bright, distorted, or likely to suffer during lossy encoding. True peak matters because sample peaks can look safe inside the DAW while inter-sample peaks clip after conversion, encoding, or playback. The clean workflow is to meter true peak after the final limiter, use oversampling, leave enough headroom, preview codec damage when possible, and judge the master by translation instead of chasing one loudness number.
True peak is one of the easiest mastering details to ignore because it does not always announce itself inside the session. Your sample-peak meter may show -0.3 dBFS and look technically safe. Then the song gets encoded, streamed, converted back to analog, or played through a phone speaker, and the top end turns gritty in a way that was not in the mix. That damage often comes from inter-sample peaks: waveform peaks that happen between the sampled points your basic meter shows.
The goal is not to make every master quiet. The goal is to stop loudness decisions from creating avoidable distortion. You can still master a competitive record. You just need a limiter and meter chain that tells the truth about the final file.
Need a streaming-safe master that keeps punch without clipping after upload?
Book Mastering ServicesWhat True Peak Actually Measures
A sample peak meter measures the highest stored digital sample. A true-peak meter estimates the analog waveform that will be reconstructed between those samples. That difference matters because digital audio is not just a staircase of sample points. When a converter rebuilds the waveform, the curve can rise above the highest individual sample. That overshoot is an inter-sample peak.
If your limiter ceiling is set close to 0 dBFS and the song is dense, clipped, bright, or aggressively limited, the reconstructed waveform may exceed full scale even though the DAW's sample meter never showed clipping. The result can be crunchy cymbals, sharp vocal consonants, distorted 808 edges, or a master that sounds smaller after upload than it sounded in your session.
A true-peak meter oversamples the signal internally and estimates those in-between peaks. It is not magic, and different meters can read slightly differently, but it gives a more realistic safety check than sample peak alone. For streaming delivery, that safety check is worth building into every mastering session.
The Practical Streaming Standard
The simple standard is this: keep the final master at or below -1 dBTP unless you have a clear reason to go lower. If the master is louder than the usual streaming normalization neighborhood, especially if it is dense and limited, use -2 dBTP as the safer ceiling. Spotify's own artist guidance recommends -1 dBTP for masters around its normalized playback level and -2 dBTP for louder masters because louder encodes are more susceptible to added distortion. SoundCloud gives similar guidance for avoiding distortion through its lossy formats.
That does not mean every platform has one identical, public, permanent rule. Some services publish clear guidance. Some describe normalization behavior without giving a full mastering table. Some change playback behavior by account setting, device, or codec. Your job is to deliver a master that survives those differences. A -1 dBTP ceiling is the baseline. A -2 dBTP ceiling is the extra safety move when the song is loud or fragile.
| Situation | Suggested ceiling | Why |
|---|---|---|
| Dynamic master with clean top end | -1 dBTP | Normal streaming-safe headroom |
| Loud pop, rap, EDM, or rock master | -1.5 to -2 dBTP | More protection during lossy encoding |
| Distorted 808s or clipped drums | -2 dBTP | Low-end harmonics and clipping can overshoot hard |
| Bright vocals, cymbals, or dense synths | -1.5 to -2 dBTP | High-frequency content often reveals codec stress first |
| Separate club/DJ master | Depends on purpose | May prioritize level, but should still be checked for clipping |
Why LUFS and True Peak Are Different Decisions
LUFS measures perceived loudness over time. True peak measures the maximum reconstructed peak. They are related because louder masters often need harder limiting, but they are not the same thing. A song can be -9 LUFS and stay below -1 dBTP. A quieter song can still clip if a transient hits too close to full scale. Do not use one number as a substitute for the other.
Streaming normalization also changes the incentive. If a platform turns down a loud master during playback, the extra limiting you used to make it louder may not create a louder listener experience. It may only reduce punch, increase distortion, and make the track feel smaller after normalization. The LUFS targets guide covers the loudness side. This article focuses on the peak-control side that prevents post-encoding damage.
The best masters balance both. You choose a loudness range that fits the genre and a true-peak ceiling that keeps the final file clean. If you are working from a mix that is already too hot, start with where your mix should peak before trying to solve everything with the mastering limiter.
Limiter Setup
Put the true-peak limiter at the end of the chain, after EQ, compression, clipping, saturation, widening, and any final tone shaping. If anything comes after the limiter, it can create new peaks. If you need a meter after the limiter, use a meter only. Do not put another gain stage, imager, EQ, or dither processor there unless you understand how it changes peak level.
Enable true-peak mode if the limiter has it. Enable oversampling, often 4x or higher if the CPU allows it. Set output ceiling to -1 dBTP for a normal streaming master. If the master is loud, set the ceiling to -1.5 or -2 dBTP. Then adjust input drive or threshold until the limiter is working only as hard as the song can tolerate.
Gain reduction is more important than the ceiling number alone. A limiter doing 1-3 dB of gain reduction on peaks may sound clean. A limiter doing 6-8 dB constantly may flatten drums, smear vocals, and exaggerate sibilance. If you need that much limiting to compete, consider earlier-stage clipping, compression, mix revisions, or a different loudness target. A limiter should finish the master, not rebuild the mix.
| Limiter control | Starting point | What to listen for |
|---|---|---|
| True-peak mode | On | Cleaner codec translation |
| Oversampling | 4x if available | Fewer aliasing and overshoot surprises |
| Ceiling | -1 dBTP to -2 dBTP | Less clipping after encoding |
| Gain reduction | 1-4 dB on loud sections | More than that can flatten impact |
| Release behavior | Program-dependent or tuned to groove | Pumping, dull drums, or smeared vocals |
Codec Preview and Export Checks
A master that sounds clean as a WAV can still reveal problems after AAC, Ogg, MP3, or another streaming encode. Codec preview tools are useful because they show how the file behaves after compression. If you do not have one, do a practical test: export the master, convert a copy to a common lossy format at a realistic bitrate, level-match it to the WAV, and listen for new harshness, low-end blur, cymbal fizz, stereo collapse, or vocal edge.
Do not listen louder to the encoded version. Level-match it. Louder almost always feels better for a few seconds. You are listening for damage, not excitement. The most revealing spots are bright vocal hooks, dense cymbal sections, loud 808 hits, distorted guitars, and final choruses where the limiter is working hardest.
Also check the exported file, not just the session playback. Some DAWs, plugins, and export settings can change gain or dither behavior. Reimport the final WAV into a new session and meter it again. Confirm integrated loudness, short-term loudness, sample peak, true peak, bit depth, sample rate, and file length. The pre-mastering checklist is a useful upstream reference when the mix arrives with issues that make true-peak control harder.
When -1 dBTP Is Not Enough
Use more headroom when the master is intentionally loud. The closer the average level gets to the peak ceiling, the less room the waveform has to breathe. Dense masters create more opportunities for codec distortion because the encoder has less clean space to represent transients. Dropping the ceiling from -1 to -2 dBTP will not fix an over-limited master, but it can prevent some avoidable overshoot.
Also use more headroom when the top end is aggressive. Bright vocals, sharp S sounds, hi-hats, cymbals, distorted synths, and clipped guitars can expose tiny artifacts fast. If the encoded preview makes the vocal spitty, lowering the ceiling may help, but do not stop there. Check the de-esser, limiter release, high-shelf boosts, stereo widening, and any clipping stage before the limiter.
Low-end distortion can also trigger problems. An 808 that is clipped into upper harmonics may look controlled but feel fuzzy after encoding. If the low end loses shape on the encoded preview, back off the limiter, reduce sub energy before the limiter, or use controlled saturation earlier instead of forcing the final limiter to do all the work.
Common Mistakes
- Setting a sample-peak ceiling to -0.1 dBFS and assuming the file is safe.
- Turning on true-peak mode but putting EQ or gain after the limiter.
- Mastering to a loud LUFS number while ignoring true peak.
- Using -1 dBTP as an excuse for heavy limiter distortion.
- Never checking the exported file.
- Never listening to a lossy preview.
- Chasing a platform target instead of making the song translate.
The worst mistake is treating true peak as a compliance checkbox. It is a listening problem. If the master sounds crunchy after upload, the listener does not care that your meter looked clean. Build the meter chain so it catches real-world damage before release day.
True Peak Workflow
- Start with a mix that has clean headroom and no master-bus clipping.
- Apply mastering EQ and compression before final peak control.
- Use clipping only if it improves tone and punch, not as hidden damage.
- Place a true-peak limiter last in the processing chain.
- Turn on true-peak mode and oversampling.
- Set the ceiling to -1 dBTP, or -2 dBTP for loud and fragile masters.
- Drive the limiter until the song reaches the loudness range that fits the genre.
- Level-match against references after normalization, not just at raw volume.
- Export the final WAV and reimport it for metering.
- Preview lossy encoding and fix any new distortion.
If you are comparing your master to commercial references, level-match them. A reference that is 3 dB louder will almost always feel better at first. Turn it down to match your master and listen to punch, vocal clarity, bass control, width, and fatigue. If your master only competes when it is louder, the problem may be mix balance, not final limiting.
Genre Notes
Rap and trap masters often need careful low-end control before the limiter. A huge 808 can eat limiter headroom and force everything else to distort. Shape the 808 in the mix, use saturation intentionally, and make sure the kick and 808 relationship is not causing unnecessary peak buildup. If you wait until mastering, you may have to choose between loudness and clean low end.
Rock and alternative masters often reveal true-peak problems in cymbals, vocal edge, and guitar fizz. If the mix is already bright, do not add a final high shelf just to create excitement. Try midrange balance, automation, or a more controlled limiter release before adding top end. Loud guitars can hide distortion until the chorus is played on earbuds.
Pop and R&B masters depend heavily on vocal polish. Sibilance that feels acceptable in the WAV may become painful after encoding. If true-peak control alone does not solve it, revisit de-essing, vocal saturation, and the final high shelf. The goal is a vocal that stays expensive at normal listening volume, not a vocal that wins a loudness race for ten seconds.
Delivery Files
Keep a high-quality master as the main archive file. A 24-bit WAV at the project sample rate is a common working delivery for distribution pipelines, though your distributor may provide exact requirements. Do not upload an MP3 as the master source unless the platform specifically asks for it and no better option exists. A lossy file that gets encoded again can compound artifacts.
Also print practical alternates when needed: instrumental, clean, performance, TV mix, and acapella. Those alternates should use the same true-peak discipline. It is easy to make the instrumental clip because the vocal is gone and the beat feels quieter. Meter every version. Name files clearly. Store the limiter settings and loudness readings with the project notes so revisions do not become guesswork.
How Platform Playback Changes the Decision
Streaming platforms do not all behave like a simple file player. Some normalize playback loudness. Some let listeners change normalization settings. Some use different codecs for different account types, devices, or connection settings. Some playback paths involve smart speakers, TVs, car systems, browser players, mobile apps, or lossless modes. That means your master is not only judged by the WAV you export. It is judged by how well the WAV survives the distribution and playback chain.
This is why true peak is more useful than a last-minute compliance number. A clean -1 dBTP master gives playback systems room to encode and reconstruct the waveform without scraping the ceiling. A clean -2 dBTP master gives even more room when the song is already dense. If a platform turns the song down during playback, that extra headroom does not make the song weaker. It simply reduces the chance that the loudest moments create distortion before the listener hears them.
Do not make separate masters for every platform unless you have a real delivery reason. Most independent releases go through one distributor with one primary master. That single master should be robust enough for the major listening environments. A streaming-safe master is not the quietest possible file. It is the file that keeps the vocal clear, low end stable, transients intact, and top end clean after normal playback processing.
What to Do When the Client Wants It Louder
Artists often ask for louder masters because loudness feels impressive during a quick A/B. The mastering decision is to separate useful loudness from destructive loudness. If another half dB of drive makes the chorus feel more finished without flattening the snare, blurring the 808, or adding vocal grit, it may be worth it. If the limiter starts changing the groove, the master is not improving. It is only getting denser.
Use references at matched loudness. If the client reference is louder, turn it down until the perceived volume is similar. Then compare bass shape, vocal size, snare impact, stereo width, and fatigue. If the reference still feels better, the mix or master balance needs work. If the reference only wins because it is louder, the request may be a monitoring illusion.
Offer two prints when needed: a clean streaming master and a louder reference master. Label them clearly. The clean master should be the release candidate if it translates better. The louder reference can help the artist understand the tradeoff. This is especially useful with aggressive rap, EDM, and rock records where the line between exciting and damaged can be thin.
Mastering QC Checklist
Before approving a master, run a short quality-control pass. Check the loudest chorus, the first vocal entrance, the densest low-end section, and the final ten seconds. Meter integrated LUFS, short-term LUFS, sample peak, and true peak. Confirm the true-peak ceiling after export. Listen to the WAV and a lossy preview at matched volume. Check the clean version if one exists. Check the instrumental if it will be used for performance or sync.
Write the readings down. A simple note like "Main master: -9.8 LUFS integrated, -1.2 dBTP, 24-bit WAV, 48 kHz" makes future revisions easier. If the artist returns two weeks later asking for a slightly quieter vocal, you can reopen the session and know what target you are returning to. Without notes, every revision becomes a new mastering decision.
End the pass with a short silent-gap and fade check. True-peak problems usually get attention during loud choruses, but bad fades, clipped count-ins, and noisy endings can also make a release feel unfinished. Clean starts and endings are part of the same delivery discipline as clean peak control.
FAQ
What true peak ceiling should I use for streaming?
Use -1 dBTP as the normal streaming ceiling and consider -2 dBTP for loud, dense, bright, or heavily limited masters. The extra headroom helps reduce distortion during lossy encoding and playback conversion.
Is true peak the same as sample peak?
No. Sample peak measures the highest stored sample, while true peak estimates the reconstructed waveform between samples. A file can show no sample clipping and still create inter-sample peaks during conversion or encoding.
Do I need to master every song to -14 LUFS?
No. Streaming normalization uses loudness targets during playback, but you should master for the song, genre, and desired punch. Use LUFS as context and true peak as a safety check, not as a single formula.
Should I use -2 dBTP for every master?
Not necessarily. -2 dBTP is safer for loud or fragile masters, but it can slightly reduce available peak level. If a dynamic master sounds clean at -1 dBTP and survives codec preview, -1 dBTP is usually fine.
Can true-peak limiting fix an over-compressed master?
No. True-peak limiting can prevent overshoot, but it cannot restore punch, depth, or clarity lost from too much compression and limiting earlier in the chain. If the master feels flat, fix the dynamics before the final limiter.
Where should my true-peak meter go?
Place the true-peak meter after the final limiter and any final gain stage so it reads the actual exported level. If you put processing after the meter, the meter is no longer showing the final file.
