Audio mastering is the final stage of music and audio production, in which a finished stereo mix is balanced, polished and prepared for distribution. The mastering engineer adjusts tonal balance, dynamics, stereo image and loudness, then creates delivery files that play back consistently on streaming services, CDs, vinyl, broadcast and every kind of speaker or headphone.
Mastering, often simply called mastering, differs from mixing in scope: mixing balances individual tracks inside a song, while mastering treats the whole program as one signal and ensures that songs on an album or release sound coherent together. It is also the last point of technical quality control before a recording leaves the studio.
What audio mastering does to a mix
Mastering applies small, broad corrections that affect the entire mix. Typical moves are measured in fractions of a decibel to a few decibels, because a well-made mix needs refinement rather than repair, and large changes at this stage alter every instrument at once.
The work has two sides. The creative side improves translation and impact: smoothing harsh frequencies, adding weight to the low end, controlling dynamics and setting a competitive but musical loudness. The technical side covers sample rate and bit depth conversion, dithering, peak control, metadata such as ISRC codes, track spacing and the creation of format-specific masters.
A crucial tool is critical listening in a calibrated room on accurate monitors. Mastering engineers compare the mix against reference tracks and check it on several playback systems, because a master that sounds impressive on one system but thin or boomy on others has failed its main purpose.
The mastering chain and its processors
A mastering chain is the sequence of processors through which the mix passes. Its order is not fixed, but most chains move from corrective processing to tonal shaping, dynamics and finally peak limiting and format conversion.
| Processor | Purpose | Typical setting range |
|---|---|---|
| Equalizer | Corrects tonal balance | Gentle moves of about 0.5 to 3 dB |
| Compressor | Glues dynamics and controls level swings | Low ratios around 1.5:1 to 2:1, 1 to 3 dB reduction |
| Multiband compressor | Controls one frequency range only | Narrow, targeted gain reduction |
| Mid/side processing | Adjusts center and sides separately | Small width or tonal changes |
| Saturation | Adds density and harmonic color | Subtle, often barely audible |
| Limiter | Raises loudness and catches peaks | Ceiling near -1 dBTP |
| Dither | Masks quantization error when reducing bit depth | Applied once, at the final stage |
Each processor should solve a problem the engineer can hear. Adding processing by habit is a common source of a flat, fatiguing sound, and many excellent masters use only two or three stages.
Loudness, true peak and delivery targets
Loudness in mastering is measured in LUFS, an integrated loudness unit that reflects perceived level, and peaks are measured as true peak in dBTP, which accounts for inter-sample overs created during digital-to-analog conversion and lossy encoding.
| Destination | Typical loudness reference | Typical peak ceiling |
|---|---|---|
| Major music streaming services | About -14 to -16 LUFS normalization | -1 dBTP, lower for loud masters |
| European broadcast under EBU R128 | -23 LUFS | -1 dBTP |
| US television broadcast | -24 LKFS | -2 dBTP |
| Podcasts and spoken word | About -16 to -19 LUFS | -1 dBTP |
| CD release | No fixed target | Below 0 dBFS, usually -0.3 to -1 dBTP |
Because streaming platforms apply loudness normalization, a master pushed far above the reference is simply turned down during playback. The extra loudness then brings no benefit, only lost dynamics and more limiter distortion. This shift has eased the pressure of the loudness war, although many genres still master well above streaming references by artistic choice.
Engineers also watch loudness range and crest factor, which describe how much the level varies over time and how far peaks rise above the average. A master with an extremely low crest factor may meet a loudness number yet sound dense and lifeless, especially after normalization places it beside more dynamic material.
Mastering compared with mixing and stem mastering
The boundary between stages determines what can be fixed where. Mastering can correct overall balance, but it cannot rebalance a buried vocal without side effects.
| Stage | Input | Main control | Best for |
|---|---|---|---|
| Mixing | Individual multitrack recordings | Every instrument separately | Balance, effects and arrangement |
| Stereo mastering | One stereo mix file | The whole program | Finishing sound and release preparation |
| Stem mastering | Grouped submixes such as drums and vocals | Limited group-level changes | Mixes with balance issues |
Stem mastering offers more control but blurs responsibility between mix and master. It is most useful when the mix engineer is unavailable or when a release must be adapted for several formats.
Automated mastering services analyze a mix and apply a preset chain toward a target loudness and tonal curve. They can give quick, acceptable results for demos, but they cannot judge artistic intent, album sequencing or problems that should be sent back to the mix stage, which remain the core value of a human mastering engineer.
Common mastering mistakes
Most mastering problems come from chasing loudness or over-processing, and from comparing versions at different volume levels, which makes the louder one seem better.
- Not level-matching comparisons, so a louder version always sounds superior.
- Excessive limiting, which flattens transients and creates audible distortion.
- Ignoring true peak, causing clipping after conversion to lossy streaming formats.
- Wide stereo bass, which causes phase problems and is unsuitable for vinyl cutting.
- Dithering more than once, or truncating to 16 bits without dither.
- Mastering in an untreated room, where room modes distort low-frequency decisions.
A good mix delivered with sufficient headroom, typically peaks several decibels below full scale and no limiter on the mix bus, gives the mastering stage room to work.
In plain words
Audio mastering is the final polish before a song is released. The mix is already finished, and mastering makes it sound balanced and complete. It should sound good on headphones, car speakers, phones and large sound systems alike.
Think of it like the last edit of a printed book. The story is written, but someone checks the layout and spacing. They also make sure every chapter matches. Mastering does the same for sound and makes all songs on a release feel like one album.
A real-world example
An independent band sends a mastering engineer ten mixed songs for an album to be released on streaming services and vinyl. The mixes peak around -4 dBFS and sound good individually, but the songs differ in brightness and loudness, and two tracks have a boomy low end.
The engineer applies gentle equalization per track, a low-ratio compressor for cohesion and a limiter with a true peak ceiling of -1 dBTP. The digital master lands at around -10 LUFS integrated, matched across songs by ear. A separate vinyl master keeps more dynamic range, sums the bass below about 150 Hz to mono and reduces harsh sibilance.
The result is a consistent album that translates well on earbuds and home speakers, plus a vinyl master that cuts cleanly without distortion.