Spotify currently adjusts tracks toward minus fourteen LUFS during normalized playback, but that figure is not a universal mastering target. The adjustment changes playback gain rather than repairing damaged audio. A louder upload can be turned down, but its flattened drums and clipped transients remain flattened.
Self-mastering becomes dangerous when extra volume gets mistaken for better sound. You already know every production decision, so familiarity can hide faults while the louder version wins every casual comparison. Begin with level-matched judgment, not a favorite plugin chain.
Keep an untouched version of the finished mix available throughout the session. Mastering should improve the whole stereo presentation without changing the song's internal relationships by accident. If the vocal, kick, or chorus needs individual treatment, the mix is still the better place to work.
The same distinction applies to dynamics. A master-bus compressor can change the movement of the whole song, but it cannot lower an overbearing vocal without also reacting to drums, bass, and everything around it. If one element causes the complaint, reopen the mix instead of forcing a stereo processor to work around it.
Clicks, abrupt edits, accidental silence, and distorted recordings are not character added by mastering. Fix them at their source whenever the session is available. When one correction requires three compensating moves elsewhere, stop and return to the mix because the first move created more trouble than it removed.
Headroom is simpler than internet folklore makes it sound. A clean floating-point mix does not need to peak at one magic number, but it must avoid intentional clipping unless that clipping is part of the approved sound. Turning down an unclipped file creates working space without changing its dynamics.
Processing order should follow the cause. If excessive low end drives the limiter into audible distortion, reducing that excess before limiting is more direct than asking the limiter to recover faster. If tonal balance is already right, adding EQ because a template includes it merely introduces another decision that can go wrong.
Bypass each meaningful change at matched loudness. A one-decibel gain advantage can make a version seem clearer, wider, and more exciting even when its balance has deteriorated. The processed version should retain its advantage after that loudness advantage disappears.
A commercial reference can expose a master that is unusually dark, narrow, or dense, but comparison needs context. Choose music with related instrumentation and energy, then reduce the reference to a comparable listening level. Copying its spectrum or loudness blindly can erase choices that make your song distinct.
Work at a repeatable, moderate monitor level and check quietly before making another move. Loud monitoring accelerates fatigue and can disguise harshness until the next day. Small reversible mastering moves are easier to judge, undo, and explain than a chain that changes five qualities at once.
For normalized streaming, pushing beyond the song's comfortable loudness often buys no playback advantage. Spotify currently offers different normalization levels and may reduce a loud master during playback. Apple also provides encoder-audition tools because a clean PCM file can clip after AAC conversion.
Raise the limiter input while listening to the loudest and most exposed sections, then watch for specific damage. Snare attacks become papery, bass notes pull the mix downward, vocals spit, and stereo depth collapses before a meter declares failure. Those audible changes matter more than reaching an inherited LUFS value.
Leave enough true-peak margin for the intended destination and audition a representative lossy encode when possible. Compare that result with the lossless master at equal loudness. If encoding exposes crackle or hardens transients, lower the ceiling or reduce the processing that made those peaks fragile.
Save the best pass, step away, and compare it with the untouched mix at matched loudness. Keep the master only when it solves the written problems without weakening punch, focus, or emotional movement. Once further changes sound merely different, the processing has stopped earning its place.
Self-mastering becomes dangerous when extra volume gets mistaken for better sound. You already know every production decision, so familiarity can hide faults while the louder version wins every casual comparison. Begin with level-matched judgment, not a favorite plugin chain.
Keep an untouched version of the finished mix available throughout the session. Mastering should improve the whole stereo presentation without changing the song's internal relationships by accident. If the vocal, kick, or chorus needs individual treatment, the mix is still the better place to work.
The mix must survive the first decision
Listen once before loading a processor and write down only problems you can hear. Separate global faults from local ones. A dull balance across the entire song may justify broad equalization, while one piercing guitar note belongs in the mix or an isolated repair.The same distinction applies to dynamics. A master-bus compressor can change the movement of the whole song, but it cannot lower an overbearing vocal without also reacting to drums, bass, and everything around it. If one element causes the complaint, reopen the mix instead of forcing a stereo processor to work around it.
Clicks, abrupt edits, accidental silence, and distorted recordings are not character added by mastering. Fix them at their source whenever the session is available. When one correction requires three compensating moves elsewhere, stop and return to the mix because the first move created more trouble than it removed.
Headroom is simpler than internet folklore makes it sound. A clean floating-point mix does not need to peak at one magic number, but it must avoid intentional clipping unless that clipping is part of the approved sound. Turning down an unclipped file creates working space without changing its dynamics.
Every processor must earn its place
Start with the smallest move that addresses the written problem. Broad EQ can correct an overall lean or heavy balance. Compression belongs only when controlling movement or density makes the song feel steadier without softening the attack that gives it life.Processing order should follow the cause. If excessive low end drives the limiter into audible distortion, reducing that excess before limiting is more direct than asking the limiter to recover faster. If tonal balance is already right, adding EQ because a template includes it merely introduces another decision that can go wrong.
Bypass each meaningful change at matched loudness. A one-decibel gain advantage can make a version seem clearer, wider, and more exciting even when its balance has deteriorated. The processed version should retain its advantage after that loudness advantage disappears.
A commercial reference can expose a master that is unusually dark, narrow, or dense, but comparison needs context. Choose music with related instrumentation and energy, then reduce the reference to a comparable listening level. Copying its spectrum or loudness blindly can erase choices that make your song distinct.
Work at a repeatable, moderate monitor level and check quietly before making another move. Loud monitoring accelerates fatigue and can disguise harshness until the next day. Small reversible mastering moves are easier to judge, undo, and explain than a chain that changes five qualities at once.
Loudness stops when the music starts shrinking
Integrated LUFS describes average perceived loudness over time, while true peak estimates peaks that can occur between stored samples. They answer different questions. A master can meet a loudness figure and still produce overs when converted to a lossy listening format.For normalized streaming, pushing beyond the song's comfortable loudness often buys no playback advantage. Spotify currently offers different normalization levels and may reduce a loud master during playback. Apple also provides encoder-audition tools because a clean PCM file can clip after AAC conversion.
Raise the limiter input while listening to the loudest and most exposed sections, then watch for specific damage. Snare attacks become papery, bass notes pull the mix downward, vocals spit, and stereo depth collapses before a meter declares failure. Those audible changes matter more than reaching an inherited LUFS value.
Leave enough true-peak margin for the intended destination and audition a representative lossy encode when possible. Compare that result with the lossless master at equal loudness. If encoding exposes crackle or hardens transients, lower the ceiling or reduce the processing that made those peaks fragile.
Save the best pass, step away, and compare it with the untouched mix at matched loudness. Keep the master only when it solves the written problems without weakening punch, focus, or emotional movement. Once further changes sound merely different, the processing has stopped earning its place.
