The First Time Anyone Hears It as a Record

Mastering has a reputation for being the mysterious stage — the part nobody quite explains. The work itself is not mysterious at all. What makes it hard to describe is that it is the first time anybody listens to the material as a record rather than as a collection of songs, and almost everything distinctive about the discipline follows from that change of scale.

The process has three plain stages: assembly, sweetening and output.

Assembly

Assembly builds the release as an object. The tracks are sequenced in order, the spacing between them is set, the beginnings and endings are cleaned up and faded, and any clicks, pops or stray noises that survived mixing are removed.

None of that sounds like creative work and all of it is audible. How a track ends is a choice between three genuinely different results:

• Silence — fading to digital black. Clean, and the one that most often reads as abrupt, because a real room never goes completely silent.
• Room tone — fading to the natural, barely-there background of the space the music was recorded in, left in deliberately. The song stops; the room does not.
• A crossfade into the next track, with no gap at all.

A release with uneven gaps or an unconsidered ending reads as unfinished no matter how good the mixes underneath are. The last lesson of this course covers the codes and track markers that make all of this actually work on a delivered product.

Sweetening

Sweetening is what most people picture: broad, subtle equalization correcting the overall tonal balance, and gentle compression and limiting managing dynamics and perceived loudness — applied to the mix as a whole rather than to any instrument in it.

The origin of the tools is worth knowing, because it explains their character. When engineers first cut master discs for vinyl, they built compressors, limiters and equalizers for a purely defensive reason: to stop the signal overloading the cutter head. Too much low end or too sharp a transient would make the cutting stylus damage the lacquer or cut into the adjacent groove.

Then they noticed the settings also made the music sound better. An entire discipline grew out of a safety device for a mechanical process, which is why mastering processing has always had a protective streak in it and why the good equipment is designed to be transparent first and characterful second.

The emphasis on subtle is not modesty. Mastering EQ moves are usually small and wide — a decibel or two across an octave — rather than the narrow surgical moves that are appropriate on an individual track during mixing. Heavy-handed mastering EQ is one of the fastest ways to undo good work underneath it, because there is no way to affect one element without affecting everything sharing that frequency range.

Output

Output is preparing the files that each destination actually needs: the right format, the right resolution, the right loudness, checked for mono compatibility and for peaks that a lossy encoder might push over.

Historically the deliverable was a CD-R that could be played on any machine, so a client could audition it and approve it, and which the replication plant then used to cut a glass master — a perfect mirror image of what the mastering house produced. That role is now filled by a file set, which the final lesson covers.

A mix that sounds perfect in your own room can still arrive at a listener too quiet, too loud or subtly distorted if this stage is skipped. It is the bridge between sounding right where it was made and sounding right everywhere it will be played.

Why it is a separate room and separate ears

There are three real reasons, and only one of them is about equipment.

Fresh ears

Whoever mixed the record has heard it fifty times, and by the end they are no longer hearing the music — they are hearing the list of decisions they made. Adaptation is not a failure of concentration; it is what ears do. Somebody who has never heard the track before hears what a listener will hear.

Consistency

This is the strongest argument and the least discussed. An album's songs are typically mixed over weeks, at different times of day, on ears that were in different states, sometimes in different rooms. Every one of those mixes can be individually excellent and the collection can still not sound like one record — the third track a little brighter, the seventh a little quieter, the closer with more low end than anything before it.

Mastering is what turns a set of good mixes into a single object. If a release only needs one thing from the process, it is usually this.

A different kind of listening

Mixing listens inward: is the vocal sitting right, is the snare too loud, does the guitar mask the piano. Mastering listens outward: what is this record's overall character, does it hold up across forty minutes, does it feel like the thing it is trying to be.

Those are two modes rather than two skill levels, and switching between them on the same afternoon is genuinely difficult. A separate person makes it structurally easier.

The equipment argument is real but secondary. Running one guitar through a limiter and running an entire finished mix through it are different acts: a mix is a balance that can be damaged as easily as improved, so the processors have to be clean enough that the only change is the one you intended.

The vocabulary problem

Feedback almost never arrives as "boost 3 dB at 2 kHz". It arrives as "make it warmer", "give the keys some sparkle", "it needs more air", "can it be bigger".

Translating those into an actual move is a practised skill and a real part of the job — not a quirk of the industry. The terms are surprisingly consistent between people who have never met, which suggests they are tracking something genuine about how frequency ranges are perceived. The lesson on the language of equalization takes that vocabulary apart properly.

Monitoring

A great many problems that arrive at mastering were created on monitors that could not be trusted. If the mixing speakers exaggerate the top end, the mixer compensates by dulling everything, and the fault arrives baked in.

There are two defensible positions on mastering monitors and they pull in opposite directions:

• Near-fields, placed close, minimise the room's contribution and resemble what most listeners actually own. They will not deliver deep bass — but neither will most consumer systems, so it is not a bad compromise, and speakers this size often image very precisely.
• Large, accurate, wideband systems in a room designed for listening give the most truthful picture of the music. Almost nobody listening at home has anything like it, and that is not really the point: if the speakers tell the truth, you can compensate from experience.

Both are chosen deliberately by people who are good at this. What is not defensible is a room nobody has evaluated.

Analogue or digital

Given the choice, a mastering facility wants source material at higher resolution than the target — 24-bit files or analogue tape for a 16-bit delivery, so there is room to work before anything is committed to the final word length.

The comparison between the two is worth stating without partisanship:

• Dynamic range is roughly comparable, with measured signal-to-noise favouring digital. Most pop and jazz has considerably less than 85 dB of range anyway, so noise is rarely the deciding issue.
• Distortion favours digital by a factor of at least ten. What analogue offers in exchange is the shape of its failure: driven hard, tape saturates into a natural compression rather than hitting a wall.
• Frequency response is flatter on digital within the band, and then falls off a cliff at the anti-alias filter. A 15 ips tape machine remains useful to around 30 kHz, and a good 30 ips machine well past 40 — an extra octave and, more importantly, a gentler roll-off. That difference in shape is what people are describing when they say analogue has "air".
• Low frequencies on tape carry small irregularities called head bumps, usually 2 dB or less, caused by the geometry of the playback head. They do not cause phase problems, and anyone mixing on such a machine compensates for them without thinking about it.

Below sixteen bits of signal in a digital system there is nothing left; on tape, musical information continues below the noise floor. That is a genuine argument, and it is also an argument for 24-bit rather than for tape.

The practical rule when a chain crosses between the two domains: convert once, carefully. Each crossing is a real stage. And if the crossing is happening, it is worth re-clocking the data on the way, since jitter shows up not as noise but as a loss of stereo focus and depth — a blurring, which is much harder to identify than a fault you can name.

Tube, solid state and digital

The argument about circuit types will not be settled, because all three approaches can be excellent when built carefully and each fails in a characteristic direction:

• The best solid-state designs come close to disappearing. They tend towards a faster, more defined, slightly less forgiving sound — they show you what is there, including what you would rather not see.
• The best tube designs approach transparency from the other side, with a sense of air. People reliably reach for the word "warmth", and tube circuits offer a wider range of options when the goal is to improve a sound rather than pass it through unchanged.
• Digital designs approach the standard set by good solid state and are less forgiving again. A harder, more closed top end is the usual criticism, along with some loss of depth and sound-staging.

None of that is a reason to prefer one. It is a reason to know which one you are working through, because the failure directions are different and the compensations you make should be too.

The mono check

Before anything else happens, sum the mix to mono and listen. If the level drops noticeably, or an instrument thins out or vanishes, there is a phase problem — usually from a delay line, a stereo guitar processor, or room microphones that were never checked against the close ones.

It matters beyond the technical tidiness, and the real-world consequence is specific: stereo separation degrades with distance from a broadcast transmitter. A listener at the edge of a station's coverage is hearing something closer to mono, so a guitar that lives entirely in the out-of-phase content simply disappears for them. The part was there in the studio and is gone in the car.

Mastering will expose that problem rather than fix it, because anything done to widen or thicken the record will emphasise exactly the content that cancels.

What to send

A short list, and following it costs nothing:

• Do not normalise, EQ or compress during assembly editing. If a local facility is sequencing the tracks, ask them to sequence and nothing else.
• Do not send a heavily limited mix. A submaster that has been through a mix-bus maximiser has had its headroom spent already, and the sonic signature is not reversible — which leaves the mastering engineer with far less room to move and a sound they cannot undo.
• If you love a particular stereo compressor, use it — moderately — and print a parallel unprocessed safety copy from the desk at the same time. Then the decision stays available.
• In general, the less stereo bus processing, the better the final result. Enhancers and exciters are a common exception in the wrong direction: they are frequently used to compensate for dull monitors, and on an accurate system the compensation is what you hear.

Alternate mixes

While the session is still open and every decision is fresh, two extra passes are worth printing:

• Vocal up — the same mix with the vocal one or two decibels higher. Vocals get buried during long sessions more often than anyone expects, and choosing between two versions later is much easier than remembering what you meant.
• Varispeed — if the multitrack allows it, run the mix again a few cents faster, up to about one percent. It will irritate anyone with perfect pitch, and it lends a subtle lift in energy that is worth hearing at least once before dismissing.

What mastering cannot do

It improves good work. It does not repair bad work, and being clear about that saves a great deal of disappointment.

Everything arrives as two channels. If the kick sounds excellent and the bass guitar sounds wrong, there is no setting that fixes one without moving the other, because they occupy the same frequencies in the same file. The same is true of a muddy arrangement, an unconvincing performance or a vocal that was never quite in tune — all of those are mix or production problems wearing a mastering costume.

The correct response is sometimes to send the record back, which is a conversation worth having early rather than after three attempts to fix it with equipment.

What a session actually looks like

Two formats, and the difference is worth knowing before booking one.

Attended — the artist or producer is in the room. Decisions are made and agreed in real time, the conversation about what the record should be happens face to face, and you leave with something everyone has heard together on a system they can trust. It costs more and it removes almost all of the revision cycle.

Unattended — files are sent, the engineer works alone, and a reference is returned for comment. Far more common now, considerably cheaper, and it makes the quality of your written notes matter a great deal. "Track 4 feels thin against track 3" is useful; "make it better" is not.

Either way there is a revision round, and the useful discipline is to listen to a returned master the next day, on more than one system, before writing anything. Notes written in the first ten minutes are almost always about the difference from what you were used to rather than about the record.

The loudness tension

A louder master competes better against a radio station's own processing and sounds more impressive on first hearing. Pushed too far, a record that is exciting for fifteen minutes becomes tiring by the fortieth, even for listeners who like it.

For a long time this had no settled answer, and labels ended up in a contest to be half a decibel louder than the last release — trading clarity and the sense of air for a competitive edge that stopped existing as soon as everyone had it. A good home system can reproduce over 90 dB of range, and that is wasted when the music is crammed into the top ten.

That argument is now genuinely settled, and not by anybody winning it. The lesson on loudness and LUFS in this course explains what actually ended it, which is more interesting than the argument itself.

What to practice

• Play the last thirty seconds of one song straight into the first ten of the next, at the gap you have chosen. That handover is a decision, and hearing it in isolation is the only way to judge it.
• Take a mix you finished today and do not open it again for a week. What you notice on the eighth day is precisely what mastering exists to catch.
• Line up your record against a commercial release in the same genre, matched for loudness rather than level. Any comparison at different loudness tells you nothing at all.
• Ask someone who has not heard the record to say what it sounds like in three adjectives. Translating that into a move is the actual skill, not operating the equipment.
• Check a mix in mono. If instruments vanish or the level drops noticeably, you have a phase problem that mastering will make worse rather than better.