What mastering does that turning it up does not

Somebody has told you your audio needs mastering, and it sounds like being charged twice for the same afternoon. The mix is finished. The voice sits over the music, the levels are steady, nothing is fighting. What is left to do?

The short version: mastering checks and prepares the finished mix for delivery. One careful master may serve several destinations; make additional versions where the delivery requirements differ. Even a single upload needs a final check.

What is the difference between mixing and mastering?

Mixing balances the parts against each other. Voice against music, music against room tone, the loud sentence against the quiet one. Every decision in a mix is about the relationship between things inside the file.

Mastering works on the finished mix: loudness, dynamics, tonal balance and delivery checks. If the relationship between voice and music needs changing, return to the mix where those parts are separately available.

One mix, one quality-controlled master, and a separate finish only where a destination needs one. That is the entire idea.

Why would one video need more than one version?

Because destinations do not treat a file the same way, and the thing you are adapting to is the destination, not the room somebody happens to be listening in, which you never know. What varies from one place to another is the loudness it normalises to, whether it normalises at all, the codec and bit rate it wants, and how much dynamic range survives its encode.

A long-form upload, a short vertical cut and a broadcast or advertising deliverable can each want a different level, a different amount of dynamic control and a different file. Where those differ enough to matter, one file sent everywhere lets each destination decide for you, and some decide badly. Where they are close enough, and for a single streaming platform they usually are, one careful master carries them all and the platform handles the rest.

What a master decides

Three things, and none of them is a knob you turn up.

The first is arrival level. Most large platforms measure your upload and turn it towards a loudness target of their own, so viewers stop reaching for the volume control between videos. What happens to your file depends on where it is going and on whether it arrived louder than that destination wanted: arrive loud and it is turned down, arrive quiet and most destinations leave it where it is, which is the outcome people do not expect. The only choice you really have is whether the level was aimed at, or ambushed by, that.

The second is how much dynamic range survives. Dynamic range is the distance between the quietest and loudest parts of your audio. Squash a file hard before uploading and the platform pulls it back down again, but the distance you flattened does not return. You arrive at the same loudness as everybody else with less shape left, which is the specific way loud files end up sounding smaller.

Check the words and important cues on small speakers as well as headphones. Fine low-frequency detail may disappear on a phone; that does not make it worthless for listeners on other systems.

Is this worth doing for one video?

A single video still needs a delivery check.

For one destination, check the actual delivery requirements, loudness, peaks and playback. If the mix already meets them and sounds right, the final pass may be brief.

When does it start to matter?

When the delivery requirements or the material call for a different finish.

A long lesson, a short cut, an audio release and a website embed may have different needs. Check those needs first. Separate edits often need their own checks, but do not create different masters merely because the destination names differ.

Document the export and checking process so repeated deliveries are manageable. Bring in help where the volume or technical requirements justify it.

What can you check before you publish?

Two things, and both are free.

Play the export on a small speaker at a normal listening level. If words disappear, compare with headphones and inspect the mix, playback level and source recording before deciding what to change.

Listen again on headphones in a quiet room. If the sound feels tiring, check the dynamics, tonal balance and harshness. Fatigue is a reason to investigate, not a diagnosis of over-compression by itself.

Use the two listening checks to identify where the sound changes, then inspect the cause. Do not automatically make the file quieter or compress it further; the right treatment depends on the mix and destination.

Sources and evidence

Standards and platform facts. Loudness is measured with ITU-R BS.1770-5 (International Telecommunication Union, 2023), the shared algorithm the Audio Engineering Society describes for playback normalisation. Delivery targets then differ by destination: the European Broadcasting Union's R128 covers broadcast, Spotify for Artists publishes about -14 LUFS with true-peak guidance, and YouTube publishes recommended upload encoding but no equivalent loudness target, re-levelling on playback instead. A standard defines how loudness is measured; it does not set one number every service wants.

Our method. Deciding the destination first and mastering towards it with headroom, rather than making one master louder, is how this studio works, not a rule any platform sets.

References.

• International Telecommunication Union. Recommendation ITU-R BS.1770-5: Algorithms to measure audio programme loudness and true-peak audio level. itu.int. Defines measurement, not a universal platform target. Authoritative standard; does not prescribe one delivery level for every service.
• Audio Engineering Society. Loudness normalization overview. aes2.org. Gain-based playback normalisation and loudness concepts. Technical overview rather than a platform policy.
• Spotify for Artists. Loudness normalization on Spotify. support.spotify.com. Spotify normalises to about -14 LUFS with settings (-11/-14/-19) and exceptions. Not a universal cross-platform target; some players/devices do not normalise.
• YouTube. Recommended upload encoding settings. support.google.com. Current codec, frame-rate, colour-space and reference bitrate recommendations. Recommendations, not hard limits; source and destination still govern.