Your vocal disappears into the beat: diagnose masking before you touch an EQ

The symptom is always the same. You push the vocal fader: it gets louder, not clearer. You push further: it jumps in front of everything else and the record falls apart. Between those two positions there is nothing that works.

This is not a level problem. If it were, the fader would already have solved it.

What masking actually means

The ear does not process each frequency in isolation. When two sounds occupy the same region of the spectrum, the louder one makes the quieter one hard to perceive — sometimes inaudible. The effect is local in frequency: a low sound will not mask an isolated high one, but it very efficiently masks whatever sits immediately around it.

Everything else follows from that. If masking is local, the correction has to be local too. The fader is global: it lifts the entire vocal, including the regions that had no problem at all. That is exactly why it never works — it answers a question nobody asked.

Diagnose before you move anything

Three listens are enough to know what you are fighting. Do them in this order, and fix nothing yet.

1. Alternate mutes. Mute the backing track, listen to the vocal alone. If it is clear, clean, intelligible — the problem is not in the vocal, and you were about to EQ it for nothing. Bring the track back, then mute elements one at a time: drums, bass, pads, guitars. The culprit announces itself; it is the one whose removal makes the vocal readable all at once.

2. The sweep. On the guilty element, move a narrow, heavily boosted bell through the spectrum while listening to the full mix, not the solo track. You are not hunting for "the frequency that sounds bad". You are hunting for the band where the vocal dies when you boost and comes back when you cut. That is where the fight is. Note the region, then flatten the EQ.

3. Mono. Collapse the mix to mono. Stereo placement genuinely helps the ear separate two sources; in mono that help disappears and the masking that was hiding shows itself. A vocal that holds up in mono will hold up anywhere.

Two landmarks for reading what you hear. Low down, the region usually called the low mids — roughly 200 to 500 Hz — is where nearly everything stacks up: the body of the voice, of the snare, of guitars, of pads. It is the first suspect whenever a mix sounds muddy. Higher up, speech intelligibility lives mostly between 1 and 4 kHz: that is where the consonants that separate one word from another live. A vocal you can hear but cannot understand is being masked up there, not down in the bass.

The real decision: who gives ground?

This is where an automatic process stops, and not for lack of computing power. A machine can detect that two sources overlap between 250 and 400 Hz. It cannot decide which of the two is carrying the song at that moment, because that is not in the signal. It is in the intent.

The question to ask is editorial: in this chorus, what does the listener need to hear first? If it is the lyric, the track steps back. If it is the riff, then accept that the vocal sits behind it and stop trying to win both.

Once the hierarchy is settled, one practical rule: taking away from the accompaniment costs less than adding to the vocal. A narrow dip in a dense, wide source goes unnoticed; a bump on an exposed, centred source is heard immediately, and it usually brings its own share of harshness with it.

The moves, in order

1. Arrangement first. Two instruments playing the same note in the same register through the verse is a writing problem, not a mixing problem. A pad you delete, a guitar moved up an octave, a vocal double you drop: that fixes more masking than any EQ.
2. Static balance next. Rebuild an honest balance with faders only, no processing. A lot of "masking" is just instruments that are too loud.
3. Space. Spread sources across the stereo field wherever the music allows. The centre is already occupied by the vocal, the bass and the kick — put nothing there that can live somewhere else.
4. Subtractive EQ, finally, on the accompaniment, in the band you found with the sweep. Narrow if you are fixing a specific clash, wide if you are redefining a role.
5. Dynamic processing as a last resort: a dip that only opens while the vocal sings — dynamic EQ, or compression triggered by the vocal. Powerful, but it is a bandage over an arrangement decision you did not want to make. Use it knowing that.

What this will not fix

A mediocre vocal take is still mediocre once the spectrum is clear: masking was not the problem, it was hiding it. A purchased instrumental that arrives already crushed at mastering will not leave you much room — you can carve into it, you cannot give back dynamics it no longer has. And a song whose arrangement asks three elements to play the same role at the same moment will never be fixed in the mix. Nobody can fix that in the mix.

In those cases the useful move is to say so. Not to open one more plugin.