Reference on something bad, deliberately

Most listening now happens on a phone speaker, cheap earbuds, a car, or a smart speaker in a kitchen. None of those resemble a treated room with good monitors.

What to check and where

Phone speaker: does the bass line still read when there is no bass? Earbuds: is the top end fatiguing? Car: does the low end disappear or overwhelm? Kitchen speaker at conversational volume: can you still hear the words?

This is not about mixing on bad speakers. It is about checking that the decisions survive translation.

Mix on the good ones. Check on the ones people actually use.

What each system is actually testing

The checks are not arbitrary. Each playback system stresses a different property of the mix, and knowing which tells you what to do when it fails.

• Phone speaker. No low end below roughly five hundred hertz, and effectively mono. Tests whether your bass line has enough harmonic content to be implied when the fundamental is gone, and whether the mix survives collapse.
• Cheap earbuds. Exaggerated bass and a harsh presence region. Tests for fatiguing top end and for low end that is already too heavy.
• Car. Unpredictable modal low end and high background noise. Tests whether the vocal holds up when there is competition, and whether the bass is controlled enough to survive being reinforced.
• Smart speaker at conversational level. Tests intelligibility and whether the mix has enough internal contrast to work quietly.
• Laptop speakers. Similar to a phone but wider. A good fast check for whether anything important disappeared into the sides.

How to do it without ruining the mix

The failure mode is obvious and common: you check on earbuds, the bass is too much, you cut the bass, and now the mix is thin on everything else. You have mixed for the check.

The rule is that the monitors make the decisions and the other systems only raise questions. If something is wrong on three of the five, it is a mix problem and you fix it on the monitors. If it is wrong on one, it is that system, and you leave it alone.

Level-match when you compare. Quietly is not a system characteristic, it is a volume difference, and it will have you chasing something that is not there.

The checks generate questions. Only the monitors get to answer them.

Timing

Not at the end. By the time a mix is finished, the structural decisions that cause translation failures — arrangement density, stereo placement, low-end management — are expensive to change.

Check early, when the balance is roughly there and nothing is committed. Then again in the middle. Then once at the end as verification rather than as diagnosis. Three cheap checks during the work beat one expensive one after it.

The reference records

Do the same tour with two or three records you know are well made, in the same session, at matched level. This is the part people skip and it is what makes the whole exercise meaningful.

Your phone speaker makes every record sound thin. Without the comparison, you cannot tell whether your mix is thin or whether the speaker is. With it, the question answers itself in about four seconds, and half the time the answer is that your mix is fine and the system is bad.