Findings

What the chart says, in words.

The chart lets you look for yourself. This page is the short version of what I saw in it.

1 · Does paying more help?

Price barely predicts sound.

If money bought sound, the dots would climb as they get pricier. They do not. For 3.5 mm IEMs the link between sound score and price has an R² of 0.001. For Type-C it is 0.008. Both are close to zero.

  • 3.5 mm

    KZ EDX Pro 2 costs ₹1,190 and scores 86.0. NF Acous NM20 costs ₹7,499 and scores 54.7.

  • Type-C

    KZ EDX Pro 2 costs ₹1,490 and scores 86.0. Final VR3000 costs ₹7,999 and scores 73.8.

See price against sound on the chart

2 · By budget

The typical score is flat across budgets.

Median Harman score by price band and plug type
Price band3.5 mmType-C
Under ₹2,00078.7median of 9279.2median of 24
₹2,000 to ₹5,00077.1median of 10080.4median of 38
₹5,000 to ₹10,00078.1median of 8980.0median of 17

Median Harman score of measured IEMs in each band. The gaps between bands are smaller than the measurement noise.

Read scores as fuzzy. Rigs, eartips and insertion depth change measurements; treat +/-3 points as noise.

3 · Find your sound

Warm or bright is taste. Distance from the target is not.

Tone says which way an IEM leans from the Harman target. Warmer means more bass. Brighter means more treble. The sound score falls away on both sides.

Not everyone likes the target. In the Harman research, about 72% prefer Harman-like tuning; about 28% prefer ~2 dB less bass and 1-2 dB more treble. If that sounds like you, favour IEMs with a slightly positive tilt.

See warm against bright on the chart

4 · How the score works

One number, three parts, all adjustable.

  • 60

    Sound

    The Harman in-ear model predicts how much listeners will like the tuning, from the measured frequency response.

  • 30

    Hardware

    Detachable cable, microphone, and whether a phone can drive it. As claimed by the store.

  • 10

    In the box

    Eartips, a case, a cable, an adapter. As claimed by the store.

Explore and set your own weightsSee how I did it

5 · What the data cannot tell you

The honest bit.

  • 345 of 415 IEMs are measured by at least one independent person.
  • 15 are collab editions where only the base model is measured. They are marked Proxy and lose points.
  • 55 have no measurement at all. They are marked Buyer beware, and are missing from any chart that needs a sound score.
  • 210 were in stock when the data was built on 2026-11-11. Sold-out ones stay listed, with a flag.
  • 19 were removed as discontinued. They are listed here.

See and download the dataRead the write-up (PDF)Find me on LinkedIn