# Headphones and earbuds

Decide wearing style and listening place first; those two answers remove most of the catalogue. Then care about tuning and seal, not codecs or driver buzzwords: measured frequency response predicts whether people like a headphone and price does not. Wireless models are consumables with a glued-in battery, so check battery hours, repairability and cable playback.

## How it works

A headphone is a small loudspeaker next to or inside your ear. A driver turns the signal into moving air; how evenly it does so across pitches is its frequency response. That shape, far more than driver technology or price, decides whether you like the sound. Harman's 31-model, 130-listener test found ratings fall the further a headphone strays from a preferred target curve. Two surveys of over 400 headphones found no link between price and frequency response. Sealed cups and tips block outside noise passively and hold bass in; open backs and open-ear designs let sound through both ways. Active noise cancelling adds microphones and an inverted signal that cancels steady low rumble, which passive sealing handles worst. Wireless models add a battery, Bluetooth radio and codec; the codec sets delay far more than fidelity. Wired models leave amplification to the source, so impedance and sensitivity matter only there. [1] [29] [30] [32] [34] [33] [11] [25] [39] [40]

| Family | What it is | Suits |
|---|---|---|
| Dynamic driver | A cone or dome on a voice coil, like a tiny loudspeaker. Physical bass, warm timbre, slightly softer edges on fast passages. | Almost every wireless headphone and earbud, and most wired ones. |
| Planar magnetic | A thin flat film with a printed conductor between magnet arrays, so the whole surface moves at once. Low distortion and fast, but heavy, seal-sensitive and harder to drive. | Wired open-back headphones for home listening. |
| Balanced armature | A tiny pivoting reed in a coil, small enough to stack several in one earpiece. Fast and detailed with tighter, less physical bass. | In-ear monitors, often several per ear with a crossover. |
| Hybrid | A dynamic driver for bass plus one or more armatures for mids and treble. Lives or dies by its crossover, the circuit that splits the signal between the drivers. | Mid-range and up in-ear monitors. |
| Electrostatic | A charged film between two grids, needing a 230 or 580 volt charge and its own high-voltage amplifier. Very fast, weak deep bass at high volume. | A small high-end niche. Needs dedicated hardware. |
| Bone conduction and open-ear | A transducer on the cheekbone, or a small speaker aimed at the ear canal, leaving the ear open. Awareness of traffic, weak bass, audible leakage. | Running, cycling and swimming, and some kinds of hearing loss. |

## How an expert narrows the choice

```mermaid
timeline
  section 1 · Wearing style
    wired in-ear / true wireless in-ear : ask Ear tips in the box
    open-ear (clip or hook) / bone conduction : ask Sweat and water rating : skip How it treats outside noise, Noise reduction, measured, Ear tips in the box
  section 2 · How it treats outside noise
    active noise cancelling : ask Noise reduction, measured, Battery life
    sealed without cancelling / open (hears everything) : skip Noise reduction, measured
  section 3 · Connection
    true wireless with charging case : ask Battery life, What can be replaced, Multipoint, Bluetooth codecs : skip Impedance, Sensitivity, Driver type, Works with a cable
    Bluetooth headphones : ask Battery life, Multipoint, Works with a cable, What can be replaced, Bluetooth codecs : skip Impedance, Sensitivity, Driver type
    2.4 GHz dongle : ask Delay, Battery life : skip Impedance, Sensitivity, Driver type
    wired 3.5 mm / wired USB-C or Lightning : ask Impedance, Sensitivity, Driver type : skip Battery life, Bluetooth codecs, Multipoint, Works with a cable, Delay
```

1. **Wearing style.** In-ear buds, over-ear cups, or open-ear (hear traffic)?
   Wearing style is what every expert asks first and a buyer can always answer. It cuts the catalogue into pools that do not compete and fixes what later questions mean: tips only for in-ears, open-back only for over-ears.
   If wired in-ear or true wireless in-ear: ask Ear tips in the box.
   If open-ear (clip or hook) or bone conduction: ask Sweat and water rating; skip How it treats outside noise, Noise reduction, measured, Ear tips in the box.
   Skip if already known: memory.headphone_form_factor, an earlier answer naming earbuds, over-ears or open-ear.

2. **How it treats outside noise.** Noisy commute or office, quiet home, or outdoors (must hear traffic)?
   This maps onto the three acoustic designs. Open designs leak both ways and are useless on a train; cancelling is wasted at a quiet desk and costs battery. In over-ears it settles open-back versus closed-back, the real expert split.
   If active noise cancelling: ask Noise reduction, measured, Battery life.
   If sealed without cancelling or open (hears everything): skip Noise reduction, measured.
   Skip if already known: memory.listening_environment, memory.commute_mode.

3. **Connection.** Phone over Bluetooth, a cable, or a PC or console (delay matters)?
   This removes less than the first two questions but changes everything downstream: wireless brings battery, codec and multipoint; wired brings impedance and the amplifier question. A gaming or video answer routes to 2.4 GHz dongles and wired.
   If true wireless with charging case: ask Battery life, What can be replaced, Multipoint, Bluetooth codecs; skip Impedance, Sensitivity, Driver type, Works with a cable.
   If Bluetooth headphones: ask Battery life, Multipoint, Works with a cable, What can be replaced, Bluetooth codecs; skip Impedance, Sensitivity, Driver type.
   If 2.4 GHz dongle: ask Delay, Battery life; skip Impedance, Sensitivity, Driver type.
   If wired 3.5 mm or wired USB-C or Lightning: ask Impedance, Sensitivity, Driver type; skip Battery life, Bluetooth codecs, Multipoint, Works with a cable, Delay.
   Skip if already known: memory.phone_has_headphone_jack, memory.gaming_console, an earlier answer about gaming or video use.

## What to surface

### Wearing style

Where the headphone sits: over the ear, on the ear, inside the ear canal, or outside the ear leaving the canal open. It sets the physics of everything else. Over-ears can be open-backed and carry big drivers and long battery life. In-ears depend on a tip seal for bass and isolation and are the only style with balanced armatures. Open-ear and bone conduction leave you able to hear traffic but give up most bass and leak enough that the person next to you hears your music. [40] [39] [16] [19]

Values: over-ear, on-ear, wired in-ear, true wireless in-ear, neckband, open-ear (clip or hook), bone conduction.
Question to ask: In-ear buds, over-ear cups, or open-ear (hear traffic)?

Used as: top-level split, filter. Data: on the listing, backfill cost none. Look for it as: In the title and the first image on every listing. The scraper already captures it as form_factor on the headphone add-on.
When the title is vague, the product photo settles it.

### How it treats outside noise

Whether the design cancels outside noise electronically, blocks it passively with a sealed cup or tip, or lets it through on purpose. Open-back headphones and open-ear buds sound spacious and keep you aware of the room, but they isolate nothing and leak music outward, so they belong at home or on a run. Sealed designs block speech-range noise reasonably well on their own. Active noise cancelling is the only thing that removes low rumble from planes, trains and air conditioning, which passive sealing handles worst, and it only works on top of a good seal: unsealed earbuds with cancelling remove far less. [32] [4] [35] [39]

Values: active noise cancelling, sealed without cancelling, open (hears everything).
Question to ask: Noisy commute or office, quiet home, or outdoors (must hear traffic)?

Used as: top-level split, filter. Data: derived from listing fields, backfill cost low. Look for it as: Combine the add-on's anc_present with form_factor and the title words open-back, closed-back or open-ear. Open-back is almost always in the title of wired over-ears.
If anc_present is true the value is active noise cancelling. If the form factor is open-ear or bone conduction, or the title says open-back, it is open. Everything else is sealed. Read the manufacturer spec page for the few over-ears whose title omits the enclosure.
Chart: Noise reduction by band.

### Connection

How the headphone receives sound: a cable, ordinary Bluetooth, a dedicated 2.4 GHz radio dongle, or true wireless earbuds with a charging case. It decides which other facets exist. Wireless brings battery life, codec, multipoint and a glued-in battery that ages. Wired brings impedance and sensitivity and never runs out. Dongles exist because Bluetooth measured 160 to 370 ms of delay in two labs, which is visible in games and video, while the one dongle headset measured so far came in at 23 ms over the air and 13 ms by cable. [6] [7] [15]

Values: true wireless with charging case, Bluetooth headphones, 2.4 GHz dongle, wired 3.5 mm, wired USB-C or Lightning.
Question to ask: Phone over Bluetooth, a cable, or a PC or console (delay matters)?

Used as: top-level split, filter. Data: on the listing, backfill cost none. Look for it as: The electronics standard's connectivity.connection_types already captures bluetooth, usb_c, audio_3_5mm and lightning. A 2.4 GHz dongle shows up as a title phrase on gaming headsets and in the box contents.
Chart: Latency scale.

### Multipoint

Whether the headphone stays connected to two devices at once, so a call on the phone can interrupt music from the laptop without re-pairing. It is the feature office users miss most and the one listings bury. Two devices is the usual limit, only one plays at a time, and on some models enabling it once forced the lower-quality codec until a firmware update. Experts check the manual rather than the box. [13] [15] [14]

Question to ask: Switch between laptop and phone without re-pairing?

Used as: filter. Data: needs research, backfill cost low. Look for it as: Rarely a structured field on retail listings and often absent from the title. Found on manufacturer spec pages and in the manual, phrased as multipoint or connect to 2 devices simultaneously.
Manufacturer page or manual for anything mid-range and up. Treat absence on a budget product as no.

### Sweat and water rating

The IP code from IEC 60529, where the second digit says how much water the product survives and an X means that position was not tested. It separates gym and rain use from desk use. IPX4 means splashes from any direction, 5 means jets, 7 means a brief dip to about a metre. Apple rates recent AirPods IP54 or IP57 and still says they are not for swimming or showering, and that the resistance wears off with use. No stated rating on a sport product is a warning. [17] [16] [18]

Values: none stated, IPX4 (splashes), IP54 (dust and splashes), IPX5 or IP55 (jets), IPX7 or IP57 (brief immersion), IP68.
Question to ask: Will you sweat in them or wear them in the rain?

Used as: filter. Data: on the listing, backfill cost low. Look for it as: On the spec page and usually in the bullets as water resistant or IPX4. The electronics standard already captures it as ingress_protection.
Manufacturer spec page. Record none stated rather than guessing from words like sweatproof.

### Works with a cable

Whether a wireless headphone has an analogue input and still plays when the battery is flat or Bluetooth is banned. A flight, a dead battery or an old games console turns a wireless-only headphone into a hat. Sony's flagship plays through its cable with the power off, without cancelling; Sennheiser ships a cable in the box. Most true wireless earbuds and some premium over-ears have no input at all. [15] [20]

Question to ask: Must they still play on a flight or with a flat battery?

Used as: filter, shown on the card. Data: on the listing, backfill cost medium. Look for it as: Spec page line for a 3.5 mm input or an included audio cable. Whether it plays with the power off is only in the manual.
Cable input is on the spec page. Power-off playback needs the manual, so backfill it for the top N over-ears and otherwise record cable input only.

### Bass level against the preferred curve

How many decibels the headphone's measured bass sits above or below the Harman target, a curve that 64 percent of listeners in Harman's tests preferred. Tuning is the single best predictor of whether people like a headphone, and bass level is what separates the three listener groups Harman found: most like the target, 15 percent want 3 to 6 dB more bass, 21 percent want 2 to 3 dB less. A headphone far above the target sounds thick and buries voices; one far below sounds thin. Knowing the buyer's group is worth more than any driver or codec spec. [1] [2] [3] [38] [37] [5]

Range: Listeners who want more bass want 3 to 6 dB above the target below 300 Hz. Those who want less want 2 to 3 dB below it. Products spread wider than that in both directions, but no source gives category-wide bounds.
Question to ask: Strong, thumping bass, or an even, neutral sound?

Used as: shown on the card, sortable, filter. Data: needs research, backfill cost high. Look for it as: Never on a listing. Comes from measurement databases: RTINGS, Crinacle's graph tool, the squig.link sites built on CrinGraph, and SoundGuys.
Backfill the top few hundred products from one measurement database per form factor and lazy-load the rest. Do not mix databases: rigs and targets differ, so a figure is only comparable to others from the same source. Record the database used alongside the number.
Chart: Frequency response against a target.

### Battery life

Hours of playback on one charge with noise cancelling on, and for true wireless earbuds the extra hours the case adds. It is the number on the box, but the condition matters: Sony's flagship runs 30 hours with cancelling on and 40 off on AAC, 26 or 36 on LDAC, and 28 or 38 on SBC, so one product carries six figures. Reviewers test at a fixed volume because louder playback drains faster. Earbuds are the ones to watch: 7.5 to 8 hours per charge on Apple's and Sennheiser's flagships, 24 to 30 with the case. [15] [16] [20] [36] [41]

Range: True wireless earbuds from Apple and Sennheiser give 7.5 to 8 hours per charge and 24 to 30 with the case, cancelling on where stated. Wireless over-ears give 30 hours with cancelling on (Sony) to 60 (Sennheiser).
Question to ask: How many hours between charges do you need?

Used as: filter, sortable. Data: on the listing, backfill cost none. Look for it as: In the title and the first bullet on wireless listings. The add-on captures single_charge_hours and total_battery_with_case_hours. Record whether the figure is with cancelling on, since brands quote both.
When the listing gives one number, treat it as the best case the brand chose to print. The spec page or manual carries the on and off table.
Chart: horizontal range bar, single charge in solid and case-extended in hatched.

### Delay

The time between a sound leaving the source and reaching your ear, which decides whether lips and gunfire line up with the picture. Ordinary Bluetooth is slow: RTINGS measures about 260 ms on SBC for earbuds and headphones alike, SoundGuys measured 308 to 369 ms across Android phones, and people start to notice between 40 and 80 ms. Gaming headsets fix this with a 2.4 GHz dongle: the one RTINGS has measured came in at 23 ms over the air and 13 ms by cable. aptX Low Latency promises 32 ms by spec if both ends support it. [6] [7] [9] [8]

Range: 13 to 369 ms. A cable measured 13 ms and a 2.4 GHz dongle 23 ms, both on the single gaming headset RTINGS and Audeze have measured. aptX Low Latency is 32 ms by spec. Ordinary Bluetooth measured 160 ms on one earbud on Windows and 260 to 369 ms as lab averages.
Question to ask: Gaming or video where lip-sync matters, or just music?

Used as: filter, shown on the card, sortable. Data: derived from listing fields, backfill cost medium. Look for it as: Derive a class from connection and codec_support: wired or dongle is low, aptX Low Latency or Adaptive is medium, everything else on Bluetooth is high. Measured values come from RTINGS reviews.
The class is free from existing fields. Backfill measured milliseconds from RTINGS for the top N gaming and video products only.
Chart: Latency scale.

### What can be replaced

Whether the parts that wear out, ear pads, tips and above all the battery, can be swapped, and by whom. Battery wear is what ends a wireless earbud: capacity halves after a few hundred cycles and most fail at two to three years, and almost none can be opened. iFixit scored every AirPods 0 out of 10 until the AirPods 5 reached 2, while Fairphone's Fairbuds scored 10 with a coin cell you swap with a fingernail. Wired open-backs sell replacement pads and cables as spares and last decades. [22] [23] [24] [19] [20]

Values: sealed with nothing replaceable, ear pads or tips replaceable, battery replaceable by a repair shop, battery user-replaceable.
Question to ask: Repairable, or fine replacing them in two or three years?

Used as: shown on the card, filter. Data: needs research, backfill cost high. Look for it as: Replacement pads appear as accessories on manufacturer pages. Battery replaceability is only in iFixit teardowns or a brand's own repair page.
Backfill the top N from iFixit scores and manufacturer spare-parts pages, lazy-load the rest. Default true wireless earbuds to sealed unless a source says otherwise; default wired over-ears with a listed spare pad to pads replaceable.

### Noise reduction, measured

How many decibels of outside noise the headphone removes in a lab, with cancelling on, usually reported by band because low rumble and speech behave differently. Listings say noise cancelling as if it were one thing. It is not: the best over-ears cut bass-range noise by more than 15 dB, while unsealed earbuds with cancelling remove far less, and price barely predicts the result across 205 products SoundGuys tested. Passive sealing contributes too, and a foam tip alone blocks 15 to 25 dB. [4] [35] [31] [34]

Range: The best over-ears and sealed earbuds remove more than 15 dB in the bass range. Open earbuds with cancelling remove far less. A foam tip alone blocks 15 to 25 dB. No source gives a category-wide ceiling, so none is stated.

Used as: shown on the card, sortable. Data: needs research, backfill cost high. Look for it as: Not on listings. RTINGS publishes per-band attenuation; SoundGuys publishes a percentage of perceived loudness removed.
Backfill the top N cancelling products from RTINGS or SoundGuys and lazy-load the rest. Store the source and its unit, because the two are not interchangeable. For the long tail, show the noise_handling class instead of a number.
Chart: Noise reduction by band.

### Bluetooth codecs

The compression formats the headphone can receive over Bluetooth, which must also be supported by the phone or the link drops to the basic SBC codec. Codecs are the most advertised and least audible spec. SBC runs at up to about 328 kbps, AAC 256 to 320, aptX 352, aptX HD 576 and LDAC up to 990, but SoundGuys found 990 kbps LDAC rarely engages and pushes Bluetooth to the edge of stability. What a codec reliably changes is delay and ecosystem fit: aptX Low Latency and Adaptive cut lag, iPhones use AAC, and LC3 beats SBC at half the bitrate and enables Auracast broadcasts. [8] [9] [10] [11] [12]

Values: SBC, AAC, aptX, aptX HD, aptX Adaptive, LDAC, LHDC, LC3 (LE Audio).

Used as: shown on the card, filter. Data: on the listing, backfill cost low. Look for it as: Manufacturer spec pages list them; Apple does not. Retail titles mention LDAC or aptX as a selling point. The add-on already captures codec_support.
Listing is justified by the add-on's existing codec_support field and by Sony's and Sennheiser's spec pages, which list codecs. Apple is the exception and publishes none, so default Apple products to SBC and AAC, which every modern product supports.
Chart: Latency scale.

### Driver type

The kind of transducer that makes the sound: dynamic, planar magnetic, balanced armature, a hybrid of those, electrostatic, or bone conduction. It predicts the character more than the quality. Dynamic drivers give physical bass and slightly soft edges; armatures are fast and detailed with lighter bass; planars are low-distortion and fast but power-hungry and seal-sensitive; hybrids depend on their crossover, the circuit that splits the signal between drivers; electrostatics need their own high-voltage amplifier. None of this overrides the tuning, which is why experts read the graph before the driver list. [29] [30]

Values: dynamic, planar magnetic, balanced armature, hybrid (dynamic plus armature), electrostatic, bone conduction.

Used as: shown on the card, filter. Data: derived from listing fields, backfill cost medium. Look for it as: Wireless listings give only a driver size in millimetres, which means dynamic, so the value is inferred. Spec sheets of wired headphones and in-ear monitors state the type and count, such as 1DD plus 4BA.
Infer dynamic when only a diameter is given. Read manufacturer pages for planar and armature counts on wired headphones and in-ear monitors.

### Impedance

The electrical resistance the headphone presents to the source, in ohms, which with sensitivity decides whether a phone can drive it and how much the source colours the sound. High-impedance headphones like the 300 ohm HD 600 need more voltage than a phone gives to play loud, so they want an amplifier. Low-impedance earphones play loud from anything but are most sensitive to the source: when its output impedance exceeds an eighth of the headphone's, the tuning bends, by up to 12 dB on some multi-driver earphones. Wireless headphones carry their own amplifier, so it only matters in cable mode, where Sony quotes 48 ohms on, 16 off. [19] [15] [20] [25] [26] [27]

Range: 16 to 470 OHM. Wireless headphones in cable mode and most consumer wired models sit at 16 to 60 ohms. Open-back studio designs sit around 300 ohms, and Sennheiser's Momentum 4 reports 470 ohms when powered on.

Used as: shown on the card. Data: needs research, backfill cost low. Look for it as: On every wired spec sheet and in the manual of wireless over-ears, as impedance in ohms. Absent from true wireless earbuds and from Apple's pages, so it is not on most listings in the category.
Manufacturer spec sheet or manual; cheap for the wired cohort. Leave blank for true wireless; it does not apply.
Chart: Drivability scale.

### Ear tips in the box

How many tip sizes and which materials ship with an in-ear product, which decides whether most ears can get a seal. For in-ears the seal makes the bass and the isolation: a leaking tip loses 10 dB or more in the sub-bass region and lets noise straight in, which is also why cancelling fails on unsealed buds. Foam blocks 15 to 25 dB against 10 to 15 for silicone but dulls treble and lasts one to three months. Apple ships five silicone sizes; cheap buds ship three or one. [31] [5] [16] [35]

Values: one size, three sizes, four or more sizes, foam tips included.

Used as: shown on the card. Data: on the listing, backfill cost low. Look for it as: The in-the-box list on the spec page, phrased as ear tips in five sizes. Only applies to in-ear and true wireless.
Spec page box contents. Leave blank for over-ears and open-ear.

### Sensitivity

How loud the headphone plays for a given input, quoted either per milliwatt of power or per volt, which are different numbers for the same product. With impedance it answers the only question that matters for wired use: will it get loud from my phone. Brands mix the units, so Sennheiser's 97 dB at 1 volt and Sony's 102 dB per milliwatt cannot be compared until converted. Experts convert to decibels per volt, because the source supplies volts. [19] [15] [20] [28]

Range: The cited spec sheets quote 97 dB at 1 V (Sennheiser HD 600), 100 to 102 dB/mW (Sony WH-1000XM5) and 106 dB SPL (Momentum 4). Check the unit before comparing.

Used as: shown on the card. Data: needs research, backfill cost low. Look for it as: Spec sheet line labelled sensitivity, SPL (sound pressure level) or efficiency, with its unit. Sony quotes dB/mW, Sennheiser dB at 1 V or dB SPL at 1 kHz. Absent from true wireless and Apple pages.
Manufacturer spec sheet or manual for wired and over-ear products. Store the unit with the number so the derived dB per volt figure can be computed. Leave blank for true wireless.
Chart: Drivability scale.

## Numbers worth computing

| Metric | Formula | Why |
|---|---|---|
| Sensitivity per volt (dB/V) | `sensitivity_db_per_mw + 10 * log10(1000 / impedance)` | Puts a 300 ohm headphone and a 16 ohm one on one scale. Sennheiser quotes per volt and Sony per milliwatt, so the spec sheets cannot be compared until converted, and the source supplies volts, not milliwatts. Skip the conversion when the brand already quotes per volt. [19] [15] [28] |
| Highest safe source impedance (OHM) | `impedance / 8` | The one-eighth rule: a source with output impedance above this bends the headphone's tuning, most severely on multi-driver earphones. A 16 ohm earphone wants a source under 2 ohms, which rules out many laptops and some phones. [25] [26] |
| Hours lost to cancelling (h) | `battery_life_anc_off - battery_life_anc_on` | Shows what the cancelling costs and exposes brands that print only the off figure. On Sony's WH-1000XM5 it is 10 hours on each of SBC, AAC and LDAC, and LDAC itself costs another 4 against AAC. [15] |

## What you give up

- Open-back and open-ear designs sound spacious and leave you aware of the room, and that feeling is noticeable even with no music playing. The cost is total: they block nothing and the person next to you hears your music, so they belong at home or on a run, never on a train. [32] [40] Board branch: "See more open designs for home listening".
- Cancelling costs battery, 10 hours of 40 on Sony's flagship, and it only works on top of a seal, so the strongest cancelling comes from sealed over-ears and snug tips, not from the earbuds that sit loosely for comfort. [15] [35] Board branch: "See more with stronger noise cancelling".
- True wireless earbuds are the most convenient form ever sold and the shortest-lived: the battery halves after a few hundred cycles, fails at two to three years, and on almost every model is glued in. Wired headphones with spare pads and cables last decades. [24] [23] [19] Board branch: "See more that can be repaired".
- The only ways under 40 ms are a cable, a 2.4 GHz dongle that occupies a port and will not pair with a phone, or a codec that both ends must support. Pick ordinary Bluetooth and you accept 160 to 370 ms, which is fine for music and visible in games. [6] [7] [8] Board branch: "See more low-delay options for gaming".
- Push the bitrate up to 990 kbps LDAC and the link drops packets in a busy room, the phone rarely chooses it anyway, and most listeners will not hear what it added. Lower bitrates hold the connection. [11] Board branch: "See more with high-bitrate codecs".
- Open-ear and bone conduction keep your ears free for traffic and let you talk, and bone conduction even works for some hearing loss and for swimming. Bass barely registers, the transducer leaks into the air around it, and the clamp can ache after a couple of hours. [39] [40] Board branch: "See more open-ear for running and cycling".
- A high-impedance headphone is indifferent to what it is plugged into, because the source's output impedance is a small fraction of its own, but it needs an amplifier to get loud. A low-impedance earphone plays loud from anything and takes on the character of whatever drives it. [25] [26] [27] Board branch: "See more that play loud from a phone".
- Bass above the target is the easiest way to make a headphone exciting in a shop, and 15 percent of listeners want 3 to 6 dB of it. Harman's own caution is that adding too much bass beyond the target may alienate many listeners, because the group that wants more bass is small and the group that wants less includes the trained listeners. [1] [2] Board branch: "See more neutral tunings".
- Foam tips seal more ears and block 15 to 25 dB of noise against 10 to 15 for silicone, at the price of a duller top end and a tip that lasts one to three months instead of six to twelve. [31]

## How to draw it

**Frequency response against a target** (other, shows Bass level against the preferred curve). A line graph of loudness against pitch on a log scale, with the product's curve drawn over a preferred target line. A buyer sees at once whether the bass sits above or below the line and whether the treble has a spike, which no table of specs conveys. It shows the flavour, not the quality: a flat gap to the target is what most people prefer, but the graph cannot show detail or imaging. Used today by Crinacle's graph tool and the squig.link sites built on CrinGraph, RTINGS, SoundGuys, and Harman's own research papers. [38] [36] [1] [37]

**Noise reduction by band** (bar, shows Noise reduction, measured, How it treats outside noise). Three bars for bass, mid and treble noise removed, with cancelling off and on. It shows that cancelling earns its keep on low rumble while a plain seal already handles speech, and it makes an open design's near-zero bars obvious next to a sealed one. Used today by RTINGS reports attenuation by band with ANC off and on; SoundGuys reports a single percentage of perceived loudness removed. [4] [36]

**Latency scale** (scale, shows Delay, Connection, Bluetooth codecs). One horizontal log scale of milliseconds with marks at 20 (nobody notices), 80 (many notice) and 150 (everyone notices), and the product placed on it by its connection or measured value. Wired and dongle cluster at the left, ordinary Bluetooth far right. This is a proposal, nobody draws it this way yet. [6] [7]

**Drivability scale** (scale, shows Impedance, Sensitivity). Sensitivity per volt on one scale, with zones for plays loud from a phone, wants a small USB amplifier, and needs a desktop amplifier. It collapses two numbers in two units into the single question a wired buyer has. This is a proposal, nobody draws it this way yet. [28] [27]

## Words you will meet

- **Active noise cancelling (ANC)**. Microphones pick up outside noise and the headphone plays an inverted copy that cancels it, which works best on steady low rumble.
- **Transparency or ambient mode**. The microphones pipe outside sound in so you can hear a conversation without taking the headphones off.
- **Open-back**. An over-ear whose cups let air and sound through, which sounds spacious but isolates nothing and leaks music outward.
- **Closed-back**. An over-ear with solid cups that block outside sound to some degree and keep music in.
- **Open-ear**. A bud that rests outside the ear canal instead of sealing it, so you hear traffic and the bud leaks sound.
- **Bone conduction**. A transducer on the cheekbone that vibrates sound into the inner ear, leaving the ear canal free.
- **In-ear monitor (IEM)**. A wired earphone that seals inside the ear canal, originally for musicians on stage.
- **True wireless**. Two independent earbuds with no cable between them, charged in a case.
- **Seal**. The airtight fit between tip and ear canal, or pad and head, that holds bass in and noise out.
- **Frequency response**. How loud a headphone plays each pitch from deep bass to high treble, drawn as a curve.
- **Harman target**. A frequency response curve from Harman's listening research that most listeners prefer, used as the reference line on graphs.
- **Tonality**. How the balance of bass, mids and treble sounds, which the frequency response graph shows.
- **Technicalities**. Everything about the sound that is not tonal balance, such as detail and the sense of where instruments sit, which graphs do not show.
- **Soundstage**. How wide and far away the music seems to come from, strongest on open-back headphones.
- **Driver**. The part that turns electricity into sound, the headphone's tiny loudspeaker.
- **Planar magnetic**. A driver made of a flat film with a printed conductor between magnets, so the whole surface moves together.
- **Balanced armature**. A tiny pivoting reed driver small enough to fit several in one earphone.
- **Codec**. The compression format that squeezes audio over Bluetooth, which both the phone and the headphone must support.
- **SBC and AAC**. The baseline codec every Bluetooth product supports, and the one iPhones use.
- **aptX, LDAC and LHDC**. Higher-bitrate codecs from Qualcomm, Sony and Savitech that need a supporting phone, mostly Android.
- **LE Audio and LC3**. The newer Bluetooth audio system and its codec, which sounds better than SBC at half the bitrate and allows broadcasts.
- **Auracast**. A Bluetooth broadcast so one source can play to any number of headphones at once, such as a TV in a gym.
- **Multipoint**. Staying connected to two devices at once, with only one playing at a time.
- **Latency**. The delay between the source playing a sound and you hearing it.
- **2.4 GHz dongle**. A small USB radio receiver that replaces Bluetooth with a faster private link, used by gaming headsets.
- **Impedance**. The headphone's electrical resistance in ohms, which with sensitivity decides whether a phone can drive it.
- **Sensitivity**. How loud the headphone gets for a given input, quoted per milliwatt or per volt.
- **Output impedance**. The source's own resistance, which should be under an eighth of the headphone's or the tuning bends.
- **DAC and amplifier**. The chips that turn digital audio into a signal strong enough for a headphone; built into phones and wireless headphones, bought separately for hard-to-drive wired ones.
- **IP rating**. A two-digit code for dust and water protection, where X means that digit was not tested.
- **Hi-Res Audio Wireless**. A Japan Audio Society certification mark for codecs such as LDAC, not a guarantee of audible improvement.

## Where we made a call

- Tuning is the single best predictor of whether a buyer will like the sound, and price predicts neither tuning nor liking. (confidence high). Harman's model predicts listener ratings from the frequency response error with a correlation of 0.86, and two surveys of over 400 headphones found no price to response correlation. This is why tuning_vs_target outranks driver_type and codec_support despite being researched. [1] [2]
- Bluetooth codec differences are rarely audible; codecs matter for delay and for matching the phone, not for fidelity. (confidence medium). Sources disagree. Marketing and the Hi-Res Wireless certification say otherwise. We side with SoundGuys' analysis that 990 kbps LDAC rarely engages, is unstable and is not true hi-res, and with blind-test reports read only as search excerpts. Differences at the aptX HD level are plausible for some listeners, so medium. [11] [10] [12]
- Noise cancelling performance does not track price, and a seal is the precondition for it. (confidence high). SoundGuys found an incredibly weak price relationship across 205 products and that unsealed earbuds with cancelling remove little. RTINGS' band data shows the bass gain that cancelling adds on top of passive isolation. [35] [4]
- There is no agreed impedance threshold for needing an amplifier, so we rank wired drivability by sensitivity per volt instead. (confidence medium). Sources disagree. The Master Switch says anything above 100 ohms needs an amplifier; Wikipedia's worked example says a 120 ohm headphone from an iPod would most likely not benefit from one; the headphones.com community says impedance alone does not predict loudness and points to dB/V. We follow the community and derive dB/V. [27] [26] [28]
- Multipoint and high-bitrate codecs were mutually exclusive on some models and now depend on firmware, so the manual is the only reliable source. (confidence medium). Sources disagree. SoundGuys wrote that multipoint only works with AAC. Sony's 2.0.2 firmware for the WH-1000XM5 later enabled LDAC during multipoint, according to a report we could read only as a search excerpt. The SoundGuys line is dated rather than wrong for its time. [13] [14] [15]
- True wireless earbuds are consumables with a two to three year battery life unless the battery is replaceable. (confidence medium). Sources disagree. iFixit quotes about 18 months before the battery stops holding a charge; MakeUseOf says the two or three year mark after a few hundred cycles. We use two to three years and treat 18 months as the heavy-use case. Both agree the battery is glued in on almost every model. [23] [24] [22]
- A frequency response figure is only comparable to others from the same measurement database. (confidence medium). RTINGS uses a Brüel and Kjær 5128 rig, Harman an IEC 60318-1 ear simulator with its own ear-shaped mould, and SoundGuys scores against its own house curve rather than Harman. Head-Fi's rig-differences thread could not be fetched. The tuning_vs_target backfill note therefore records the source database with the number. [5] [36] [1] [37]
- The listing-field hints rest on manufacturer spec pages, because every retail listing we tried was blocked. (confidence low). Amazon returned only a page head carrying battery hours and the cancelling claim; seven other retailers returned errors or a captcha. Every facet marked listing therefore rests on Apple, Sony and Sennheiser spec pages plus fields the add-on already captures, so check the hints against the scraper's fill rates. [21] [16] [15]
- Bluetooth delay is 160 to 370 ms depending on the lab and the phone, and we give the span rather than one number. (confidence high). Sources disagree. RTINGS averages about 260 ms on SBC; SoundGuys measured 308 to 369 ms across four Android phones. The 160 ms floor is one earbud on Windows and the 23 ms dongle figure one gaming headset. Labs differ by method and source device, not product, so one figure would mislead. [6] [7]
- The bass offsets that define the listener groups are 3 to 6 dB above and 2 to 3 dB below the target. (confidence high). Sources disagree. Olive's 2022 article says 4 to 6 dB more and 2 dB less; SoundStage Solo's summary of the same research says 3 to 6 and 2 to 3. We use the wider span. [1] [2]

## References

1. [Sean E. Olive, The Perception and Measurement of Headphone Sound Quality: What Do Listeners Prefer? Acoustics Today, Spring 2022](https://acousticstoday.org/wp-content/uploads/2022/03/The-Perception-and-Measurement-of-Headphone-Sound-Quality-What-Do-Listeners-Prefer-Sean-E.-Olive.pdf) (academic). Supports: 64 percent prefer the Harman target, 15 percent want 4 to 6 dB more bass, 21 percent 2 dB less; ratings fall as the response deviates from the target; model correlation 0.86; over 400 headphones showed no correlation between price and frequency response; 31 models and 130 listeners in the main test.
2. [SoundStage! Solo, Where are we at with the Harman curve?](https://www.soundstagesolo.com/index.php/features/217-where-are-we-at-with-the-harman-curve) (specialist review). Supports: the 64, 15 and 21 percent groups; more-bass group wants 3 to 6 dB more below 300 Hz; less-bass group wants 2 to 3 dB less.
3. [SoundGuys, Harman Target, SoundGuys Preference Curve most preferred for in-ears](https://www.soundguys.com/harman-target-soundguys-preference-curve-validated-125420/) (specialist review). Supports: 72 percent of 32 listeners preferred the 2019 in-ear target or the SoundGuys curve.
4. [RTINGS, Our Headphone Tests: Noise Isolation](https://www.rtings.com/headphones/tests/noise-isolation-cancellation-passive-active) (specialist review). Supports: the best cancelling headphones attenuate bass-range noise by over 15 dB; passive isolation struggles with bass and handles speech-range noise; attenuation is reported by band with ANC off and on.
5. [RTINGS, Our Headphone Sound Quality Tests: Frequency Response Consistency](https://www.rtings.com/headphones/tests/sound-quality/frequency-response-consistency) (specialist review). Supports: glasses, hair and head shape change the bass you hear on over-ears; tip size and placement decide consistency on in-ears; RTINGS measures on a Brüel and Kjær 5128 rig.
6. [RTINGS, Test Bench 1.6 Latency research article](https://www.rtings.com/headphones/learn/research/latency) (specialist review). Supports: average SBC latency about 260 ms for earbuds and non-gaming headphones; gaming headsets average 170 ms on SBC; Audeze Maxwell measured 13 ms wired and 23 ms via dongle.
7. [SoundGuys, Android's Bluetooth latency needs a serious overhaul](https://www.soundguys.com/android-bluetooth-latency-22732/) (specialist review). Supports: measured averages of 308 ms SBC, 316 ms aptX, 324 ms LDAC, 369 ms AAC across four Android phones; people cannot perceive delay below 20 ms and many miss 40 to 80 ms.
8. [Feasycom, SBC vs AAC vs aptX vs LDAC vs LC3: Bluetooth audio codec comparison](https://www.feasycom.com/bluetooth-audio-codecs-comparison/) (primer). Supports: kbps: SBC max 328, AAC 256 to 320, aptX 352, aptX HD 576, Adaptive 279 to 420, LDAC max 990; codec latency ranges by spec; a link falls back to SBC when either side lacks the codec.
9. [Wikipedia, aptX](https://en.wikipedia.org/wiki/AptX) (primer). Supports: aptX 352 kbps, aptX HD 576 kbps, aptX Adaptive 279 to 420 kbps; aptX Low Latency offers 32 ms end to end.
10. [Wikipedia, LDAC (codec)](https://en.wikipedia.org/wiki/LDAC_(codec)) (primer). Supports: LDAC steps between 330, 660 and 990 kbps by connection strength; the Japan Audio Society's Hi-Res Audio Wireless certification of LDAC in 2019.
11. [SoundGuys, The ultimate guide to Bluetooth headphones: LDAC isn't Hi-res](https://www.soundguys.com/ldac-ultimate-bluetooth-guide-20026/) (specialist review). Supports: 990 and 660 kbps push Bluetooth to the limits of stability; phones rarely pick 990 kbps without developer settings; LDAC is not true hi-res and many will not notice the difference; dropped packets rise with bandwidth.
12. [Bluetooth SIG, LE Audio](https://www.bluetooth.com/learn-about-bluetooth/feature-enhancements/le-audio/) (standards body). Supports: LC3 improves on SBC even at a 50 percent lower bitrate; Auracast broadcast audio to unlimited receivers; LE Audio adds hearing aid support.
13. [SoundGuys, How to set up Bluetooth multipoint on your headphones](https://www.soundguys.com/how-to-bluetooth-multipoint-headphones-set-up-62931/) (specialist review). Supports: multipoint means more than one connected device at a time, usually two; only one stream plays at a time; the dated claim that multipoint only works with AAC.
14. [Notebookcheck, Sony WH-1000XM5 receive multipoint LDAC and head-tracking support via new update (search excerpt only)](https://www.notebookcheck.net/Sony-WH-1000XM5-receive-Multipoint-LDAC-and-head-tracking-support-via-new-update.736850.0.html) (specialist review). Supports: firmware 2.0.2 enabled LDAC during multipoint on the WH-1000XM5, which earlier forced a choice between the two.
15. [Sony, WH-1000XM5 Help Guide (PDF)](https://helpguide.sony.net/mdr/wh1000xm5/v1/en/print.pdf) (spec sheet). Supports: battery hours, cancelling on and off: AAC 30 and 40, SBC 28 and 38, LDAC 26 and 36; impedance 48 ohms on and 16 ohms off; sensitivity 102 and 100 dB/mW; plays through the cable with the power off, without cancelling; supports SBC, AAC, LDAC and two-device multipoint; mass about 250 g.
16. [Apple, AirPods Pro 3 technical specifications](https://www.apple.com/airpods-pro/specs/) (spec sheet). Supports: 8 hours per charge and 24 hours with the case with cancelling on; IP57 rating; five ear tip sizes in the box; no impedance, sensitivity or codec list published; features named Transparency mode, Adaptive Audio, Hearing Aid, Loud Sound Reduction.
17. [Apple Support, About the sweat and water resistance of AirPods](https://support.apple.com/en-us/105046) (spec sheet). Supports: IP54 on AirPods Pro 2 and AirPods 4, IP57 on AirPods Pro 3 and AirPods 5; resistant but not waterproof, not for swimming or showering; resistance diminishes over time.
18. [Standard Clarity, IEC 60529 explained: the IP code](https://standardclarity.com/standards/iec-60529/) (primer). Supports: first digit solids 0 to 6, second digit water 0 to 9; X means not specified, not zero; water level 4 splashing, 5 jets, 7 immersion to about 1 m.
19. [Sennheiser, HD 600 product page](https://us.sennheiser-hearing.com/products/hd-600) (spec sheet). Supports: 300 ohm impedance and 97 dB at 1 V sensitivity; open dynamic design, 260 g; replacement ear pads and cable sold as spares.
20. [Sennheiser, MOMENTUM 4 Wireless product page](https://us.sennheiser-hearing.com/products/momentum-4-wireless) (spec sheet). Supports: 60 hours with cancelling on; SBC, AAC, aptX and aptX Adaptive; 470 ohms active and 60 ohms passive, 106 dB SPL; 3.5 mm cable included and replacement pads sold.
21. [Amazon, Sony WH-1000XM5 listing (page head only; product table not returned)](https://www.amazon.com/dp/B0BXYCS74H) (retail listing). Supports: the listing title carries battery hours and the cancelling claim; retail product tables could not be read in this run.
22. [iFixit, AirPods 5 are the least unrepairable AirPods yet](https://www.ifixit.com/News/119587/airpods-5-are-the-least-unrepairable-airpods-yet) (specialist review). Supports: earlier AirPods scored 0 of 10; AirPods 5 score 2 of 10 with a shop-replaceable battery; Fairphone Fairbuds scored 10 of 10.
23. [iFixit, AirPods Pro 3 teardown: still a tragedy](https://www.ifixit.com/News/113507/airpods-pro-3-teardown-still-a-tragedy) (specialist review). Supports: 0 of 10 repairability; battery glued in; about 18 months before the battery stops holding a charge; disposable by design.
24. [MakeUseOf, Wireless earbuds all break over the same thing, and almost nobody sells the fix](https://www.makeuseof.com/wireless-earbuds-all-break-over-the-same-thing-almost-nobody-sells-fix/) (primer). Supports: capacity halves after a few hundred cycles; failure at the two or three year mark; glue and ultrasonic welding mean most earbuds were never designed to be opened; Fairbuds use a user-swappable coin cell.
25. [NwAvGuy, Headphone and amp impedance](http://nwavguy.blogspot.com/2011/02/headphone-amp-impedance.html) (primer). Supports: output impedance should be under one eighth of headphone impedance; higher output impedance bends frequency response, up to 12 dB on a multi-driver earphone.
26. [Wikipedia, Headphone amplifier](https://en.wikipedia.org/wiki/Headphone_amplifier) (primer). Supports: high output impedance causes response fluctuations, a 5 dB error with 120 ohm output; damping factor example and that a 120 ohm headphone from an iPod would likely not benefit from an amp.
27. [The Master Switch, Headphone impedance and sensitivity explained](https://www.themasterswitch.com/headphone-impedance-sensitivity-explained) (primer). Supports: up to 32 ohms plays from a phone, 33 to 100 is a grey area, above 100 needs an amplifier; sensitivity is loudness at a given power.
28. [headphones.com community, Impedance vs loud sound](https://forum.headphones.com/t/impedance-vs-loud-sound/25764) (enthusiast community). Supports: impedance alone does not predict loudness; look at dB/V and dB/mW; damping factor of 8 or more, ideally 10 to 20.
29. [headphones.com, How do headphone drivers work?](https://headphones.com/blogs/features/headphone-driver-types-explained) (primer). Supports: how dynamic, planar, electrostatic and balanced armature drivers work; electrostatics need 230 or 580 V bias and an energiser, weak deep bass at high level; planars are costly, seal-sensitive and move less.
30. [Campfire Audio, Balanced armature vs dynamic vs planar: IEM driver types explained](https://www.campfireaudio.com/blogs/insights/iem-driver-types-balanced-armature-dynamic-planar) (primer). Supports: dynamic drivers give physical bass and softer transients; armature bass is tighter and less visceral; planars are power-hungry; hybrids depend on the crossover.
31. [Campfire Audio, Ear tips and IEM fit guide](https://www.campfireaudio.com/blogs/insights/iem-ear-tips-fit-seal-guide) (primer). Supports: a lost seal costs 10 dB or more in the sub-bass; foam blocks 15 to 25 dB, silicone 10 to 15; foam lasts one to three months, silicone six to twelve.
32. [headphones.com, Open vs closed headphones](https://headphones.com/pages/open-vs-closed-headphones) (primer). Supports: open-backs leak and do not isolate; the openness is audible even with no music; closed-backs isolate to varying degrees.
33. [Razer, Hybrid ANC explained](https://www.razer.com/technology/hybrid-anc-active-noise-cancelling) (primer). Supports: feedforward uses an outward microphone, feedback an inward one, hybrid both.
34. [Soundcore, How does active noise cancelling work](https://www.soundcore.com/blogs/earbuds/how-does-active-noise-cancelling-work) (primer). Supports: cancelling works on long predictable low-frequency waves and fails on abrupt high-frequency sounds; microphone placement for feedforward and feedback.
35. [SoundGuys, We tested 205 ANC headphones to find out if it's worth spending more](https://www.soundguys.com/you-can-have-the-best-anc-headphones-with-one-simple-trick-139427/) (specialist review). Supports: an incredibly weak relationship between cancelling performance and price across 205 products; unsealed earbuds offer very little effective cancelling.
36. [SoundGuys, How we test headphones](https://www.soundguys.com/how-we-test/) (specialist review). Supports: measurements on a Brüel and Kjær 5128 head scored against the SoundGuys house curve; isolation measured by recording shaped noise with headphones off and on; battery tested at 75 dB SPL.
37. [Crinacle, The tonal-technical dichotomy: the IEF evaluation system](https://crinacle.com/2021/06/04/the-tonal-technical-dichotomy-the-ief-evaluation-system/) (enthusiast community). Supports: tonality is the frequency-domain part of sound; technicalities are everything non-tonal; interpreting a graph is subjective even though measuring it is not.
38. [CrinGraph README (GitHub)](https://raw.githubusercontent.com/mlochbaum/CrinGraph/master/README.md) (enthusiast community). Supports: CrinGraph compares frequency response measurements against each other and target curves on log frequency and dB axes; the code is free and underlies the squig.link sites.
39. [Shokz, Bone conduction headphones: how the technology works and its real limitations](https://shokz.com/blogs/guides/the-technical-difficulties-of-bone-conduction-headphones) (primer). Supports: bone conduction struggles with deep bass; the vibrating transducer leaks sound into the air; vibration can cause discomfort over time; the open ear keeps situational awareness.
40. [SoundGuys, Bone conduction vs open earbuds](https://www.soundguys.com/bone-conduction-vs-open-earbuds-160735/) (specialist review). Supports: bone conduction bass barely registers; it suits swimmers and some hearing loss; open earbuds leak enough for a neighbour to hear and sound better than bone conduction.
41. [Sennheiser, MOMENTUM True Wireless 4 product page](https://us.sennheiser-hearing.com/products/momentum-true-wireless-4) (spec sheet). Supports: 7.5 hours per charge and 30 hours with the case for a true wireless earbud; no codec list or IP rating on the page.