
Audio specs and measurements explained
Frequency response is the audio spec that best predicts how a headphone or speaker will sound. Impedance, sensitivity and power output say how hard it is to drive, while SINAD, THD and SNR say how clean an amplifier or DAC is. Those cleanliness figures stop mattering once they pass audibility, which most modern DACs do.
24 terms Updated
Decibels and levels
Almost every number on a spec sheet is in decibels, and many of the rest describe a voltage. Start here and the other figures read more easily.
- Decibel (dB)
- The decibel is the logarithmic unit audio uses for level and ratio. Each 10 dB is a tenfold change in power and sounds roughly twice as loud; about 1 dB is the smallest change most people notice. Almost every audio specification is written in it.
- Spec sheets use it two ways: as a level against a stated reference, such as dB SPL or dBV, and as a ratio between signals, as in SNR; a level without its reference means nothing.
- Line level
- Line level is the standard signal voltage passed between audio components, stronger than a microphone or turntable signal and weaker than a speaker signal. Home equipment works at a nominal −10 dBV, about 0.316 V; professional equipment at +4 dBu, about 1.23 V.
- Levels must match at each connection: a line-level source overloads a phono or microphone input, and a turntable plugged straight into a line input plays faint and thin.
- Gain
- Gain is how much an amplifier or preamp multiplies the signal it is given, usually stated in decibels of voltage. It sets how far the volume knob turns to reach a given loudness, not how much power the amplifier can deliver.
- Gain and power output are often confused: gain sets how loud a given knob position is, power sets the ceiling, and a gain switch changes only the first.
Impedance and power
These figures decide whether an amplifier suits a headphone or speaker. Read them in pairs: impedance with sensitivity, power with the load it was measured into.
- Impedance
- Impedance, given in ohms (Ω), is how much a headphone or speaker resists the current an amplifier sends it. High-impedance headphones need more voltage to play loud; low-impedance ones draw more current and are easier for phones to drive.
- On its own it predicts little about loudness; pair it with sensitivity to estimate the voltage and power needed, and with output impedance to check the tonal balance survives.
- Sensitivity
- Sensitivity says how loud a headphone or speaker plays for a given input: so many decibels per milliwatt or per volt for headphones, per watt at one meter for speakers. Higher numbers play louder from less power.
- For speakers it is one of the most useful pairing figures on the sheet: every 3 dB of extra sensitivity halves the amplifier power needed for the same loudness.
- Efficiency
- Efficiency is how much sound a transducer produces from the electrical power it is given. In everyday use it means the same as sensitivity: an efficient headphone or speaker plays loud from a small amplifier.
- Spec sheets print sensitivity rather than efficiency, so when a forum post calls a speaker efficient, the sensitivity figure is the number behind the word.
- Power output
- Power output is how much an amplifier can deliver into a given load, in watts for speakers or milliwatts for headphones. The figure only means something with the load stated, since the same amplifier delivers different power into 32 Ω and 300 Ω.
- The rating is a ceiling reached only on peaks; for headphones it means little until you read it at your own pair’s impedance and convert it with sensitivity into decibels.
- Headroom
- Headroom is the margin between the level a system normally works at and the most it can deliver before it clips. On audiophila. it means how many decibels above 100 dB SPL an amplifier can drive a given headphone; 10 dB or more is enough.
- This is where the figures in this group meet: power and sensitivity give the maximum level, and the gap between that and your loudest listening is the headroom.
- Output impedance
- Output impedance is the resistance an amplifier presents to the headphone or speaker it drives, in ohms. A low value, ideally under an eighth of the load’s impedance, keeps the tonal balance and bass control the designer intended.
- The amplifier’s side of the impedance pairing, and not always printed; when it is missing, low-impedance headphones and multi-driver IEMs are where the unknown matters most.
- Damping factor
- Damping factor is a speaker’s impedance divided by the amplifier’s output impedance, a measure of how firmly the amplifier controls the speaker cone’s movement. Values above about 20 are considered adequate; the very high figures on some spec sheets add little.
- The speaker-world form of output impedance, turned upside down so that higher is better; figures in the hundreds or thousands look impressive and change little you can hear.
Frequency response and targets
The measurement that best predicts how a headphone or speaker sounds, and the reference curves headphone graphs are read against. A headphone graph means little until you know which target it is compared with.
- Frequency response
- Frequency response is how loudly a product reproduces each frequency from bass to treble, shown as a line on a graph. It is the measurement that best predicts how a headphone or speaker will sound, though it does not capture everything.
- It is the only spec on this page that describes how a headphone or speaker sounds rather than how clean or how loud it is; most of the others are about electronics.
- Diffuse-field
- Diffuse-field is a headphone measurement target that describes how a flat sound source would measure at the eardrum if it arrived evenly from every direction. It is a reference for what "neutral" should look like on a headphone measurement rig.
- Plotted against it, a headphone most people hear as balanced shows a bass rise of several decibels, because the curve on its own sounds lean and bright.
- Harman target
- The Harman target is a headphone frequency response curve derived from listener preference research at Harman International, with a moderate bass shelf and slightly less treble than the diffuse-field curve. Many headphones and IEMs are tuned toward it, and many measurement sites compare against it.
- When a graph is labeled against Harman, a flat line means a match; check which version was used, since the over-ear and in-ear curves differ and should not be mixed.
Noise and distortion
How much an amplifier or DAC adds to the signal. These are the figures where modern electronics most often pass what anyone can hear, and where spec sheets most often leave out the test conditions.
- THD
- THD, total harmonic distortion, is how much a device adds overtones that were not in the signal, given as a percentage or in decibels. Lower is cleaner; below about 0.1 percent it is generally inaudible on most music, though listeners argue about the exact threshold.
- Percent and decibel forms are interchangeable: 1 percent is −40 dB, 0.1 percent is −60 dB and 0.01 percent is −80 dB, each step ten times cleaner.
- THD+N
- THD+N is total harmonic distortion plus noise: everything a device adds to a test tone, distortion and background noise together, as a percentage of the signal. It is the figure most spec sheets quote and the inverse of SINAD.
- Of the distortion figures here, this is the one most spec sheets print, while measurement charts more often quote the same result as SINAD.
- SINAD
- SINAD is signal-to-noise-and-distortion ratio: how much louder a test tone is than everything the device adds to it, noise and distortion combined, in decibels. Higher is cleaner. It is the headline number on many DAC and amplifier measurements.
- Above about 90 dB, the level generally regarded as transparent, a higher SINAD shows better engineering more than an audible difference, though where audibility ends is still argued.
- Intermodulation distortion
- Intermodulation distortion (IMD) is the extra tones a device adds when two or more frequencies pass through it together, at their sums and differences. These tones bear no musical relation to the notes, so IMD can sound worse than harmonic distortion of the same size.
- Measurement reports usually show it as a twin-tone test, two high tones close together, or the SMPTE test, one low tone and one high; consumer spec sheets seldom list it.
- SNR
- SNR, signal-to-noise ratio, is how much louder the signal is than the background noise a device produces on its own, in decibels. Higher is better; above about 100 dB, noise is usually inaudible at normal listening levels.
- Check whether the figure is A-weighted, which discounts noise the ear hears poorly and usually reads a few decibels better than an unweighted result of the same device.
- Noise floor
- The noise floor is the faint hiss or hum a system produces with nothing playing. A low noise floor, sometimes praised as a "black background", lets quiet passages and reverb tails emerge from silence instead of from hiss.
- The absolute counterpart of SNR: a microvolt figure for the hiss with nothing playing, and the one to check before plugging sensitive IEMs into a desktop amplifier.
- Dynamic range
- Dynamic range is the span between the quietest and loudest sounds a system or recording can hold, in decibels. For equipment it is set by the noise floor; for a recording it depends on how much the mastering compressed it.
- Check which kind you are reading: a device figure, often above 110 dB for DACs, or an album’s DR score, which reflects its mastering and nothing in your system.
- Crosstalk
- Crosstalk is how much of one stereo channel leaks into the other inside a device, in decibels below the signal. Lower leakage keeps the stereo image stable; it is one of the measurable things a balanced connection improves.
- It is the measured form of channel separation, not of the instrument separation reviewers praise; electronics keep it far below audibility, so it seldom explains a difference in stereo image.
| Spec | What it measures | What good looks like | When it matters |
|---|---|---|---|
| Frequency response | Level at each frequency, shown as a graph | Close to a target you like; a bare “20 Hz to 20 kHz” says nothing | Always, for headphones and speakers |
| Impedance | The electrical load a headphone or speaker presents, in ohms | No good or bad value; read it with sensitivity | Choosing an amplifier |
| Sensitivity | Loudness for a given input, per milliwatt, volt or watt | Higher plays louder from less power; check whether dB/mW or dB/V | Phones, portable players, low-powered amplifiers |
| Power output | Watts or milliwatts the amplifier delivers into a stated load | Enough for 10 dB or more of headroom above a 100 dB SPL peak on your headphone | Insensitive headphones and speakers |
| Output impedance | The amplifier’s own resistance at its output, in ohms | Under an eighth of the headphone’s impedance; most solid-state amps are under 1 Ω | Low-impedance headphones, multi-driver IEMs |
| SINAD | A test tone against all added noise and distortion, in dB | Above about 90 dB is generally regarded as transparent; many modern DACs reach 110 to 120 dB | Screening DACs and amplifiers |
| THD | Added harmonics, in percent or dB | Electronics well under 0.01%; below about 0.1% is generally inaudible on music | Headphones and speakers at loud bass levels, where it runs far higher |
| SNR | Signal against background noise, at a reference level | Above about 100 dB | Sensitive IEMs, where the absolute noise floor matters more |
| Crosstalk | Leakage from one channel into the other, in dB | −80 dB or better, which most modern electronics manage | Rarely; cheap 3.5mm jacks with a shared ground |
| Jitter | Timing error in a digital clock | Sidebands more than about 110 dB below the test tone | Hardly ever with modern DACs, which reclock incoming data |
Digital audio
Numbers that describe a digital file or stream and its clock rather than an analog circuit. They set the format’s limits, and CD quality already covers what home listening needs on both bit depth and sample rate.
- Bit depth
- Bit depth is how many bits describe each audio sample: 16 bits for a CD, 24 bits for most hi-res files. More bits lower the noise floor of the digital format, giving a wider dynamic range, rather than adding detail to the music.
- Compare it with the device figures above: 16 bits allow about 96 dB, a good DAC manages 120 dB or more of dynamic range, and 24 bits allow 144 dB that no DAC reaches.
- Sample rate
- Sample rate is how many times per second a digital system measures the audio waveform: 44,100 for a CD, 48,000 for video, 96,000 or 192,000 for hi-res files. It sets the highest frequency the format can carry, at just under half the rate.
- Set it beside frequency response: a 44.1 kHz file reaches 22.05 kHz, past the right-hand edge of a typical headphone graph at 20 kHz.
- Jitter
- Jitter is timing error in a digital audio signal: the samples arrive slightly early or late relative to a perfect clock. In a DAC it adds noise and distortion sidebands, though in modern designs it usually measures far below published audibility thresholds.
- The one spec on this page about timing rather than level; in modern DACs it shows as faint sidebands around a test tone and is rarely the weak link in a system.