What You Can Actually Hear: Sample Rate, Bit Depth and Bitrate

Before IEMs, players and libraries, one question decides how much of the rest matters: what can a human ear actually hear? Most of the numbers printed on audio gear (44.1 kHz, 24-bit, 192 kHz, 990 kbps) answer a different question. This post sorts them out, so the rest of the series can spend money where it makes a difference.

The limits of hearing

Human hearing covers roughly 20 Hz to 20 kHz. That is the textbook range for a young person; the top end falls with age, and most adults hear little above 16 kHz. Above 20 kHz is ultrasound. Bats hear up to about 100 kHz; people do not.

Loudness spans a much wider range. The quietest sound you can hear and the loudest you can bear without pain are more than 120 dB apart, but real listening happens in a far narrower band: a quiet room has a background of 30 dB or so, and nobody listens to music near the pain threshold for long.

Keep those two facts in mind. Every digital format is a trade-off against them.

Sample rate: how often the sound is measured

A digital recording measures the sound wave many times a second. Each measurement is a sample; how many you take per second is the sample rate, written in kHz.

Sample rate	Where you find it
44.1 kHz	CDs, most music downloads and streams
48 kHz	Video, phones, many streaming services
96 kHz	Hi-res downloads
192 kHz	Hi-res downloads, studio masters

A sample rate of 44.1 kHz means 44,100 measurements of the waveform every second. That raises the obvious question: why does a CD at 44.1 kHz store sound only up to about 22 kHz?

Why the highest pitch is half the sample rate

To capture one vibration of a sound wave, you need at least two samples: one near the peak and one near the trough. With fewer, a fast wave cannot be told apart from a slower one. So the highest frequency a sample rate can store is half that rate. This is the Nyquist limit, from the sampling theorem.

Sample rate	Highest frequency stored	Audible?
44.1 kHz	About 22 kHz	Covers the whole range of human hearing
96 kHz	About 48 kHz	The extra range is ultrasound
192 kHz	About 96 kHz	Bat territory

That is why the CD’s 44.1 kHz is not an arbitrary number. It is just over twice the top of human hearing, with a little room left for the filter that removes everything above the limit.

Bit depth: how precisely each sample is stored

Each sample is a number. Bit depth is how many bits that number has, which sets how finely the loudness of each sample can be described. More bits mean a lower noise floor and a wider dynamic range: the gap between the loudest sound the format can hold and its own background noise.

Bit depth	Dynamic range	Compared with listening
16-bit (CD)	About 96 dB, more with dither	From a quiet room’s background to louder than you should ever listen
24-bit	About 144 dB	More than any playback chain or ear can use

Sixteen bits already put the format’s noise floor below the background noise of a quiet room at any sane volume. Dither, a tiny amount of added noise, pushes the useful range of 16-bit audio further still, and makes the reduction from 24-bit to 16-bit clean.

Bitrate: how much data per second

Bitrate, in kbps, is not a measure of resolution. It is how much data the file uses per second of music, after compression.

• For uncompressed audio it is simple arithmetic: 44,100 samples × 16 bits × 2 channels is 1,411 kbps, the CD’s bitrate.
• For lossless formats such as FLAC it is whatever the compression achieves. A CD-quality FLAC usually lands between about 700 and 1,100 kbps depending on the music. It tells you nothing about quality, only about how compressible the recording was.
• For lossy formats such as MP3 and AAC, bitrate is the quality knob. The encoder throws away detail it judges inaudible, and the lower the bitrate, the more it throws away.

So a FLAC file’s bitrate is trivia, while an MP3’s bitrate is the most important number about it.

Lossy versus lossless

Lossless formats (FLAC, ALAC, WAV) give back exactly the samples that went in. Lossy formats (MP3, AAC, Opus, Ogg Vorbis) discard information to save space.

Modern lossy encoders are very good. At 256 kbps AAC or 320 kbps MP3, most people cannot reliably tell the file from the CD in a blind test. At 128 kbps and below, artefacts become audible on cymbals, applause and reverb tails. The real problems with lossy files are practical: a low-bitrate rip from a video site, or a file converted from one lossy format to another, loses detail every time. That is where the “screech” in old rips comes from.

Lossless matters most as a master copy. You can make any lossy version from it later, at any bitrate, without stacking losses.

So why does hi-res exist?

If 16-bit/44.1 kHz covers human hearing, why sell 24-bit/96 kHz and 24-bit/192 kHz?

• The studio needs headroom. Recording, mixing and mastering process audio many times. Extra bit depth keeps rounding errors and noise far below the music while engineers work. Once the final master is done, that headroom has served its purpose.
• Different masters. Hi-res releases are often new remasters. People hear the better mastering and credit it to the sample rate.
• Marketing. Bigger numbers look better on a box and a store page.

The listening evidence is thin. Meyer and Moran’s 2007 study inserted a CD-quality bottleneck into high-resolution playback, and listeners could not detect it. Reiss’s 2016 meta-analysis of 18 studies did find a small effect: trained listeners told the formats apart about 60 % of the time, where guessing gives 50 %. That is a real but subtle difference under careful test conditions, nothing like the one the marketing implies. Mastering, the earphones and the fit all matter far more than the number after the slash.

What my library looks like

My own library has 1,512 FLAC files. 84 % are 16-bit/44.1 kHz, ripped or bought at CD quality. The rest is a mix of 24-bit at 44.1–96 kHz and 121 tracks at 24-bit/192 kHz, mostly recent remasters such as the 2023 The Dark Side of the Moon.

One album shows how labels mislead: a copy of Metallica’s 1991 album is 16-bit at 96 kHz. Genuine hi-res masters are 24-bit, so this is almost certainly a CD rip that was upsampled. It is bigger than the CD and contains nothing more.

The player decides what you get anyway

A file can only be as good as the device that plays it.

• No iPod has ever played above CD quality natively. The iPod Video’s Wolfson codec runs at 44.1 kHz. Rockbox plays 96 kHz and 192 kHz FLAC, but resamples them to 44.1 kHz in software, which costs battery. The iPod part of this series covers that player.
• Phone DACs are fine. A modern phone’s converter measures well. The problems are the missing headphone jack and the habit of listening over Bluetooth.
• Dedicated hi-res players do output hi-res, and differ from older players in power, output impedance and noise rather than in anything you would hear from the sample rate. The sources part of this series covers that.

Bluetooth: compressed again, whatever the file

Bluetooth earphones receive audio over a radio link with limited bandwidth, so the phone encodes the music again with a Bluetooth codec before sending it. Even a lossless file arrives at your ears as a lossy stream.

Earphones	Codec on an iPhone	Codec on Android	What reaches your ears
AirPods Pro (1st generation)	AAC, about 256 kbps	AAC	Lossy. Apple’s own lossless catalogue is converted to AAC for AirPods
Sony WH-1000XM4	AAC	LDAC, up to 990 kbps	Lossy on iPhone; close to CD quality on Android at 990 kbps, which often falls back to 660 or 330 kbps in busy radio areas

Both also apply their own processing: noise cancelling, equalisation and the tuning of the drivers. That changes the sound far more than the source file’s resolution does.

None of this makes Bluetooth bad. On a bus or a street, good AAC is hard to tell from wired. For quiet, careful listening, a wired chain removes one more variable.

What this means for the rest of the series

• 16-bit/44.1 kHz lossless is the honest ceiling for a listening library. Buy or rip at CD quality; keep hi-res only when it is a better master.
• Avoid low-bitrate lossy files and anything converted from one lossy format to another. That is where real, audible damage lives.
• Spend on the parts you can hear: the signature of your earphones, their fit, and a source that suits them. Those are the next parts of this series.

Sources

• Hearing range (Wikipedia)
• Nyquist–Shannon sampling theorem (Wikipedia)
• Audio bit depth (Wikipedia)
• Meyer and Moran, “Audibility of a CD-Standard A/D/A Loop Inserted into High-Resolution Audio Playback”, JAES 2007
• Reiss, “A Meta-Analysis of High Resolution Audio Perceptual Evaluation”, JAES 2016
• Apple: About lossless audio in Apple Music
• Sony: LDAC
• Bluetooth codecs, Audioengine