Hi-res audio explained: what the numbers mean and when they're worth paying for

Every spec sheet in the catalogue lists the sample rates and bit depths a device accepts, and the marketing around those numbers has built an industry tier called "hi-res" - badges, premium subscription levels, premium file prices. To evaluate any of it, you need just two pieces of math, both mercifully short.

What sample rate actually does#

Sampling measures the audio waveform thousands of times per second, and the Nyquist theorem sets the only rule that matters: a sample rateHow many audio samples are captured per second; 44.1kHz captures frequencies up to 22.05kHz (Nyquist limit), well above the 20kHz hearing limit. captures all frequencies up to half of itself, completely. Not approximately - completely, for everything below that limit. CD's 44.1 kHz therefore carries everything up to 22.05 kHz, which already extends past the ceiling of young, undamaged human hearing and far past the ceiling of most adults'.

What higher rates like 96 or 192 kHz add, then, is bandwidth above the audible band - content for ears nobody has. The honest engineering argument for them lives elsewhere: gentler anti-aliasing filters (the brick-wall filtering at 44.1 kHz is harder to do invisibly, a genuine if small consideration), and real utility in production, where processing benefits from headroomThe decibel margin between the loudest expected signal and an amplifier's clipping point. 10-20dB of headroom is generally needed for unclipped reproduction of dynamic recordings at realistic listening levels.. As delivery formats for listening, the audible case for high rates has never survived level-matched blind testing - which is why this catalogue's methodology leans on tools like the ABX test rather than format faith.

What bit depth actually does#

Bit depthThe number of bits per sample, which determines dynamic range (approximately 6dB per bit, so 16-bit gives ~96dB, 24-bit gives ~144dB). sets the distance between the loudest undistorted signal and the quantization noise floor - roughly 6 dB per bit. Sixteen bits give about 96 dB of dynamic rangeThe decibel span between a system's maximum undistorted output and its noise floor; 16-bit audio has ~96dB, 24-bit has ~144dB of theoretical range.; with the ditheringAdding controlled low-level noise to audio data when reducing bit depth, which randomizes quantization error and prevents the audible distortion of truncation. that's been standard practice for decades, the effective floor sits below audibility in any real room at any survivable volume. Twenty-four bits stretch the theoretical range to ~144 dB, beyond what any DACDigital-to-Analog Converter, a device that translates binary audio data into an analog electrical signal that can be amplified and heard.'s analog stageShort for soundstage; the perceived three-dimensional acoustic space of a stereo recording. Often used to describe headphone presentation specifically ("the Arya has a deep stage"). - let alone any listening room - can reproduce.

As with sample rate, the genuine 24-bit case is production headroom: engineers tracking live performances need a noise floor far below the action. There's also one playback scenario where bits become your problem: heavy digital volume attenuation discards resolutionA system's ability to retrieve and reproduce fine detail in the recording; high resolution reveals micro-dynamics, spatial cues, and timbral nuance. one effective bit per 6 dB - the bit depth calculator shows exactly what your system volume costs. The fix is gainThe multiplication factor applied to a signal by an amplifier, expressed in dB; proper gain staging is critical for minimizing noise. structure, not bigger files.

Where hi-res honestly earns money#

So why does the hi-res tier persist - and why do hi-res versions often sound better in casual comparison? Three honest reasons. The big one: mastering. Hi-res releases frequently come from different, more careful masters than the loudness-war CD pressing - you're hearing better mastering, sold under the format's badge. It's a real improvement with a misattributed cause. Second: lossless delivery, which historically rode in with the hi-res tier; the audible upgrade from lossy streaming to FLACFree Lossless Audio Codec, an open-source lossless compression format that reduces file size by 40-60% with zero degradation to the audio data. is modest but real on revealing chains. Third: it's cheap insurance - storage is free, every modern device decodes it (a $179 box like the SMSL DL100 handles DSD256 without blinking), so there's no penalty to keeping the bigger numbers.

What that adds up to: pay for masters, not for math. When a hi-res version sounds better, it usually is better - just not for the reason on the badge.

When the numbers lie#

The hi-res market has a fraud problem the badges don't mention: nothing prevents an upsampled CD master - or worse, a decoded MP3 - from being packaged in a 24/96 FLAC container and sold at the premium price. The container's numbers describe the box, not the contents, and a surprising fraction of "hi-res" catalogue fails the inspection.

The inspection is easy, because real high-rate recordings carry energy above 22 kHz and upsampled fakes show an empty shelf where that content should be - a hard cutoff at the original rate's Nyquist limit. The hi-res audioA marketing certification (JAS / Hi-Res Music) for audio with sample rates ≥ 96kHz and bit depths ≥ 24-bit. The audible benefit over 16-bit/44.1kHz is debated; the legitimate benefit is reduced quantization error in the mixing chain. checker runs that FFT analysis in your browser, locally, on any file you feed it. Habitually checking premium purchases is the single most practical use of everything on this page. (Format trivia like MQAMaster Quality Authenticated, a proprietary lossy compression format claiming to preserve master-quality information in a smaller file. Critics argue it adds distortion and serves rights-holder DRM more than fidelity; Tidal removed MQA from its catalogue in 2024.'s compromises and the streaming tiers' history is covered in the glossary - the short version is that lossless availability, not resolution, was always the part worth wanting.)

The honest summary for buying decisions: 16/44.1, done well, was already a complete delivery format for human hearing - that's the measured reality. Hi-res costs nothing to accept, sometimes carries better masters, and occasionally carries fraud; transducers, rooms, and recordings still own the differences you can actually hear. Spend accordingly, verify what you pay premiums for, and let the chain's weakest link - never the file format - direct the next upgrade.