Hi-Res Audio Explained

“Hi-res” describes two numbers: how many bits describe each sample, and how many samples per second there are. Both have a precise technical meaning. Neither is owned by a standards body that manufacturers must answer to, which is why what actually gets published is a maximum supported rate and nothing more.

So rather than argue about definitions, this page does what the rest of the site does: takes the numbers the manufacturers publish and works out what they imply.

Bit depth is dynamic range, and it is calculable

Each bit of depth adds about 6 dB of range between the loudest signal a system can encode and the noise floor beneath it. The relationship is exact for an ideal converter, and the arithmetic below runs it against the actual published signal-to-noise figures of every DAC we cover.

This is not a criticism of any of these DACs. It is a statement about where the limit lives. A 16-bit file already provides 98 dB of dynamic range; the converters that play it resolve around 118 to 126 dB. The format is not the bottleneck in either direction, and the extra eight bits of a 24-bit file describe a noise floor no converter reaches.

Sample rate is bandwidth, and it is also calculable

A sampled system reproduces frequencies up to half its sample rate. So:

Now put that beside the rest of the chain. Ortofon publishes 20 Hz – 20 kHz (+3/−1 dB) for the 2M Red. Audio-Technica publishes 20–22,000 Hz for the AT-VM95E. Klipsch publishes 62 Hz – 21 kHz ± 3 dB for the R-51M; ELAC publishes 44 Hz – 35 kHz for the Debut B6.2 with no tolerance at all.

The rest of the equipment is designed and specified to roughly 20 kHz. A 768 kHz-capable DAC is carrying a band 384 kHz wide into speakers whose makers publish a ± 3 dB figure that stops at 21 kHz. The specification is real; the part of it above about 20 kHz has nothing downstream that claims to reproduce it.

So does hi-res do anything?

Three honest answers, in descending order of confidence.

1. The master is usually the real variable, and nobody publishes it

A hi-res release and a CD release of the same album are frequently different masters — different equalization, different dynamic range compression, sometimes a different mix. If the hi-res version sounds better, that can be why, and the format is along for the ride.

We cannot tell you which release you are getting, because no streaming service or label we are aware of publishes mastering provenance in a form a buyer can check before pressing play. That is a genuine information gap and we are not going to fill it with a guess.

2. Higher sample rates give the converter’s filter more room

At 44.1 kHz a converter must remove everything above 22.05 kHz while leaving everything below 20 kHz alone — a steep filter in a narrow gap. At 96 kHz that same filter has from 20 kHz to 48 kHz to do its work, so it can be gentler. This is a real engineering argument and it is about the converter, not about the extra content.

3. Extra bit depth matters when you are producing, not when you are listening

Recording at 24-bit means you can set levels conservatively and still have far more range than you need — which is exactly the argument we make on digitizing records. Once the file is mixed and mastered, that headroom has already done its job.

DSD, briefly

A different encoding entirely — one bit at a very high rate rather than many bits at a low one. Three DACs we cover support it: Topping publishes DSD64–DSD512 native on the E30 II, iFi publishes DSD512 on the ZEN DAC 3, and SMSL publishes DSD up to 22.5792 MHz on the SU-1. Schiit publishes nothing about DSD for the Modi 5 at all, which our spec table records as a gap rather than as a “no”.

If you have a DSD library, check the box supports it. If you do not, this specification costs you nothing and changes nothing.

Oversampling and upsampling are not hi-res

Almost every DAC sold today oversamples internally, and it is unrelated to the resolution of your file. The converter runs its modulator at a far higher rate than the incoming data so that its reconstruction filter has an easier job — the same argument made above for higher sample rates, applied inside the chip rather than to the recording.

Manufacturers describe this in different amounts of detail. Schiit publishes “Schiit Mesh — their digital filter plus a delta-sigma modulator, in this case an ES9018” for the Modi 5, which is unusually specific. Topping publishes the converter parts directly: two AK4493S in the E30 II, an ES9038Q2M in the D10s. Bluesound publishes an ES9039Q2M with Hyperstream IV. iFi publishes only “Burr Brown” for the ZEN DAC 3 and does not give the part number at all.

None of that turns a 16-bit file into a 24-bit one. Software upsampling does not either — it interpolates between the samples you already have, which cannot add information that was not captured. Both are legitimate engineering; neither is a resolution increase, and a product page that implies otherwise is describing a filter as though it were a source.

MQA, and the one device here that plays it

MQA is a container format that packs a higher-rate recording into a stream a standard player can decode partially, with full decoding requiring hardware or software that knows about it.

Of every device we cover, exactly one publishes support: Bluesound lists MQA playback for the Node Nano, alongside AirPlay 2, Spotify Connect, Tidal Connect, Qobuz Connect and Roon Ready. WiiM publishes no MQA support for the Pro Plus, Ultra or Amp. None of the standalone DACs — the Topping, Schiit, SMSL, FiiO or iFi units — publish it either.

So this is a compatibility question with a one-line answer: if your library or your streaming tier is MQA and full decoding matters to you, the Node Nano is the only box on this site that says it can. If it is not, this specification costs you nothing. We are not going to arbitrate the long-running argument about whether MQA is worth having, because that argument turns on listening comparisons we have not made.

The thing that actually limits your system

Look back at the numbers on this page. Every DAC we cover publishes distortion at or below 0.0007% and signal-to-noise at or above 113 dB. Those figures are close enough to each other that the choice between them is not a fidelity choice.

Meanwhile: the speakers we cover publish frequency responses spanning 44 Hz to 68 Hz at the bottom end — a 24 Hz spread in where the bass stops. Sensitivity spans 83 dB to 95 dB. Room treatment changes what arrives at your ear by more than any of it.

Buy the DAC for its inputs, its outputs and whether it will fit your system. That is the argument our DAC roundup makes at length, and the hi-res specification is not the reason to pick one over another.

One practical trap worth knowing

Your input choice can cap your sample rate before the DAC does. FiiO publishes 384 kHz over USB and only 96 kHz over optical for both the K7 and the K11 — a fourfold difference decided by which cable you plug in.

If a hi-res file is playing at a lower rate than you expect, the connection is the first thing to check. That is the subject of optical vs coaxial vs USB.