Playing with high-resolution audio

Some time ago, I’ve found out that under special circumstances I can hear differences between 320 kbps mp3 and CD audio. Now, I’m curious whether I can hear any difference between CD and high-resolution audio.

There is a lot of discussions whether high-resolution audio is a gimmick or a useful thing. Maybe the answer is simple: some people can hear the difference against CD audio, some can’t and some think they can while they can’t. Either of that is fine, one enjoys the music the way he likes it.

However it’s not that easy to find relevant arguments about high-resolution audio. One of the most pro-arguments is that it’s better because it has a higher sampling frequency and a higher bit depth. Yes, it has both, but an explanation whether it actually matters and why is usually missing. On the other hand, there are cons-arguments about „Nyquist theorem“ and „dynamic range“. But they also usually miss meaningful explanations. Yes, according to the Nyquist theorem, CD audio is good enough to capture frequencies up to 22 kHz, i.e. covering the frequency limits of human hearing. But music is not only about frequency but also about volume and tone curve shapes. Imagine what happens e.g. to a sine or another 10 kHz or 11.025 kHz sound curve when it’s sampled on 44.1 kHz. Yes, you get exactly the same tone frequency, but possibly with a lower volume and sounding like a different instrument. And regarding dynamic range, it’s about resolution in the quiet parts. While 16-bit whole-range resolution covers the whole audible specter, less than 10-value range remains to represent quiet but still audible sounds, which looks a bit coarse.

So purely numerically, high resolution audio adds information within nominal human listening range. Whether we can recognize the added information is a different question as well as whether we can percieve inaudible frequencies some way. I guess my tentative answer above holds, but let’s return to the original problem, whether I can hear the difference. The best way to find out is to try it, which is not completely trivial.

The first step is finding true high-resolution audio. Some popular hi-res sites sell overpriced recordings, which I’m not going to buy. Another option is utilizing streaming services offering hi-res audio but I’m not their subscriber. Fortunately, there is a selection of Czech music of various genres in hi-res, not much more expensive than their CD quality versions. I bought five albums of different kinds there that I’d probably buy anyway sooner or later (three contemporary recordings, two older recordings from analog tapes; one classical orchestral work, two single-instrument classical works and two popular music albums with a lot of vocals; one 192 kHz, two 96 kHz, one 88.2 kHz, one 48 kHz and all 24-bit). For a total additional price of about one pizza I could obtain their hi-res versions. This is worth the fun, isn’t it?

The second step is how to play hi-res audio. Besides the obvious need of having an audio system capable of playing hi-res audio, there are also software issues. The PulseAudio thing clearly tries to solve different problems than playing hi-res audio (well, I wonder again which ones — my recent experiences include heavily distorted sound on normal play and inability to switch to a different sound input for a running application) and it’s best to avoid it. Fortunately, there is a way to prevent PulseAudio from touching a USB DAC. Then it’s possible to use the DAC safely with a player that can send sound data to the DAC directly, without any sound processing and conversions. I used VLC with ALSA output. Another option is to use Volumio on a dedicated device (such as Raspberry Pi), which does the right thing out of the box.

The third step is to get something to compare the hi-res version to. The mp3’s bundled with the hi-res recordings are lossy and I can’t be sure they weren’t processed differently. I used sox utility to convert the audio to a CD like quality, i.e. 16 bit depth and 48 or 44.1 kHz frequency (depending on the original hi-res frequency):

sox FILE.flac -r 48000 -b 16 FILE-cd.flac

The fourth step is finding a calm moment when it’s possible to listen to the music carefully, without disruptions and background noise. This is the most tricky thing!

Finally, the fifth step is actually listening to the music.

And the result? I couldn’t hear any difference.