Digital Music Quality Revisited

Since I love music, I was looking for a way to test my hypothesis about a better music experience. There are two major parameters, the bit depth per sample, and the sampling rate. CD quality music is encoded with 16 bits, and played at 44.1 kHz. Again, I don’t even consider MP3 or AAC encoded music here, with their lossy compression, as well as any other compression degradations and artefacts down the playback chain, such as wireless Bluetooth headphones.

Music Sources and Player

I needed to find music that is not compressed (or only losslessly compressed), and mastered at higher sampling rates, such as 96 kHz or more. I’d consider 96 kHz the lower limit to minimise the detrimental impact of the required low pass filter before the sampling stage, as well as for the reconstruction upon playback. There are several on-line sources to purchase these music files, such as https://www.hdtracks.com. A 24 bit 96 kHz album is about one GB of data.

I also needed a player device. As with everything in the realm of “audiophile” music playback, you find devices in price ranges that make you wanna cry. You know, like loudspeaker cables for 100 Euros per metre. Or more. After reading reviews, I settled on a simple player for about 120 Euros, the HIFI WALKER H2 to get started. I may want to get a better player going forward. A 120 Euros player simply cannot have highest quality components.

Of course I use my old-fashioned wired in-ear monitors.

Experience

Well, I have never experienced Peter Gabriel’s “Intruder” with more punch, apart from hearing it at live concerts, of course¹. The same holds true for other music I have since purchased. As predicted by the Laplace theorem, the biggest difference is when you have sudden changes in amplitude, leading to high slew rates of the signal, which require more frequency bandwidth to be present in the music signal. That is, the higher sampling rate actually does result in better timing precision. I guess I am Laplace compatible.

There are differences of quality improvement between different albums, maybe due to the source master material that was available to draw the higher sampling rate files from. Unfortunately, this information is not disclosed by the sellers of the music files.² Also, let’s not forget that just using a different player can change the listening experience due many factors in the playback chain, from digital pre-DAC³ processing (eg. upsampling) to DAC type to anti-aliasing (anti-imaging) low pass filtering in the analogue domain and amplifiers. And the remastering process was most likely different.

Conclusion

I am aware of confirmation bias, or more technically, the predictive processing of our brain with its (proximate) Bayesian inference through generative models, which of course would “turn up” my corresponding priors in favour of these findings. :)

Be it as it may, my experience has been to be blown away by the punch and timing precision when I listened to the music with the new rig and music files. Whatever brings about the experience. Now, of course, my brain’s priors have been updated, and my listening expectations have been spoiled forever. Maybe it’s time to buy one those 1,500+ Euro portable music players to cope with my new addiction. Just kidding.