Audio Experience

This is – finalman¹ – the third post in a three part series, after Rabbit Holes and Audio Reconstruction.

With good playback equipment available, I was ready to do some more in-depth assessments. As we have established, transients in music can cause frequency components that are well above the 20 kHz limit ascribed to healthy, young humans, and transients are crucial for the listening experience. Hence above-CD quality music should sound better, right?

Alas, it’s not that simple.

Variations

I was really surprised how different the experience can be from day to day, from situation to situation, even playing exactly the same music. And I mean hearing the music differently, different details, different timbres, and so on. I guess it’s the power of predictive processing, and changing precision weighting of priors-based predictions and actual audio input, ie. the corrective signals. I am tricked by my mind.² It’s the nature of our perception. The invisible gorilla again and again.

If there are this kind of differences in my own experience, by extension, there must be similar or even greater variances compared to what you hear. So: YMMV.

Interestingly, if I very consciously apply a focused meditation-type mind-set, I can pierce through some of the variations. As Andy Clark describes, it’s possible to hack your predictions. Which I have tried, also in real life outside of music listening, with more or less success. When I find I don’t hear the music the same as, say, yesterday, I can focus, try to remember, and the previous experience comes back. It’s quite astonishing. And eye – err, I mean, ear – opening.

The Files

In Digital Music Quality Revisited I had complained that the sellers of hi-res music don’t reveal the sources from which their files are drawn. I was wrong.

www.prostudiomasters.com actually provides some of this information. You know if the original recording was analogue, or 44.1 kHz digital PCM, or whatnot. As it turned out – read on –, this information is relevant and useful.

Here’s another site: www.highresaudio.com. Alas, not as much technical information, but some good titles, and detailed search functionality.³

The Chain

What we experience is the end product of a chain of technical and creative activities and decisions, apart from the actual performance by the musicians:

1. multi-track recording
2. mixing the tracks
3. mastering
4. media production

In general, the most important factor for our listening experience is the quality of this process as a whole, independent of the technologies used. No fancy high sampling rate can rescue bad positioning of microphones when recording a classical concert, for example. All crimes committed during recording are carried over. As I like lots of old stuff, the corresponding available sources are analogue recordings. Digital recordings started towards the 1980s. Due to technical limitations of the time, early digital recordings were made using 44.1 kHz sampling rate, or 48 kHz. All bandwidth lost there is gone forever.

All decisions during mixing and mastering⁴ determine the sound quality of, and experience via, the end product that we consumers can get our hands on. Before the digital era these were vinyl LPs, or tape cassettes, then CDs.⁵ Today we also have downloadable files in different formats, among which FLAC and ALAC are two prominent examples. Both provide losslessly compressed music. I talk about what music enthusiasts use who value the quality of their experience, and also have the equipment to play it back. The big majority of music consumers downloads – or streams – lossily compressed music files. Of course the same person may use a mix of both, eg. one set to use on the go on their storage-limited phone and Bluetooth earphones, and the lossless set when using a higher quality playback system. The file size difference is about 1:10, give or take.

The bottom line here that at the point where we get our hands on the final medium we listen to, lots and lots of decisions have been made, both artistically and technically, both good and bad, depending on our personal tastes and preferences. For my listening comparisons I don’t usually have different recordings, just one final product.⁶ Hence no direct A/B comparison of the same musical base material is possible, eg. at different sampling rates. Which makes the assessment of the different music and file formats difficult. It’s for sure not objective. Maybe if the mixing engineer had done things differently, the sound would be better for my ears, even from the same same base recordings. And even if I know that the recording was done analogously, I have no idea what quality the tape recorder and the ensuing results were, which form the basis for what I hear today. Analogue or digital recording, a well made 44.1 kHz 16 bit music file can sound much better than a sloppy 92 kHz 24 bit one. And your experience might be different – “well” and “sloppy” are very much in the ear of the listener.

As an aside, one has to admire the skills that went into producing high quality analogue vinyl records, through all the steps. Which were the highest quality music sources available for us mere mortals in the fully analogue world of yesteryear. Cassette tapes are not even a contest, and who had one of these huge reel-to-reel tape machines at home?! Preparing a master tape for the production of a good vinyl record was an art in itself, since the record players are such a collection of compromises.⁷

Analogue Sources

The above sites offer lots of older music, eg. from the seventies or even sixties of the last century, often in “hi-res” 92 kHz 24 bit or better formats. Prostudiomasters will point out that these files were transferred from the original analogue sources. What we don’t know, though, is if the original analogue track recordings were used, or just the analogue masters. The former would require a digital remix. We also lack the detailed info about these “transfers”, in particular about the specific sampling rate used – also for the remix, but of course we can reasonably assume the sampling rate to be equal or higher than on the final product we can purchase. But we don’t know, maybe lower rate tracks were “upsampled”. ⁸ Which would be cheating, but hey, “hi-res” sells.⁹ Some albums are labelled as “remastered”, but I don’t know if that includes a remix of the original tracks, or if the technology magic was just applied to the stereo master tracks.

But digital recordings, even at CD sampling frequency, are for sure better than the ancient analogue recordings, right? No, not necessarily. The Ampex ATR-100 professional audio tape recorder had an upper bandwidth limit of 28 kHz with only -2 dB level loss, at a tape speed of 30 inches per second. It’s unlikely that many recordings were made at that costly breakneck tape speed, but it shows the potential.

Importantly, though, analogue recorders don’t require any artificial up-front bandwidth limitation to work properly, unlike digital audio recorders with their mandatory anti-aliasing filters. The analogue bandwidth is limited by the components and the construction of the devices themself. But here’s a little “secret” that often gets forgotten: analogue equipment has useful “data” well above what is specified as upper frequency limit, if at an attenuated level. But the high frequency information is there, provided it’s above the noise floor. It’s how analogue devices and filters work. The audio level falls off gradually, and is not cut off by a steep anti-alias filter just above 20 kHz. Hence, a good analogue signal can be digitised using a sufficiently high sampling rate, and the resulting digital data will contain all that “analogue” high frequency content well above 20 kHz.

Consequently, high resolution music files created from an ancient analogue source can sound better than a “modern” digital recording using 44.1 or 48 kHz sampling frequency, possibly even 92 kHz, depending on the analogue tape equipment used back in the day, and the conservation status of these tapes today.

The Experience

I know, lots of ifs, uncertainties and caveats, many variables. But that has been (and still is) exactly my experience. Let me try to express my essential findings anyway.

Timbre

High resolution music files, played on capable hardware, can reproduce the timbre of each single instrument in a delightful way, much more nuanced than CD quality. Which should not be surprising, since the transients at the beginning and the end of a musical note are decisive, as explained for the sound of an instrument, much more than the spectral composition.

Precision

With the enhanced transient playback reproduction comes a precision in the overall experience. Again, not surprising. It’s a precision of timing of a whole track, as well as single notes.

“Openness”

I might be veering into dangerous “audiophile” argumentation territory here. There’s an “openness” to the experience with well-produced high resolution (92 kHz or more) music. I was listening to a 92 kHz album, but all of a sudden, a new song appeared constrained to my ears (and brain). And sure enough, going to the technical information on Prostudiomasters I found out that this single song was only mastered at 44.1 kHz. This “openness” (“airiness”, “clarity”) is hard to describe, but I also experience it – or the contrary, actually – when listening to 44.1 kHz music files, which often sound constrained. I cannot really explain this in technical terms of transients and whatnot. It’s like somehow being aware of the higher bandwidth, which only manifests itself through the transients, not continuous sine signals, which I am sure are capped at 15 kHz or even lower, considering my age.

So…

Thinking about it now, my findings appear to be fully in line of what you can even expect from a high resolution (above CD) quality music recording, based on the theoretical foundation. The above are the traits that can make high resolution recordings sound better than CDs, all other factors being equal. There cannot be more to it. Unless I have missed some fundamental theory or technology.

Is It Worth the Efforts and Cost?

As usual, it depends. If you listen to music, say, while working, that is, being focused on other stuff, you probably will not experience a big difference. Our mind has only so much attention it can pay to details, as the precision weighting of priors and corrective signals is a zero-sum affair, according to the predictive processing theory.

However, when I sit down and enjoy to actually listen to the music, ie. pay attention, it’s well worth it for me. Again, the overall experience is strongly dependent on the recording and processing capabilities, decisions, and preferences of the people involved. Some albums sound airy and light, with well defined instruments on the soundstage, and nuances in the dynamics. Others sound mushed together, both regarding instrument separation and loudness, using the same playback hardware, even if the corresponding music recordings are 92 kHz 24 bit.

I really enjoy to listen to single notes and instruments, their timbre, and their well-defined position in time. But I need to pay attention. Actually listen. For blasting sound into my ears while doing other stuff I can also use lossily compressed music, and even my Bluetooth AirPods Pro.

A Dream Come True, But…

I find it exciting to have technical means to record, produce, and play music at a quality level we could only dream of many years back. Thanks to the better technology, I can experience old analogue master recordings exactly as the audio engineer intended, any bad remastering notwithstanding. I have better technology than these audio professionals back then. There is no quality loss from the the master tapes to the final medium anymore, as before with vinyl records. I don’t even mind some noise from these tapes.

I am not sure if these technical means are actually always used to the full extent of their potential on the recording and production side these days, based on my experience with the different music files I have purchased. They for sure are not on the playback side, considering what we have in the phones everyone uses for their music listening. Maybe it’s considered not to be worth the efforts and costs in a world where most music is streamed in low quality, and with paying customers nonetheless. One has to find the jewels.

It’s somehow ironic that with much more capable technical possibilities to enjoy music better than ever before, the mass markets are not taking advantage of them. In the video world, everyone wants ever higher resolutions and refresh rates, but in the audio realm the requirements and expectations are much lower than ever before. It’s like being content with blurry 720p videos.