Science and technology

Does hi-res audio sound higher than CD high quality?

In response to a current article, a reader wrote:

Is there nonetheless any want for compressed knowledge? As far as I can inform, nobody has ever in a double-blind check been capable of acknowledge something higher than CD-quality uncompressed audio. MPEG provides options that may be recognized with apply. In concept, FLAC would not, however the mere act of decompression can result in jitter. As these days there isn’t any have to compress CD high quality, then why trouble? It could also be price having greater bitrate and depth on audio recording/processing, however for listening functions, CD is nearly as good as you want or certainly can get.

These are all fascinating factors and properly worthy of consideration. Let’s take them point-by-point.

Do we want compressed audio knowledge?

For me, no less than, the reply is a transparent sure. I do most of my day-to-day work on my high-quality, gracefully growing older, System76 Gazelle laptop computer. Some years in the past, I changed the 500GB 7200RPM onerous drive with a 480GB SSD. As far as I may decide on the time, my finest guess, by way of efficiency and reliability, was a SanDisk ExtremePro, which particularly appealed to me due to the included 10-year guarantee.

I additionally journey rather a lot. When I am away I actually get pleasure from listening to my music assortment, which is usually FLAC, fairly a little bit of it ripped from CDs, but in addition greater decision stuff, usually as much as 24-bit / 96 kHz pulse-code modulation (PCM) knowledge, with a small scattering of MP3 recordsdata that had been the one choices for buying sure music. In complete, this assortment takes up:

clh@avignon:~$ du -s --block-size=G Music
237G Music
clh@avignon:~$

So, of the 480GB offered by my SSD, 237GB is taken up by music. That leaves me one other 200GB for the system and my work, with a little bit of a buffer. If I moved from FLAC to WAV, that may roughly double the quantity of area wanted, which would depart me no room for work (or worse).  I may purchase a 1TB SSD however, or I may change the CD/DVD ROM drive so I can rip the occasional CD and create the occasional DVD-RW backup. Practically talking, due to this fact, I will be dwelling with FLAC recordsdata on my laptop computer for now.

Recently, I purchased an exquisite standalone digital audio participant, the xDuoo X3 II, from Massdrop. This is a fine-sounding machine, constructed round an AKM4490 DAC chip. In phrases of performance, it is primary—no contact display screen, no WiFi, no Android—nevertheless it works properly and sounds nice. It performs PCM (FLAC, WAV, no matter) as much as 32-bit / 384 kHz, in addition to DSD stream recordsdata (DSF) as much as DSD128, so it greater than covers my wants. It makes use of a MicroSD card to retailer music recordsdata and my laptop computer has a MicroSD author, so I am good to go. I can copy my music library from my laptop computer to my new 256GB MicroSD card and revel in my music with out having to tear it to MP3 or another lossless format. But that 256GB MicroSD card is fairly darned full, so good factor I am utilizing compressed audio knowledge, i.e., FLAC format.

In abstract, I want round 250GB of storage for my music assortment and 500GB SSD and 256GB MicroSD playing cards are at an honest worth level. So, sure, compressed audio nonetheless issues to me.

In a double-blind check, can anybody acknowledge higher than CD high quality?

There are two points right here. The first is the concept double-blind testing is the final word manner of detecting variations in audio replica. The second is having the ability to acknowledge variations between CD-quality and better resolutions. Let’s take them one by one.

Double-blind testing

I get the significance of double-blind testing usually, however I additionally get that it is quite simple to demand it because the gold commonplace in detecting audible variations (between completely different enter knowledge, between completely different tools, between completely different rooms…) with out essentially being conscious of all of the shifting components.

First, let’s take into consideration what “music enjoyment” means. Evidently, a number of folks get pleasure from their music over $5 earbuds and smartphones. At the opposite excessive, some folks spend many 1000’s of dollars to equip their listening environments. I submit that not one in every of these folks spends a substantial amount of their time doing double-blind A/B/X testing to guarantee themselves that their supply recordsdata or tools or listening atmosphere or the section of the moon is making any distinction to their pleasure. Instead, they’re in a zone, maybe taking the bus or exercising or cooking to background music, maybe listening intently to the music and avoiding different distractions.

So how does the act of double-blind listening mimic, in any significant manner, the act of having fun with music?

Turning that query round, who feels comfy figuring out that a panel of individuals with completely different and unspecified ranges of schooling, musical curiosity, familiarity with devices’ sound (be they actual or digital or one thing else), curiosity in listening critically, and “taste” (no matter that is) can one way or the other inform us that there are (or aren’t) significant variations between element A and B or file X and Y?

As I become older, I discover I actually detest background music besides once I’m cooking—how may an individual probably make empanadas that look this good with out Violeta Parra playing in the kitchen—or when folks cease mid-sentence and say one thing like “Wow! I’m really digging this music you’re playing.” Instead, I save my music listening for once I’m alone and may savor what I am listening to. Moreover, I acknowledge my means of music appreciation has modified, and I’ve little or no curiosity in whether or not a panel composed of people that take heed to Spotify or MP3s on low-cost earbuds can inform the distinction between a CD-quality model of a given album and the high-resolution equal. It’s simply not related to me.

I’ve performed guitar since I used to be seven years outdated and I am virtually 63. I do know what a guitar feels like as my head is hanging over it and I hear what I am enjoying. I do know that it sounds great within the lavatory, much less so within the bed room. When I hear a well-recorded guitar, whether or not it is on LP, or CD, or high-resolution digital, I do know that it is well-recorded. When it is not well-recorded… no shock, it would not sound like a guitar; it feels like a mediocre recording of a guitar. But I do not know what a cello, or a violin, or a Moog XYZ actually feels like. I have never spent a lot time listening to those devices, and I definitely have not critically or fastidiously listened to recordings of them to detect variations. Have these focus teams listened critically and thoroughly and with plenty of prior expertise? Who is aware of?

Beyond that, what about recordings made, for example, with out nice care? Is the standard of these recordings in 24/96 going to be a lot better than the 16/44.1 variations? Is there a real distinction within the content material between the high-resolution and the CD model?

And lastly, what concerning the tools? Is it actually as much as reproducing the variations that would be detected in a double-blind check? Could, as some folks counsel, the tools’s bandwidth make the high-resolution recording sound worse (quite than the identical or higher)?

So, double-blind testing. From my perspective, except all of the variables (together with ones I have never thought-about) are fastidiously and correctly managed, who is aware of? When somebody figures out a option to double-blind check the best way I take advantage of music, possibly I will be extra satisfied.

Has anybody acknowledged a distinction?

Here we’re on extra strong floor. Apparently, folks have acknowledged the variations, as steered by Joshua Reiss’ AES meta-analysis. There’s additionally “Inaudible high-frequency sounds affect brain activity: hypersonic effect,” an article I discover fascinating maybe as a result of it is so splendidly medical.

Reiss’ paper is especially fascinating to me as a result of it (scientifically) discards an early paper that purports to point out there isn’t any audible distinction in medium- and high-resolution recordsdata. It says, “[r]esults showed a small but statistically significant ability of test subjects to discriminate high-resolution content, and this effect increased dramatically when test subjects received extensive training.” In different phrases, the typical particular person can hear a distinction, and individuals who have invested in fastidiously coaching their listening to can clearly hear variations.

Why we might (or might not) hear variations in so-called high-resolution recordsdata

Before we will decide whether or not we will hear a distinction between a high-res, CD, and MP3 model of the identical recording, we have to ask: “Is there actually a difference”?

Open supply to the rescue! Spectrum analyzer Spek is one in every of my favourite instruments for checking the so-called high-resolution music recordsdata I purchase to verify they will legitimately be known as high-resolution (and sure, I’ve complained to the obtain shops once I’ve discovered their high-res recordsdata wanting). I do not know if Spek’s Fourier transforms are completely executed (though as a result of it is open supply, I may discover out if I needed to). Nevertheless, Spek tells some fascinating tales.

Here is Orchestra Baobab’s “Woulinewa” at 24/96:

Look on the overtones on the market as much as 40kHz+. Think about filtering these to suit into 16/44.1 PCM. Now take into consideration analog-to-digital and digital-to-analog conversions that may—or may not—cope properly with the overtones at 15kHz to 25kHz. Slow filters (that do not eliminate aliasing errors), quick filters (that eliminate aliasing errors however might introduce section shifts)… who is aware of what occurs when producing a CD-quality model of this file?

But there’s additionally the reverse drawback. Here’s Trentemøller’s “One Eye Open” at 24/96:

Wow, have a look at that cutoff simply above 20kHz! There has to be some critical filtering taking place to make this monitor appropriate for 16/44.1 CD high quality.

Let’s take into consideration this a bit extra. First, in each music recordsdata, the overtones are at a reasonably low stage: -80dB or under. I hear skeptics on the market grumbling “of course we can’t hear that.” But we have to have a look at the variations between the high-level and low-level music to place this in context. A whole lot of the basics are within the -40dB to -45dB vary, so the distinction between the lower-level fundamentals, at -45dB, and the higher-level overtones, at -80dB, is just 35dB (contemplating voltage, each six dB is kind of twice as loud, so the distinction right here between -45dB and -80dB is a couple of issue of 100 – -80dB is 1/100th of -45 db.)

Moreover, we’re accustomed to greater frequencies at a decrease stage; have a look at the “green stuff”—it is within the -60dB to -70dB vary, that’s, solely 10dB to 15dB louder than the overtones (once more, an element of two to a few instances as quiet or as loud). Do we hear the inexperienced stuff? The yellow stuff? The cyan stuff? I do not know.

I may paste a bunch extra Spek screenshots that present content material above 20kHz, in addition to ones that present a brick wall filter. But my most important level right here is: Don’t anticipate your 24/96 (or greater) recordsdata to have a ton of high-resolution content material—generally, they do not!

According to this Drummer World report, cymbal frequencies run manner out to the inaudible. Cymbals make musical overtones that—in concept—solely bats can hear. But my level isn’t that ultra-high-frequency stuff is audible; it is that the issues we do to eliminate it may create a number of nasty unwanted effects. Of course, the speculation says it will not. But that is the speculation. Can I belief that every one ADC and DAC implementations get this proper? Hmmm…

What about evaluating CD-quality and high-resolution variations of the identical music?

I haven’t got plenty of music in each 16/44.1 and high-resolution variations, however I do have some Led Zeppelin duplicates. I used Spek to investigate the 16/44.1 model of “Immigrant Song” from the album Mothership and the 24/96 model from Led Zeppelin III (Remastered). I additionally stretched the Spek window for the 24/96 model so the 20kHz traces are on the similar vertical place. Here’s what it appears to be like like:

In the 24/96 model, there’s fairly a little bit of content material above 20kHz that lies between -60dB and -70dB, which isn’t all that far down. That content material was filtered out within the 16/44.1 model to get it under the Nyquist restrict.

So, utilizing the Spek check, this monitor appears to be like like an honest candidate for evaluating CD high quality and excessive decision. However, there might be different variations between the 2 variations.

For instance, one model might be compressed greater than the opposite or its general stage might be set greater. Compression (which entails shifting the sign stage nearer to the utmost—0dB—differentially, so very quiet passages are louder and loud passages are proper on the restrict)  is commonly performed to make tracks sound louder on broadcast media, which is deemed by some broadcasters to be interesting (see Wikipedia’s entry on loudness wars). This is clear within the model of “Communication Breakdown” on Mothership, the place the peaks throughout the intro seem between -3dB and 0dB, in comparison with the one on Led Zeppelin I (Remastered), the place the peaks seem between -6dB and -3dB. In different phrases, the intro on Mothership is twice as loud because the intro on I. You can see this distinction through the use of an amazing open supply audio participant (like Guayadeque) that has stage meters.

Another instance: Older music was typically mastered with the bass turned down as a result of heavy bass was not playable on low-cost file gamers within the 1970s, and when it was remastered the bass might have been turned again up. This might be examined with a spectrum analyzer that reveals ranges at a given instantaneous. Audacity has a plugin that seems to supply this type of measurement.

Or, there might be different modifications that make two music variations sound fairly completely different. “Communication Breakdown” on Mothership has been remixed in order that the intro guitar riff is solidly within the middle of the stereo picture, quite than off to the left as it’s on I (Remastered). It’s comparatively—and unambiguously—easy to detect stability reconfiguration with my ears, nevertheless it may also be perceived as stage variations between channels when evaluating one model to a different.

In abstract, it’ll be actually tough to make sure that “the same song” pulled from completely different sources at completely different resolutions sound completely different as a result of one is CD high quality and the opposite is excessive decision. A greater method is perhaps to start out with a high-resolution monitor with actual high-resolution content material, then filter and downsample it to CD high quality. But this solely addresses the query “should I buy the CD-quality or high-resolution version?” while you’re speaking concerning the high-resolution authentic with true high-resolution content material that is been filtered and downsampled to CD high quality.

My conclusions, for now…

There is not a simple option to kind this out, no less than from my viewpoint, however I desire to be cautious: I would quite purchase high-resolution recordsdata when doable (sure, I all the time purchase my music). My absolute favourite combo is 1) shopping for a wonderful LP, and a couple of) discovering a coupon inside for a high-resolution digital copy as part of the acquisition worth (thanks, Nils Frahm and others).

Then, in the future when I’ve a number of spare time and fancy tools, possibly I will have the ability to persuade myself I can hear the distinction. In the meantime, I do know I am not being cheated of the complete glory of the efficiency as a result of I chosen the 16/44.1 as an alternative of the 24/96 model.

And extra music

OK, sufficient ranting about excessive res vs. CD vs. MP3. Let’s speak concerning the music as a result of that is what actually issues, proper?

I’ve talked about Lenine previously, particularly his album O Dia Em Que Faremos Contato. My first contact with Lenine was a CD I picked up at Starbucks “quite some time ago” (for those who bear in mind shopping for CDs in your espresso store, you possibly can in all probability work out how way back this was), which included a monitor from one other Lenine album Na presão, “A rede.” Give this a pay attention; what a unbelievable monitor, what a high-quality album! Mine was initially on CD, however I’ve since ripped it to FLAC and saved the backup within the basement.

If you (like me) have a particular spot in your coronary heart for Jimi Hendrix, you really want to listen to Tangerine Dream’s version of “Purple Haze.” Oh pricey, that is past loopy! You can discover it on their 1992 tour album, Live in America, on 7digital, an exquisite, Linux-friendly digital obtain web site.

Finally, after listening to a radio edit of “Sharing” on a compilation album for a variety of years, I picked up the Bugge Wesseltoft album Sharing on 7digital; though it seems to have vanished in Canada, it is nonetheless accessible—no less than within the United States and New Zealand. This is past high-quality music; I encourage you to take a look at the remainder of Bugge’s work, as anybody that gives a New Conception of Jazz deserves some critical research.

Most Popular

To Top