in ,

Beginner’s talk: why CD quality still matters

  • The language of the digital audio world is important, especially when it comes to talking about what streaming services do and do not offer and the associated qualitative implications.

    Apple Music and Spotify use lossy compression in order to keep file sizes small(er) so that the end user experiences minimal lag between hitting play and hearing music. This in turn increases the likelihood of glitch-free playback experience, especially when bandwidth is narrow. Another benefit of lossy compression is that ISP data allowances don’t get chewed after only a few albums. Lossy compression typically uses up to 10x less data than its lossless counterparts. Talking of which…

    Tidal and Qobuz’s higher quality streaming options might each be branded “Hi-Fi” but high fidelity doesn’t mean high resolution. The devil is found in the details of the marketing spin.

    Consider the landing page for Qobuz’s streaming services.

    The promo graphic speaks of a “high definition” experience but click through to the other side and you’ll see this sneaky wordplay translates to plain Jane lossless Redbook. It’s of higher audio quality than Spotify and/or Apple Music but it is emphatically NOT hi-res audio (HRA). Qobuz knows that the mainstream has been ensnared by lossy audio for so long now that only the recent talk of HRA (and its associated language) will extricate some people from the net.

    Like Tidal Hifi and Qobuz Hifi, the digital audio stream comes down the pipe at 16bit/44.1kHz. That’s the same resolution as a CD but compressed with the FLAC codec to use less bandwidth and ISP data allowance.


    Don’t be alarmed by the C-bomb drop: compression doesn’t always mean audible sacrifice. With FLAC, the audio is compressed without discarding data. That’s why Tidal and Qobuz’ s Hifi services sound better than their own lossy services as well as Spotify and Apple Music. The catch? Half-decent hardware is required to reap the audible benefits of shifting from half- to full-fat audio. Take the Tidal test to see if your hifi is ready to put on a little weight.

    What qualifies as half-decent is beyond the scope of this article. Eliminating Bluetooth speakers is a solid start. Not simply because of Bluetooth’s own lossy compression but because the hardware is severely limited. I’m looking at you Beats Pill and UE Boom. Besides, acquiring an integrated amplifier and a pair of standmount speakers will give you a proper stereophonic experience that these all-in-one Bluetooth devices never will.

    The next step is to source music that hasn’t been subjected to a lossy compression algorithm like MP3, M4A (Apple Music) or Ogg Vorbis (Spotify). However, be wary of those who would hijack the language associated with hi-res formats to sell you what is essentially CD quality. The newcomer is encouraged to bring proper scrutiny (and some healthy skepticism) to phrases likes “as the artist intended”.

    Last year the major labels and the Recording Academy met to formalise what was already accepted knowledge amongst many audiophiles: that “hi-res audio” means any digital audio file capable of bench pressing more than CD’s 16bit/44.1kHz. That is, anything with a bit depth greater than 16 and a sample rate higher than 44.1kHz.

    24bit/96kHz and 24bit/192kHz are bit depths / sample rates typically seen over at HDTracks, the Qobuz download store and Pono Music. However, many newer releases are creeping in at 24bit/44.1kHz or 24bit/48kHz. They meet the HRA classification because they offer more data than Redbook’s 16bit/44.1kHz but whether or not that slender slice of extra data is of audible benefit to the consumer remains debatable.

    Neither Qobuz nor Tidal yet offer HRA streaming. That means a user acting on HRA’s promise of superior sound quality (compared to CD) is required to pony/Pono up ~US$15 for an album download. It’s yours to own but US$15 represents three quarters of a month’s subscription to Qobuz Hifi or Tidal Hifi.

    Moreover, listening tests conducted by yours truly have yet to expose ANY audible benefit in shifting gears from 16bit/44.1kHz to 24bit/44.1kHz. Don’t those additional 8 bits merely serve to lower a noise floor that’s already well beyond the range of human hearing? Redbook detractors might argue that our bodies respond to frequencies beyond 20Hz – 20kHz but ask yourself this: 1) is your hifi system good enough to reveal those differences and 2) is the hi-res download vis a vis a CD ripped to FLAC worth the extra cash?


    I sometimes (but not always) hear a difference when comparing files encoded at 24bit/96kHz (or 24bit/192kHz) to the CD equivalent but complete exposure to this delta demands the use of really good and (often) expensive equipment, like that which audiophiles generally own.

    In the case of 48kHz sample rates and above, not only does the file provide additional resolution but it also pushes the digital filter, required by the D/A conversion process, above the upper limit of human hearing. With 44.1kHz audio, the filter sits at half the audio’s sample rate: 22.05kHz – that’s only a smidge above 22kHz. With 48kHz, the filter is applied at 24kHz. With 96kHz audio, 48kHz is the filter’s location. Guess where it acts on 192kHz audio? Yup – at 96kHz.

    In launching his Pono Player last year, Neil Young could be heard kicking Redbook almost as much as lossy codecs like MP3. As I see it, the conversation should not be boiled down to a matter of HRA vs. MP3. It’s an over simplification that omits the middle ground of CD/Redbook audio.

    Audiophiles love to mass debate all manner of issues: vinyl vs. digital, DSD vs. PCM, even 24/96 vs. 24/192. Not only is such tech talk a niche (hi-res) within a niche (digital audio) within a niche (audiophillia), to your average mainstreamer it’s overwhelming and a little off-putting.

    Some audiophiles are their own worst enemy. Their advocacy for hi-res, DSD or MQA confuses a simple message: that lossy audio should be the first casualty in the push for better digital audio playback.

    Less frequently discussed is the magnitude of difference between Spotify and Tidal Hifi or Apple Music and Qobuz Hifi and how those deltas compare to that between Redbook and HRA. For this listener, with all other things held equal, the former is easier to pick over the latter: the move from MP3 to lossless audio is more noticeable than the change from Redbook to HRA. However, with 99% of the world’s music supply topping out at 16bit/44.1kHz, the argument is purely academic, whereas the promotion of lossless over lossy audio is not.

    Redbook rocks.

    Further information: Wikipedia

    John H. Darko

    Written by John H. Darko

    John currently lives in Berlin where creates videos and podcasts and pens written pieces for Darko.Audio. He has also contributed to 6moons, TONEAudio, AudioStream and Stereophile.

    Darko.Audio is a member of EISA.

    Follow John on YouTube or Instagram


    1. Agreed. Lossy codecs had a place in the past, but with 4G phones with 32GB of data, and broadband and TB drives there era of lossless codecs is largely made irrelevant. There is very little case to made – other than lowest common denominator compatibility – to made the case for lossy codecs. And there is no reason to take a step back in source material from that made widely available in the 80’s on the lowly CD. It’s isn’t even a matter of ABX testing or any other nonsense. Give me all the data I pay for with the CD’s I purchased, and don’t let some computer programmers perceptual algorithm decide which 80% plus of my data to throw in the garbage can. Yes, FLAC/ALAC should be the new normal. Not prehistoric lime wire, kazaa, napster era music files.

      • What planet are you on? Where it is, I’d like to sign up for that unlimited data plans you got there.

        Lossy codecs are extremely relevant and a god send for those who want to keep their cell phone bill down.

    2. I’m fairly convinced that 16bits/44.1 khz is generally sufficient, however (and please correct me if I’m wrong) I was under the impression that:

      * the master is oftentimes recorded at 24 bits.
      * the CD version (16 bits/44.1 kHz) can have the dynamic range severely compressed (friggin’ loudness wars)

      Thus, one can observe a better quality at 24 bits, even though it is for the wrong reason.

      Just my opinion, of course.

      • An artist records an album and it is mastered. The mastering engineer produces a file encoded at (for the sake of argument) 24bit/96kHz. From that same master, a CD is made (by downsampling and dithering to 16bit/44.1khz), an MP3/OV/M4A version is encoded from that CD downsample for Spotify et al, the vinyl is pressed and a hi-res version (that same 24/96 file) is given to HDTracks and the Pono store. Each format will feature the same amount of dynamic range compression.

        • A little too simplified.

          I have seen several albums where the different formats were derived from the same master, but different levels of volume compression were used. Off the top of my head, I’ve seen this with both older and recent Paul McCartney albums and also Green Day’s “American Idiot”.
          The hi-res releases had significantly less VC added compared to the mp3 and Redbook releases. This alone could account for them sounding better, even if you don’t think hi-res intrinsically sounds better. There are more albums like this that I can’t think of at the moment, and some of them are recent releases, not some special mastering from catalog files.

          • Yes, you have a point. There will always be exceptions…but then we’re back to conversing about marginal concerns: this album here, that album there. My POV here stems from being sufficiently confident that the majority of albums DON’T enjoy finessing for vinyl and/or HRA formats. Or am I wrong about that? Happy to be corrected.

            • No you are not wrong, I wish you were. What really annoys me is expensive hi-res versions of albums sold as “audiophile” that have been heavily volumc compressed. If you want to sell to the audiophile market and reap the extra profits per album sale, at least have the sense to lay off the heavy-handed limiting. Audiophiles usually know and appreciate the value of a volume control.

      • *Lossless data compression is fine no problem there.
        *Lossy data compression tosses out data (CD to 236kbps AAC for example tosses out ~82%). Data is gone forever but no need to worry because a computer engineer created a perpetual encoder that decides what data is least offensive to remove. 🙁
        * Dynamic range compression and limiting just makes the music louder for cars, ear buds, loud environments, and to get your attention. Analogous to demo TVs being set-up in “torch mode” or TV commercials before congress outlawed this practice in the US.

    3. Good article. What you fail to mention is that a high-resolution bucket does not define the quality of it’s contents. You can digitize a 1910 Edison Cylinder recording to a 192khz/24bit file and it will never sound any better than the original lousy quality. This is THE MAIN reason why people will fail CD vs Hi Res comparisons. The original recording was done using sub-CD quality equipment (analog tape, narrow frequency response microphones, etc.). There are very few actual high resolution recordings available commercially. The ONLY mainstream (Popular music) high resolution recording currently available is the new James Taylor album.

      P.S. Neil Young has NEVER recorded any of his music in high resolution!

      • Hence why I said “with all other things held equal” in the final para. I didn’t wanna get into what goes on in the studio or at the mastering stage. I also realise that in some instances an alternative, less dynamically compressed master is used for vinyl or hi-res releases, but that doesn’t happen anyway near as frequently as people like to think. Moreover, nowadays the studio master encoding ‘standard’ (if you can call it that) for large majority of contemporary releases is 24/44.1. THAT’s why we see so many 24/44.1 encoded popping up on HDTracks and Pono Music.

        • Yeah the source material argument is a red herring. Discussing formats assumes high-quality source material and mastering otherwise it is not a bottleneck and isn’t really germane.

        • It may be true that a lot of these files are coming out at 44.1khz 24 bits, but the recording equipment that they are using can’t even record with 24bits of dynamic range anyway!

          • Yup – which is why 16bit/44.1kHz is often* more than good enough.

            *There will always be people who insist that 24bits enhances the music somehow.

          • That is an overly simplistic interpretation of the benefits of bit depth… something that I typical see when something xiph sourced. Other important effects include quantization noise (error) where rounding has to occur between rungs. Not a big deal if you are talking loud sounds, but quiet passages the rounding can result in a large % error. This is one of the reasons quiet passages in classic and jazz can sound grainy. Dither helps, but 24-bit just makes this stuff a non-issue and were only talking a 100MB delta per “CD”. Not worth tightening the belt here if you can help it. Sure good mastering can help but 24 bit just makes this so much easier.

            Greater sampling frequencies just make the digital filtering so much easier and the slopes can be gentler resulting is less phase issues and ringing.

            Sorry to sample on but good CD quality makes a great baseline IMO and 24-bit >=48KHz just really reduce the stringent constraints to get the most out of redbook. But gawd just no AAC or MP3 please! ;(((

            • Thanks for the another tech viewpoint on those additional 8 bits. Alas, with the music I listen to, I just don’t hear the benefit. Besides, as you say, of greater priority is to source 16/44.1 lossless that hasn’t been sullied by lossy codec. Not that MP3 or OV is unlistenable but with Tidal/Qobuz Hifi so affordable, it’s just so easy to punch in a lossless version.

      • How do you come up with that? There are definitely albums recorded in 24/48 and 24/96. May not lots, but to say they don ‘t exist is silly.
        As far as high-res, then we can argue over the definition. I have dozens of albums originally recorded to tape that have been remastered to hi-res digital. Some of them are the best sounding recordings I own and sound better than CD and LP versions of the same material.
        You can argue all you want about the limitations of tape and microphones, but the fact remains that very high quality recordings were being made even back in the late 50’s. Many of those recordings are superior in many ways to good recordings being made today. They may lack a bit of frequency extension at the extremes, but that doesn’t mean they don’t sound great – often more “real” than a recording made with equipment that is better on paper. The skill and knowledge of the recording/mastering team is more important than the format and equipment.
        I have no problem paying for a well done hi-res transcription of a good quality analog recording. If the original recording was a good one, I want to get the best quality digital version I can.

        • The definition of high-resolution is THE important part! There are many GREAT recordings that have been done on fabulous tape machines and microphones but NONE of them are high resolution. The BEST tape machines and tape media do not cover the range and depth required to be defined as high resolution. Just because something is not high resolution doesn’t mean that it can’t be great. Nor does a High Resolution recording mean great sound quality. High resolution is NOT a measure of the quality of the recording. It is a measure of the POTENTIAL of the quality of the recording. This is why the provenance of the distributed song is so important and why it is almost never revealed. A recording could be listed as Studio Master, High Fidelity, Master Quality, etc., but unless it was recorded with the potential to capture above CD quality, it should not be labeled or marketed as High Resolution!

          • Well we sort of disagree. I have no problem with a transcription from analog being called hi-res, as long as it is clear that the source is analog. It is hi-res, in relation to a non-hi-res transcription of the same material.

    4. I much prefer the more detailed looseless sound from Tidal but I am constantly plagued with ever frequent buffering and skips on both the iPhone app and desktop Mac, even though I have the fastest possible wifi speed and can easily stream HD video with no issues. I wish Tidal would improve it’s technical end and can’t wait for NBN to reach me. Seriously considering going back to Spotify just to stop the frustrating interruptions. Rock – hard place!

      • Hi Tagjazz – many Australian ISP’s seriously underestimated the impact of the local launch of Netflix on their networks. I’m with iiNet and their bandwidth fluctuates wildly now that their customers get to enjoy UNMETERED Netflix. Very frustrating.

    5. I just thought I would throw this out there since I do it… For others who are trying to find the best CD pressing they can get their hands on… A lot of times the US version of a CD will score very badly with dynamic range testing because the US versions were optimized for radio airplay aka loudness factor so they stand out.

      Is a database with literally thousands upon thousands of CD and Record dynamic range scores… I will often hunt down and pay $12 for German Elvis Costello CD (For instance) because the DR score is 13 vs. the US version for $8.00 has a score of 9.

      This cuts both ways as sometimes you can’t find the Japanese or German or whatever version with higher scores so you go in knowing that your version is not the best…. even so I thought it was a resource worth putting out there.

      • Interesting. I once sold a UK CD of Radiohead’s The Bends to an Aussie who was delighted at how much better it sounded than his locally-sourced CD. And yes, the DR database is a useful resource. Just imagine if Tidal or Qobuz could provide the same kind of info! We can but dream.

      • Thanks for the link!

        A reasonable way to test dynamic range compression across the board: take a number of albums that have both cd and vinyl, and compare the dynamic ranges. Combine the numbers using a popularity weighting.

        Sigh … I like the Killers, but their CDs suck … big time. Now I have an objective metric for that.

    6. Not sure if this is pertinent, but my CD player allows me to up sample to 96/24 or 192/24. I cannot hear any difference at all…

    Amarra For Tidal: a better streaming sound for OS X, Windows

    Streaming royalty payments: where does the money go?