Saturday 4 November 2023

MUSINGS: Multichannel mixes are more dynamic than 2-channel, stereo releases. (Going beyond 2-channel stereo legacy.)

Last week, in the Part II review of the Topping PA5II+ amplifier, I talked about a few albums like The Rolling Stones' Hackney Diamonds and the sound quality of the album. For years now, I'm been unhappy with the status quo of ubiquitous Loudness War, dynamic compressed albums. We've even explored using DSP to artificially improve the content for example.

A complaint I have with the Stones album is that the dynamic range of the CD is poor at DR6; and yes, this is an audible limitation. However, if you like this album, I recommend listening to the multichannel/Atmos stream or grabbing the limited edition BluRay import box with 7.1 TrueHD-Atmos mix which is much less compressed, with album dynamics averaging DR12!

[For simplicity, I'll stick with the DR "crest factor" result due to familiarity for many of us. These days we can talk about LUFS, EBU R128, etc. as discussed before as measurements of volume and for normalization. Check out the DR Database for results from many albums.]

In the last few years, I think some of us will have noticed this pattern of much more gentle use of dynamic compression in multichannel content. This is why a few months ago, I suggested that multichannel streaming simply sounds better than standard stereo 2.0 to my ears (this possibility was also suggested back in late 2022 and discussing this Steve Wilson multichannel mix).

Let's have a look at some examples.

Example 1: The Rolling Stones - Hackney Diamonds BluRay - 24/48 7.1 vs. 24/96 2.0:

So let's pick a song off Hackney Diamonds to examine. I like "Whole Wide World":

Notice comparatively just how much has been clipped off the top in the 2-channel version. As with heavily compressed recordings, it looks like a lawn mower has shaved off the waveform extensions. Clearly, that 2.0 mix is in no ways utilizing the potential of 24-bit digital! In contrast, the 7.1 mix, even with the Dolby Atmos metadata ignored (no height channels), we see a much more dynamic recording with clear use of the center channel, a little bit of LFE (low frequency bass effects), small amount of side surround channels, and more rear surround content.

The beauty of having the multichannel data is that we can easily downmix that 7.1 version into a 2.0 version - here I'm just using default parameters in Adobe Audition to collapse it down:

IMO, this automatic downmix from multichannel is a much superior sounding version than that nasty stereo CD or 24/96 2.0 mix. The nuances are noticeable. You can turn up the volume and experience more impact from percussion transients. It's less fatiguing on the ears. The vocals are clearer, better isolated from the instrumental parts which makes the whole experience less congested.

Let's look at another example we talked about last week...


Example 2: Ferry Corsten - Blueprint BluRay - 24/96 5.1 DTS-HD Master Audio vs. 24/96 2.0


On top we have the 2-channel 24/96 mix off the BluRay, and below, a rip of the 5.1 24/96 DTS-HD Master Audio track. By the way, this is the Auro-3D data but since I don't have a decoder, there's no expansion with extra height channels.

We can see the center channel being used at the start of the track for the narration and dialogue to advance the story.

At least the stereo DR8 is better than DR6 above, but again, the lawn mower has done it's job with keeping all the peaks limited.

Let's downmix the multichannel to stereo:

Nice reconstitution of the peaks using the multichannel data with a DR value IMO at what it always should have been for the stereo mix. The effect upon listening is similar to what happened with the Stones. More spacious, with a tangible improvement in separation between the gentle male narration of the first part and the ethereal female synthpop/electronica of the latter portion. 

So far we've looked at a couple of examples from BluRay physical media. But these dynamic range benefits also extend to more popular multichannel streamed content.


Example 3: Taylor Swift - 1989 (2023 Taylor's Version) stream - Apple Music EAC3-JOC Atmos extracted to 24/48 5.1 vs. 24/48 2.0 HDtracks download:

Simply awful - that DR5 HDtracks download is LOUD. Not much subtlety there, audiophiles. Granted, for pop music like this, I'm sure many listeners like it this way since we've become desensitized to this kind of sound for at least two decades now in the rock and pop word, sometimes intruding into soundtrack scores.

Yes, the compressed stereo is "punchy", aggressive, there's clearly an accentuation in bass and highs. The problem with this kind of sound is that it's fatiguing and qualitatively I think most listeners will recognize in A/B listening that "shades of grey" are missing within the volume, spatial width, and even temporal flow since each note and percussion beat blasts monotonously in our ears.

This excess compression I think in particular diminishes the sweetness of slower ballads like the song "Wildest Dreams".

I am somewhat disappointed that the 1989 multichannel mix for some reason decided to not use the center channel. We'll talk about this more below. Like the examples above, let's down mix to 2-channel:

Again, the downmix is clearly more dynamic than the supposedly "hi-res" 24-bit version of the album from HDtracks. To me, it's a scam that 24-bit music can be so dynamically squashed down to DR5 and still sold to consumers as if it's in any way higher resolution! Based on total amounts spent over the years by consumers, purchasing music like this off HDtracks and other sites like ProStudioMasters, this might even be the largest waste of dollars in all of hi-fi even beyond snake oil hardware. I've tried my best to provide fair warning that this could be a problem since 2014.

As I discussed years ago, be careful about "hi-res" albums these days, many if not most mainstream releases are not worthy of all those extra bits, nor your money. Unless you know for sure the quality of a hi-res recording you want, I would stay away.

Regarding the center channel...

I've heard that some music labels don't like to mix with a center channel and there's discussion here with Andrew Scheps claiming that center channels sound "really weird". I dunno guys, I tend to like my music with the center channel used and am not sure what he means by "phantom center sounds like records... (center channel) just sounds unmixed". I see that they're just talking about issues around binauralization with headphone listening, not in a high-quality multichannel speaker system.

This sentiment is unfortunate. I would think that it's better to do things "right" which is to utilize the full hardware capabilities of the system, including the center channel with the hope that in time, the binauralization issue they speak of with poor comb-filtering during playback can be improved rather than being stuck with baked-in phantom center just because it might sound better with current hardware and algorithms in November 2021 (when that video was streamed) over headphones.

By the way, an album like Billy Joel's An Innocent Man (1983) which extensively uses the center channel for vocals in the Atmos mix sounds really good on the multichannel speaker system!


To have the main vocals front-and-center anchored in the room sounds "right" to me. Plus for listeners sitting off-center, the fact that the voice is still coming from that central axis in the room I find beneficial. Whether this sounds "really weird" or "unmixed" with headphones, I guess I'll leave to folks who mainly listen to headphones to decide!


Example 4: Megadeth - Rust In Peace stream - Apple Music Atmos vs. CD & download versions over the years!

Most audiophiles I know would not think much about heavy metal albums as within the "audiophile repertoire". Not so fast, friends! Years ago in 2014, an audiophile friend contributed to this blog on the importance of dynamic range to his metal-listening pleasure.

One of the local metal-audiophiles was able to get me examples of the Megadeth album Rust In Peace with various masterings over the years for the "classic" track "Holy Wars... The Punishment Due". First, let me just show you what the Atmos "Spatial" stream looks like in 5.1, along with the downmixed 2.0 version which measures DR13:


Let's compare that DR13 downmix to versions over the years...


Back in 1990, if you bought this CD, you'd be treated with a DR13 version. In 2004, the mix/master got compressed down to DR6; this is the common pattern for many albums as they got the "remastered" treatment in the early 2000s and very much to the present. As you can see, they used this same DR6 version for the HDtracks "hi-res" version 10 years later but the average volume level got raised a little bit more.

And finally at the bottom, with the multichannel mixdown to 2.0, we're back to DR13 dynamic range! While the sound has been remixed for the multichannel compared to the 1990 CD, so there are slight changes, there's something poetic about this sequence of events in a "circle of life" kind of way.


Fantasy of a real acoustic space.

Summary...

In the past, I would sometimes collect vinyl-to-digital rips because of their more dynamic character which probably are the result of a number of factors likely stemming from the limitations of vinyl or even the result of playback using the mechanical system. These days, I believe that digital multichannel mixes are superior sounding and the 2-channel downmixes that can be created from them, especially with new music, have been excellent; for the most part, much more dynamic than the corresponding CD releases.

Of course there are still many old albums only available as LPs out there worth digitizing - for example I love the old '80s remixed singles, many not reissued or at least not easily available on CD.

Years ago, I suggested that music be released as two versions - the "Advanced Resolution" and "Standard" releases differentiated mainly by dynamic range since just the technical specs like 96kHz samplerate with hi-res frequency response expansion and 24-bits alone do not add much to audibility if the mixing/mastering is poor. We're seeing that divergence in sound quality now. Multichannel/Atmos mixes are typically done to a higher standard of dynamics perhaps akin to the excitement we're used to experiencing in multichannel movie soundtracks, compared to the routinely "crushed" 2-channel releases. (Hey, why should the videophiles have all the fun with HDR content?!)

Rejoice in this excellent development, audiophiles! The more refined music listener can now find the dynamics again in these multichannel mixes whether directly decoded to multichannel speakers, downmixed to stereo, or virtually "binauralized" as "spatial" 3D headphone playback.

Sadly, I suspect the culture of "loudness wars" 2-channel mainstream releases on CD and stereo downloads (including hi-res) may be foreseeably crystallized within the psyche of artists and record labels. Having said this, although not to audiophile standards, low-DR, loud music does have its place for consumption especially in cars and mobile applications where ambient noise levels are higher and we need the music/sound to rise above that. 

Notice that even though lossy, Dolby EAC3-JOC Atmos streams have much more sonic data spread across those 5.1 bed channels. While not ideal, and I would love to hear the lossless version if possible, I would not be concerned at all about the quality of lossy multichannel at very good typical bitrates like 768kbps. 

Yes, the multichannel content also sounds very good downmixed to stereo and played back in ambiophonics as well.

A note to audio engineers... Please don't mess this up! Fine, go ahead and give the artists and record labels (who don't know better) their loudness-compressed stereo mix. But when it comes to the multichannel content which most likely will be streamed through volume normalization so overly loud stuff will actually be reduced in volume ("wimpy" as per Mitch Barnett), or sold as premium BluRay packages to those who more than likely care about quality, please respect the consumer and give them a premium resolution and high dynamic range experience with all the benefits of the expansive immersive audio. Thus far, I have not heard many issues, but there are some pretty poor multichannel remixes out there which we might talk about down the road.

[Apple's Immersive Audio Source Profile specifies the integrated loudness value not to exceed -18 LKFS/LUFS (Loudness, K-weighted relative to full scale), and true peaks to not exceed -1dBTP as per ITU-R BS.1770-4. Enforcing these expectations will go a long way towards preventing the undisciplined loudness madness seen with CDs and stereo hi-res downloads.]

Noticing these changes to music quality, over much of this past year, I've been adding more multichannel content to my music library either to play as multichannel or letting Roon downmix to 2-channel; it does a fine job. The future of my music library more than likely will be increasingly multichannel. With that extra surround information, I hope applications can evolved DSP playback in ways that will push the audio experience forward. Perhaps more sophisticated and customizable downmixing options. Better binaural "generic" playback with customizations, ultimately even measuring and applying our own custom HRTF when headphone listening. 

[Hey Roon: How about Dolby TrueHD and AC3/EAC3 (.m4a) 5.1/7.1 bed channels decoding for those of us who have multichannel content even if Atmos is not licensed, height channels not decoded, and we can't do bitstreaming? Lots of potential in the Muse DSP system for multichannel, plus implementing ambio/crosstalk cancellation natively would be awesome.] 

While it might be hard to see in 2023 because 2-channel PCM audio will remain foreseeably the base, lowest-common-denominator format for audio delivery, I think the days are coming when 2 flat channels will be seen as a legacy format rather than the preferred or definitive version created by artists.

Oh yeah, one last thing. If you're listening to Dolby EAC3-JOC streamed lossy Atmos mixes (like off Apple Music, TIDAL, or Amazon), typically at 768kbps and 48kHz, these generally have good frequency extension out to 20.5kHz. For example, here's what the 5.1 FFT looks like for Taylor Swift's recent "Wildest Dreams" (Taylor's Version):

Notice the 1989 mix is essentially 4.0 audio content. We see the pair of front L&R signals highest level, then the two rear channels lower, then some center channel content significantly below, very quiet. LFE channel basically nothing when compared to fronts and surrounds.

Music I've seen encoded and then decoded from MPEG-H / 360 Reality Audio typically do not have content reaching 20kHz, usually rolling off by about 18-19kHz as I recall. Not that I think this is even perceptible, but as an audiophile desiring reproduction capable of the limits of human hearing, the ability to encode a full spectrum 20Hz-20kHz would be better.

--------------------

Especially for Beatles fans, their single "Now And Then" is out along with "Love Me Do" (2022 remaster) on the B-side. It's also available for those who enjoy multichannel streaming. For many, I think this will be a tear-jerker. They used AI/machine learning to help separate Lennon's vocals, piano, and reduced noise.

We don't really know if there's much else in the Beatles' vaults from back in the day, or if commercially many music lovers still care since music interests and styles have moved on over the decades. This could be the sentimental, last "new" farewell song. The sweet music video suggests that this is indeed a final goodbye from the boys now that Paul is 81 and Ringo 83.

The Beatles were around at the transition between mono to 2-channel stereo back in the 1950s. Perhaps as their releases come to the end with the recent remixes to multichannel/Atmos over the last few years, this also marks a transition to the next level of audio quality. They were pioneers in studio techniques and I appreciate that they continue to innovate with technology these days whether multichannel or AI reconstruction of the music (and video). This is in part what relevance, and brilliance look like.

A final message to 2-channel-only audiophiles, the "slick" audiophile magazines, and purveyors of "High End" anachronistic technologies. Why do some of you think that spinning vinyl disks or audio constructed from magnetic tape represent the ultimate evolution of sound quality? Why should there only be 2 channels when our ears and mind have the ability to perceive in 3D audio (to make matters worse, typically idealized with speakers in an equilateral triangle)? Why do some seem to take a hard stand and would rather spend enormous amounts of money on old technology and paraphernalia that could never truly reproduce the sound of an accurate acoustic space without coloration?

There's nothing wrong with enjoying the familiar, but do not hamper technological innovation. As a "hi-fi" hobby, let's make sure to open our minds to possibilities. Let's put value into the things we can objectively point to that make a meaningful difference. Be mindful of what is knowledge, what are opinions; even if supposedly these opinions are justified by some as based on "my experience" - is that so special, are your ears/brain so remarkable? Especially if you're a 70+-year-old audiophile reviewer looking to leave some kind of thoughtful legacy... Maintain the mindset of youth, the embrace of what is possible, imagine and add to the discussions of what is very much, simply better - even if these things might not be what certain advertisers have to sell, what you personally might be able to accommodate, or the predominant viewpoints of the current hobby culture.

I hope you're enjoying some excellent-sounding content, dear audiophiles, as we enter November with the Holiday Season fast-approaching again this year! Time flies...

12 comments:

  1. Hi Arch-
    For those of us who are going to stay 2 channel (no room for more): what's the simple relatively inexpensive way to playback atmos, etc, mixed down to 2 channel?
    Is it just software, or do we need some special PB equipment?
    I assume I could buy a blu-ray and rip the multi-channel to my HD.

    ReplyDelete
    Replies
    1. Hey there Danny,
      Yeah, depends on your set-up and source. For example, I can play Apple Music from my AppleTV through HDMI to my receiver and set the receiver to only 2 channels which will make the receiver downmix to stereo.

      In time, I hope more audio hardware able to stream the Atmos/MCH material off Apple Music, Tidal, Amazon, etc...

      On a computer, yes, you could rip a BluRay disc and convert the multichannel/Atmos soundtrack using DVD Audio Extractor to FLAC stereo. I would personally extract to 5.1 or 7.1 and keep that in my library so when it's time to playback, I can just use something like Foobar or JRiver to downmix to 2 channels. I'm sure many players can downmix. Otherwise I can still play them on a multichannel system and use Dolby UpMixer or DTS Neural X for Atmos/height channel extraction (not as good as native Atmos but still excellent).

      I use Roon and that has no problem with identifying whether the player hardware is able to render stereo or multichannel with the 5.1 and 7.1 stuff in my library; downmixing as needed.

      Delete
    2. Just for my reference.
      I do have the blu ray of the album Hackney Diamonds, and wondered if in stereo downmix in the song 'sweet sounds of heaven' the voice of Lady Gaga should be at the same level of Mick or really only a back vocal? I watched a video clip where she sings with him live and there her voice is more pronounced. The stereo version on the provided cd is compressed so for me difficult to compare as this is another master.

      Delete
  2. Excellent article, I think lower DR value for MCH is done by design, considering that you add all the 6 audio channels during the playback.
    And you're right, the instruments separation in MCH is much superior compare to a two tracks mix.

    ReplyDelete
    Replies
    1. Thanks Blogue,
      I assume you mean higher DR value for MCH. If they can make sure to keep loudness values on the channels from getting too high, yup, when folded down the composite DR will definitely remain good!

      Delete
  3. Stereo was supposed to have a center channel in the beginning. One of those things you have to be really old to know.

    ReplyDelete
    Replies
    1. Thanks for the note Steve,
      Did that ever happen to any extent back in the day? I know there are a number of 3.0 recordings with discrete center from decades back. Mercury Living Presence, Nat King Cole albums among the ones I have here (typically on multichannel SACDs).

      Delete
    2. The other Steve is referring to Bell Labs work from the 1930s, which led to the industry standard adoption of a center channel in movie theaters. I know of no notable penetration in the home consumer market until the advent of 'home theaters' and Dolby Pro Logic. Before then it was a rare 'hobbyist' thing to try a center channel or a 'mid-speaker' , though AFAIK no three-channel release formats actually existed for three-channel recordings, so this was really an early form of 'upmixing' 2-channel mixes.

      Delete
  4. The last Rolling Stones Blue and Lonesome album was DR7. I can't find a hires multi channel version online. Interesting to note the new remastered Steely Dan album Aja is DR15 24/192. Rare to see anything at this level.

    ReplyDelete
  5. The issue with using downmixed surround mixes as the alternative to low-DR stereo masters, especially for 'classic' albums, is that the mix you get, will not, of course, be 'the same' as the original stereo mix you know and love, despite perhaps having better DR than a modern dreck stereo remaster of it. Instrumental and vocal balances you are used to may or may not be preserved; reverbs may be different; parts may be missing or new. I often find that to the extent I know and love an original stereo mix, I am irked by the surround mix, and a downmix of same. Not always, for sure: Greg Penny's 5.1 mixes of Elton John's 70s catalog are spectacularly good and faithful to the sound of the originals. (But you know what? They are notably compressed ...check out Yellow Brick Road in any of the three formats, DVDA, SACD, BluRay) Ditto Ken Scott's remix of Ziggy Stardust. (No DR issues there...)

    (And this btw is where I differ from the general slobbering over Steven Wilson's remixes of classic prog albums. I find his work extremely hit-or-miss, even from album to album within the same band's output.)

    The other alternative, one I tend to favor (though I'll still always buy a surround release of something I love) is experimenting with upmixers. There, you know you're starting with the original mix. No worry about missing parts or alternate takes or different reverbs, and the balances can be at your command. For eyars I doted on the Dolby Pro Logic IIx upmixer in Music mode, but Dolby in its wisdom has eradicated DPL from AVRs since around 2014, replacing it with the inferior yet Atmos-friendly Dolby Surround Upmixer. So currently I'm playing with the old FreeSurround plugin for foobar2000 (which seems to have tried hard to reverse-engineer DPL IIx) and the more modern SpecWeb implementations.








    ReplyDelete
    Replies
    1. (and no worry about compromised DR...unless you start with a low DR stereo mix)

      Delete
  6. This has been true for many multichannel releases on DVD and SACD. Although sometimes you'd just get a crushed stereo mix with some reverb. Downmixing can occasionally introduce phasing artifacts when naturally panned (with a delay) voices exist in more than one channel. Most noticeable with percussion.

    I appreciate the lossy nature of EC-3. The file sizes of hires+multichannel are astronomical, and the work involved to reduce them detracts from the enjoyment. About 70 kbit/s is spent on the Atmos magic out of 768 kbit/s. It is really crude if you solo individual channels in Dolby Reference Player. The basic surround is good. I like that the bitstream allows for trimming without transcoding.

    ReplyDelete