On Stereophile's "Dolby Atmos: A Bleak Shadow?" - really? 🤔

I saw Stereophile's "Dolby Atmos: a Bleak Shadow?" article the other day. Honestly, I found it a bit irritating. The ideas seemed poorly contextualized with little background or meaning for what I imagine would be the average audiophile reader. It basically consisted of hearsay-level cherry-picked concerns as if these are significant (including that title).

Let's be clear about this. It's not like any of us must or even should love Atmos or multichannel. We can easily get through a lifetime of music enjoyment with 2-channels. Even though I like and even prefer multichannel/Atmos for many of my albums, it's just an option for those like myself who want to explore the surround-sound music experience as a superset of modern audio reproduction technology. Some artists and audio engineers these days are also exploring the extensive flexibility and ways to best create the mix. So if a traditional audiophile magazine believes that staying at 2-channels is all that's needed, well that's fine. But one doesn't have to write articles that seem to be trying to induce some kind of emotional response, stirring intrigue or concern even in just that title. (These days, sound bites and headline titles might be all that many pay attention to.)

So what do you mean by "bleak shadow", Mr. Lindberg (as reported by Jim Austin)? Are we talking just about numbers like bit-rate going down when a lossy stream is delivered to the consumer, or are we implying that sound quality changed significantly in a negative, "bleak", way? If it's only bit-rate reduction, that's totally fine, isn't it? Since the consumer isn't keeping the data, there's no point wasting network bandwidth! Streaming services like Netflix or Disney+ all stream lossy video and audio. Many customers, I'm sure including discerning ones, will not complain if they understand the nature of the technology and the quality limits depending on how it's consumed.

If Lindberg and Austin are talking about sound quality though, do you not think there's some hyperbole in that headline? Is it really that bad or is this just FUD being fed to audiophiles, stirring more anxiety than honest concern? Maybe Stereophile is just trolling and needed a headline like this as click-bait - I see their forum section has disappeared without notice recently, so maybe this can boost traffic.

If we are to try a blind listening test between 768kbps lossy EAC3-JOC Atmos and 24/48 TrueHD lossless Atmos (with the same 5.1 bed channels) of actual music, I wonder whether writers and readers of Stereophile would be able to show a clear preference; and if that preference is even for the lossless TrueHD! There are examples in the past to suggest that it would not be very noticeable. For example, 320kbps 2-channel MP3 sounds great compared to lossless CD and 16-bit vs. 24-bit is usually inaudible (TrueHD typically at 24-bits and EAC3 decodes to 16-bits). As for multichannel 5.1, back in the DVD days around Y2K, there were listening tests between lower-bit-rate AC3 (384-640kbps common back then) compared to DTS (1.5Mbps) and as far as I can recall (can't find an old link that works still), despite DTS' 2-4x increase in bit-rate, though generally consider superior, it wasn't a slam dunk and 640kbps AC3 was generally considered equivalent to DTS once level matched. The concept of "diminishing returns" does not only apply to hardware price-to-performance, but also to bit-rates of psychoacoustic perceptual audio codecs.

Fast forward 20 years to multichannel music streaming today. EAC3 is a more advanced version of AC3 released around 2004 said to be perceptually equivalent at lower bit-rates (Dolby at one time suggested 192-256kbps EAC3 is equivalent to 384kbps AC3 for basic 5.1), and the bit-rate with Apple "Spatial Audio" and other services is at 768kbps which is a reasonable number for high-quality music content. By the way, EAC3 can in fact scale beyond 5.1 (up to 15 full-band channels) and up to 6Mbps if ever needed even though currently the services are capped at 768kbps 5.1 with Atmos metadata (JOC).

I've always found it interesting what audiophile magazines choose to focus on when it comes to numbers and technical details. So, they think it's "bleak" that multichannel is streamed at 768kbps rather than Lindberg/2L's "... Dolby TrueHD bit-rates average around 6000kbps with peak data rates up to a maximum of 18,000kbps for high sampling rate multichannel content"? What's the big deal?

I think it's ironic that a more objective-focused audiophile like myself would have to remind subjective-leaning Stereophile to be careful about judging "good" or "bad" based on just numbers. Did they actually listen to the music or just want to obsess about numbers?!

As discussed over the years, big numbers like lossless 24-bit/192kHz were never that amazing when we listen and do A/B comparisons with lesser samplerates or bit-depths. I think it's great that 2L captures the sound at 24/352.8 and can archive at this data rate, translating this to high bit-rate TrueHD for their Blu-ray releases. However, this kind of hi-res specification might have been impressive 10 years ago when we were starting the hi-res journey, but not in 2024 since we know from experience now that hi-res is not audibly revolutionary even if subtly better compared to good CD at 16/44.1.

If we care about numbers having audible significance, why doesn't Jim Austin focus on some real hi-fi inadequacies like the fact that LPs only have at best ~70dB of dynamic range, and maybe -30dB crosstalk with a good phono cartridge. These and other fidelity issues with vinyl are even audible across YouTube lossy streams despite sometimes using obscenely high-priced gear as discussed last week. Surely, that's more of an inconvenient truth for many audiophiles than this minor EAC3 bit-rate nitpick! As you know, many of the products reviewed in Stereophile are for vinyl playback and thus susceptible to such levels of poor performance regardless of price. Where are the "As We See It" articles lamenting that many audiophiles are stuck with listening to such poor quality media in the 21st Century?!

I hate to say it since it's 2024 and I was hoping not to bring up the dreaded three-letter acronym, but feel I have to given the prominence of Morten Lindberg's quotes in the article. Did he notice what happened to the data rate from his precious 24/352.8 2L hi-res masters when he declared his love for MQA? It went down 8:1 with lossy MQA files, and that's only with 2 channels before further lossless compression.* Was the MQA file not also a "bleak shadow" of his hi-res master? Did he complain then?

[*With multichannel music, surround channels can be low level, sometimes there's no center channel content, and the LFE channel is bandwidth limited or even not used. Furthermore, multichannel albums have less dynamic range compression. As a result, we can often see better compression ratios with multichannel recordings than 2-channel. This is to be expected even with lossless compression.]

So, in recent audiophile history, Lindberg could have championed for lossless hi-res and instead sold out to MQA. He even presented a positive spin in the bankruptcy announcement. I'm afraid his name is linked to that money-losing, lossy scheme that nobody asked for, so forgive audiophiles like myself for questioning Lindberg's credibility when commenting about codecs! Nonetheless, congrats on the Grammy nominations and win. Awesome. For the record, I did say 2L recordings are some of the best for Realism recently so I think I'm still very fair to Mr. Lindberg.

BTW, I see 2L still charges MQA stereo files at full lossless hi-res prices; they might want to consider changing that after all these years.

Unlike the questionable FUD around bit-rate numbers, the jump from 2-channel to multichannel immersion, and the improved dynamic range of "Spatial" albums (including when downmixed to 2 channels) are definitely audible and likely beneficial for enjoyment assuming the music was produced well. Listen and decide for yourself.

One last thing from the article. Austin writes: "... this form of Atmos supports up to 128 virtual channels" as if to impress upon us how much content there is (oooohhhh, 100+ channels!) considering the bit-rate cap. The 128 virtual channels are actually objects plus bed content that can be imported and manipulated dynamically in the Atmos audio workstation. In the final Atmos render and encode, whether lossy EAC3 or lossless TrueHD, these objects are "clustered" down with detailed positional/level information defined as metadata. At most, even TrueHD can have "only" up to 16 lossless PCM channels which can be manipulated by metadata to represent the final rendered content +/- center with LFE, +/- surrounds, and +/- height channels. Ultimately, the authored Atmos file is "lossy" whether as EAC3 or TrueHD since neither contain absolutely 100% of the full-quality master as defined in the Dolby Atmos Master ADM file.

No doubt, there's plenty of fancy DSP happening in AV receivers and surround processors! Don't be afraid dear rational audiophiles, DSP is the present and the future of hi-fi audio.

In Summary...

IMO, there's nothing to be concerned about. Sure, the 768kbps EAC3+JOC Atmos stream will sound subtly different, but far from a "bleak shadow" of TrueHD-Atmos quality wise even if 10+:1 compressed down from the TrueHD or ADM file. 😂 I've done some Atmos editing and conversions over the past year and listened for myself already. Maybe at some point I'll put some demo material out for audiophiles to have a listen.

Absolutely, we can and should advocate for higher bit-rate TrueHD-Atmos versions of favorite albums for our collections. For the ultimate Atmos multichannel audiophiles, maybe they can even grab some truly lossless Atmos Master ADM files (300-800MB/track, ADMs cannot be streamed)! Perhaps in time the streaming services can transmit TrueHD-Atmos as well, but for practical reasons, I don't see that as a priority in 2024 just like 4K resolution isn't necessarily a high priority for many of us with Netflix accounts even though modern Internet speed can accommodate it (the 1080P tier is enough for me even with multiple 4K displays at home, but I will happily buy the 4K UHD Blu-rays for my favorite movies).

Whether music labels and companies like Super Deluxe Edition issue more Blu-ray multichannel/Atmos albums would be a matter of economics, driven by demand. Likewise hardware developments catering to multichannel can only grow if there's interest. I guess traditional magazines like Stereophile and TAS don't seem to be encouraging audiophiles to demand multichannel/Atmos based on what seems like tepid interest and IMO unfounded negativity towards EAC3 multichannel streaming. If they want to stay within their 2-channel niche including promotion of vinyl and other sorts of anachronistic hardware (including tubes, wire ruminations, unfounded tweaks), so be it. However, that might not be a good attitude to take for the sake of growing the high-fidelity hobby going forward especially if the traditional 2-channel "high-end" side of the hobby stagnates and the vinyl "revival" declines (IMO, this is inevitable).

Enjoy the music folks! I think it's important to be more insightful and realistic than the typical audiophile magazine editorials with their undisciplined perspectives based on cherry-picked hearsay of supposed audiophile heroes, expressed with unnecessary dramatics.

Addendum - October 2024:

Thought I'd just add a comment about listening tests and multichannel lossy codec. Way back in 2007, the EBU (TECH 3324) evaluated a number of these and we can see that Dolby Digital+ (EAC3) performed very well with 5.1 channels at only 448kbps already (significantly better than first generation Dolby Digital at 384kbps). Here's the score sheet:

Notice that every one of the samples was rated by listeners tested at 8 sites in Europe in the "EXCELLENT" range using the MUSHRA methodology which includes the original uncompressed reference audio for comparison.

These days, multichannel/Atmos is being streamed at 768kbps with a base layer of 5.1 channels plus Atmos data. One would anticipate very little issue with audible deterioration compared to the uncompressed signal!

Often, I will encode multichannel 5.1 (no Atmos of course) to AAC. Even at only 320kbps, the score doesn't look bad:

It doesn't perform as well on the "Applause" test sample which is an edge case of a highly complex signal with quick transients and reduced correlation between the channels. Already, even at 320kbps, the multichannel playback is averaging "Excellent" with the music samples.

These days when I encode 5.1 AAC, I recommend doing it at 512kbps using the excellent Fraunhofer FDK AAC encoder which will low-pass at 20kHz; retaining the full audible frequency range. dBPowerAmp is an easy way to get the encoding done.