My Week of Audio: Part 1 – Dolby 96k Upsampling Explained

While last week wound up being a lot busier around here in The Bonus View than I initially expected (thanks to our staff of regular bloggers and special guests), I was absent for much of the week to attend not one, but three audio-themed conferences. As someone who has always considered himself a videophile much more so than an audiophile, I found this very instructive and educational. My first stop was a two-day visit to Dolby Labs in San Francisco, where I attended a tour and experienced some very interesting product demos from the company. Our site’s Mike Palmer was also present and will write up more coverage of the event for the HDD front page. What I wanted to do here is to focus on the technical side of the new products that Dolby unveiled, and explain what they are and why you should care. The first is a new process that the company calls “Advanced 96k Upsampling.” Sounds intimidating, doesn’t it?

As I’m sure that most of our readers know, most Blu-rays today are encoded with movie soundtracks in either the Dolby TrueHD or the competing DTS-HD Master Audio formats. Both of these are lossless codecs that deliver compressed audio files which (once decoded) are equal in quality to the original studio PCM masters. Thus, since both are equal in quality to the studio masters, it stands to reason that they’re also equal in quality to each other. Fanboy favoritism notwithstanding (and how absurd is it for someone to be a fanboy for a codec? But I digress…), that is indeed the case. “Lossless is lossless is lossless” and “Lossless means no loss” are the mantras that we home theater writers have tried to impress upon our readers.

Nonetheless, as you would expect from any competitive businesses, both of these companies (Dolby and DTS) have worked to differentiate their products from each other, both in the eyes of consumers and on the authoring end that the average Blu-ray viewer will never see.

Having lost a lot of market share on Blu-ray for reasons that I don’t have the time or space to go into here, the folks at Dolby asked themselves what they could do to make Dolby TrueHD better. This of course begs the question: If TrueHD is lossless, and lossless is equal in quality to the studio master, how could a TrueHD lossless track possibly be any better than it already is? The answer, in a quite logical sense, is to make the studio master better.

So, how do you do that? Dolby believes that Advanced 96k Upsampling is the answer. If you’re asking yourself what that means, you’re not alone. Let me try to explain to the best of my ability. If I get some of the details wrong, I hope that our more technically-savvy readers can correct me.

Unlike analog audio, which records a smooth and continuous waveform in real time, digital audio must capture audio samples in a series of discrete steps numerous times per second. The more often you can sample, the smoother and more faithful the recording will be to the original analog sound. (Remember, sounds in the real world that our ears hear are analog.) A standard music CD has a sampling rate of 44.1 kHz, whereas high-res audio can go up to 96k or 192k. (Most tests have concluded that 192k is overkill without much practical benefit.) Aside from the occasional concert movie, almost all feature films and television shows are mixed at a rate of 48 kHz. This is unlikely to change in the foreseeable future due to a variety of logistical issues in the film production pipeline. For example, 96k files require twice as many mixing resources as 48k, which means that only half the number of channels on the console would be available. Not to mention that the average movie soundtrack is comprised of numerous audio elements recorded at a variety of quality levels. 48k is a common standard that works and doesn’t seem ready to change.

However, once a soundtrack is mastered at 48k, it can be upsampled to a higher rate, much like standard-definition video can be upconverted to high-def resolution. Also like this video analogy, just as upconverted SD isn’t quite as good as true high definition, neither will upsampled audio be quite as good as recordings natively captured at the higher rate. Nevertheless, upsampled audio could in theory be better than its original 48k capture rate – emphasis on the words “in theory.”

Unfortunately, theory doesn’t always match practical reality. In this case, most A/V receivers and even pro gear designed to upsample 48k soundtracks to a higher rate have suffered interference from an artifact called “pre-ringing” that’s introduced in the audio’s Analog-to-Digital conversion step. Upsampling the soundtrack winds up emphasizing and exaggerating this pre-ringing, thus adding a layer of audible and distracting noise to the audio. At least, it may be audible and distracting to picky audiophile listeners. Whether the average movie-watcher will notice is up for debate. In any case, until now, upsampling has been problematic, and attempts to employ filters that would reduce or eliminate this pre-ringing have negatively affected parts of the audio spectrum that you wouldn’t otherwise want to be filtered.

Dolby believes that it has solved this problem via a newly advanced “apodizing filter” that will mask the pre-ringing artifact during the upsampling process without (so they say) negatively affecting the rest of the audio. (The apodizing filter only works in conjunction with upsampling. It can’t perform the necessary math at only 48 kHz.) By removing the pre-ringing, Dolby’s Advanced 96k Upsampling should not only be superior to other upsampling techniques, it should be superior to the original 48k master.

This apodizing filter is based on and licensed from technology developed by Meridian Audio in the UK, the manufacturer of many acclaimed pieces of high-end audio gear. Until now, the process was only available during decoding and playback. As such, it was only available to owners of Meridian’s (typically expensive) products. What Dolby has done is to move the apodizing filter and upsampling to the start of the signal chain, during the audio mastering stage, which should make the upsampled 96k quality available to any consumer at little to no additional expense. It will be up to the content providers to decide whether they want to charge extra for upsampled soundtracks, but Dolby has integrated the technology into the latest version of its Media Producer encoder software, which studios can easily upgrade for only a minimal fee.

After performing this upsampling (which is as simple as a one-step button push), the new 96k master can be losslessly compressed with TrueHD and authored onto any standard Blu-ray disc. The soundtrack will play in any existing Blu-ray player piped through any existing A/V receiver, and is fully backwards compatible even with hardware that doesn’t support 96k playback. In fact, Dolby claims that downsampling the 96k master back to 48k will still benefit from the elimination of the pre-ringing artifact.

The first Blu-ray titles to be authored with this new technique include the concert films ‘Joe Satriani: Satchurated – Live in Montreal‘ (already available, and it’s even in 3D) and ‘San Francisco Symphony at 100‘ (street date June 12th, 2012).

All right, then. Now you know the concept behind it. How’s the execution? Dolby sat a group of us home theater writers down in a demo room to listen to a series of test clips, both at 96k and the original 48k back-to-back. These included part of a Joe Satriani song, as well as scenes from ‘The Dark Knight’ and ‘Kung-Fu Panda’. Dolby also provided a hand-out with a list of suggestions for things to listen for in the content. I’ll quote directly from that hand-out here:

  • Clarity & naturalness to sound
  • Longer “ring out” to reverb and ambience
  • Consistent audible quality as high frequencies decay
  • Better definition between instrumentation
  • More natural-less granular quality to some voices

One of the Dolby reps described the process as removing “mid-range mush” and making the audio less fatiguing.

So, did I personally hear any of this? I’ll be honest here; the difference is extremely subtle. I couldn’t tell a damn bit of difference on some of the clips. Even on those where there was a clearly audible change, I’m hard-pressed to say that one was truly better than another, as opposed to just being different. The fact of the matter is that both the 48k and 96k versions of all of the clips sounded really good. (The next day, Dolby held a screening of the ‘Satchurated’ film where we listened to the theatrical audio mix at 48 kHz, and it sounded great in all respects.)

Some of the others in the room (especially those who write for audiophile publications) seemed to be more wowed by this than I was. Personally, I remain a bit skeptical about how easily these results could be swayed by placebo effect. For each clip we listened to, a video screen indicated the audio sampling rate. Would we have been able to reliably identify the “better” version in a blind listening test? I’m not sure that I would, and I’m not sure that many home theater viewers would either.

Further, I would suggest that the elimination of the pre-ringing artifact is less distinct than the noise floor inherently present in the average home theater room due to air conditioning, or street traffic outside the window, or a variety of other factors. The sound of a cooling fan in a projector or PS3 is much more overt than the pre-ringing in a 48 kHz recording. When we sit to watch a movie, our brains are smart enough to tune out this noise and focus on the rest of the soundtrack, and I think that’s the case with pre-ringing as well.

With that said, none of the 96k upsampled clips we listened to sounded worse than the original 48k versions. I didn’t find there to be anything detrimental to this process that would suggest that Blu-ray buyers should avoid titles that have been upsampled. However, I suspect that 96k upsampling will be a very hard sell for Dolby, other than to dedicated audiophiles or the contingent of Blu-ray spec-hounds who will seek out these discs simply because 96 is a higher number than 48. (We all know they’re out there.)

16 comments

  1. Very interesting read. It remains to be seen what the opinion of dedicated audiophiles on these boards is. As far as I’m concerned, I too hear the noise of the projector’s cooling fan, but it doesn’t really bother me. Heck, I’m even a fan of MiniDisc, which is a sin to most purists.

      • EM

        Speak, Fido, speak!

        I sometimes wonder about the sensory resolution of A/V technologies vis-à-vis other animals. What do they perceive? and how realistic is it to them?

        Even when considering only humans, there is value in playback of information human senses cannot detect. At least, that’s what years of Star Trek have taught me; how many times have Starfleet personnel relied on recorded details beyond their capacity to perceive with the naked eye, ear, or other sensory organ? For most consumers’ needs, that level of recording is not necessary. And of course, that’s not what this upsampling technique provides anyway—it’s more analogous to those miraculous computer-generated “enhancements” on modern police procedurals.

        Even if the upsampling enhancement is better, is it desirable? Personally, I’m rather wary of a future where I can’t tell whether it’s live or it’s Memorex.

  2. JM

    Since there’s no new equipment to buy, it sounds like we’re getting higher quality blu-rays for free.

    Will 96k upsampling improve Vudu?

  3. William Henley

    I am skeptical myself. I am sure everyone here can agree that 96k audio is superior to 48k, however, I am skeptical about upconverting. From what I have deduced from the two articleson this site so far, it sounds like its just some fancy noise filter. I would actually like to play with the filter myself, maybe run some DVDs through it or something – I mean, take some lossey audio and run it through there and see if it improves the audio any.

    As to the fanboy statement, I can admit I used to be a DTS fanboy, and still am to a small degree. While my main reciever is now lossless, the one in the bedroom is my old HTiB, and its not lossless. I am sure I am going to butcher the technical aspects of this, but DTS-MA has a “core” which I believe is encoded at 640kbps, and can be transmitted over Toslink, whereas Dolby reverts back to a seperate Dolby Digital track, usually encoded at 448kbps or lower. (Josh, I know you know about this, feel free to correct me if I am misquoting this). So, on older recievers, DTS-MA produces higher qualtiy sound than TrueHD does.

    There are other codecs that I have been at one time, or am currently, a fanboy of. Back in the day, I was a fanboy of Indeo 5 (vastly superior to Quicktime and Cinepeak, its rivals). Then Divx came out and put everything else to shame, followed by xvid (I remained a fanboy of Divx, mainly because the tools for Divx were easeir to use than xvid, and the technologies practically achieved the same thing and worked in the same way). Now I am a fan of AVC over VC-1 – mainly from a production standpoint – practically everything supports AVC, and a lot of times, I will actually export my stuff as AVCHD as opposed to Blu-Ray (not sure if VC-1 is compatable with AVCHD – I would assume not). So, for those reasons, I think its perfectly okay to be a fanboy of a codec.

  4. The whole premise of this is flawed and purely market driven. First, 48 KHz is absolutely adequate for fully capturing the entire audible frequency range. Second, there is no information “lost between the samples” as many believe. Even if there were, upsampling afterward cannot possibly restore lost content. Further, if there really was an audible difference after the conversion, and not just perception and placebo effect, it could only be a degradation.

    Dolby is usually pretty good on the science, so it’s disappointing that they feel they need to go this route to differentiate themselves.

    • Josh Zyber
      Author

      To be fair, Dolby has never claimed that the upsampling adds any information to the audio. What they claim is that the Meridian apodizing filter removes (or masks) the pre-ringing artifact from the analog-to-digital stage, which is something that audiophiles have complained about for years. Previous pre-ringing filters have negatively affected other parts of the audible spectrum, but this apodizing filter (allegedly) doesn’t.

      In theory, removal of the pre-ringing allows for better clarity and audibility of other sounds in the track. However, as I described in the article, I found the difference to be so incredibly subtle that I believe most listeners will not find much, if any, practical benefit in it. The inherent noise floor in most home theater rooms is more audible than this pre-ringing. If anyone watches Blu-rays on a PS3, for example, the sound of the PS3’s fan will more than negate any advantage of pre-ringing removal. That fan is an order of magnitude more audible than the pre-ringing in any movie soundtrack.

      The audiophile audience with dedicated noise-controlled listening spaces may disagree. This process is really meant for them.

      [Edit: So, I wrote this not realizing whom exactly I was responding to. I expect that I have exposed the limits of my audio knowledge. The best I can do here is try to simplify and clarify for our readers the explanation that Dolby gave me, and offer my own reaction to the results.]

  5. Pedram

    If they’re not going to charge more for it, I say why not? Go ahead and upsample Blu Rays to 96 khz. At least it wouldn’t hurt, and might even be better.

    Here’s my logic in thinking that it *could* sound better, which may be totally flawed since it applies more to video. When I had a 480p projector, showing 720p content on it looked better than regular DVDs. Since an upsampled DVD would look better than a regular DVD (not quite 720p, but somewhere in between), it would be logical to assume that when scaled back down to 480p it would look a little better than a regular DVD.

    So by the same logic, I would think that something upsampled (which would make a good “guess” as to what would be filled in) then downsampled back could look/sound at least a little better.

Leave a Reply

Your email address will not be published. Required fields are marked *