 Give a very warm welcome to Christoph Engelmann and Anton Schlesinger. Yes, thank you. This is a bit intimidating, large space. Hello everybody. The topic of this talk is a new audio format called MQA, Master Quality Authenticated. And as the title of our talk suggests, it's more than just an audio format. And in order to explain how we arrived at the title, the clever stealth DRM Trojan, we have a lot of ground to cover and our talk is structured as you see here. So I'm going to talk a little bit about audio files and audio files. Then Anton here is going to examine some of the claims that are being made about Master Quality Authenticated. Then we will move on to 24 bits, got some room to spare to have a look where they actually hide their DRM. And finally we're going to make some remarks why we think MQA is actually quite clever. So wait. So it's good to be back here. Some of you may know me already as I've spoken at the conferences a couple of times before and mostly on policy related issues on identification, digital identification and authentication related issues. This time I have to make a confession. I am an audiophile and I believe that this cable sounds different than this cable. So I should be a millionaire. As some of you may know, the James Randi Educational Foundation had offered a million dollars to anybody who was able to identify the difference between two audio interconnects like these ones. And it controlled a big testing. Here is a 10 year old report on this challenge. Of course, nobody managed to find these differences and win the million dollar. And I'm pretty sure I would fail at the test too. But in my own experience, if I put one of these cables in the system for a week or two and then the other cable for another week, there will be some preference that emerges. And I cannot explain that, but I guess if you ask me to ABX a mattress, the subtle differences in texture and softness won't be noticed by me by just laying down on them. But if I get to sleep on each of these mattresses for a week or so, I might develop some sort of preference. I make this point here and I'm sure Anton, who is more on the objective side of things when it comes to audiophilia, will be skeptical about it. But what I want to illustrate is that we are talking about a market that operates and exploits the limits of perceptibility and where environmental and habitual factors play a role in what you sense and what you notice. And I believe that certain markets have developed around those perceptible limits and exploit something like long tail phenomena in the distribution of perceptibility among humans. In other words, in a saturated consumer electronics market, there are a certain class of devices which offer just noticeable differences and these can be commercially attractive and command premium prices. To illustrate this point further, let me ask you, who of you here is a gamer? Quite a few. Okay, so how many of you have a 240 Hertz monitor? Oh, two. Okay. How many of you have a high polling rate mouse? Quite a few more. Do you believe this makes a difference in your gaming experience? Okay, these are all objectivists here, skeptical people. Okay, values like 240 Hertz refresh rates are astonishingly high. And if you tell somebody 20 years ago that people are able to discern 60 Hertz refresh rates from 100 Hertz refresh rate or 240 Hertz, they'd have told you that the latency of the human visual system is far exceeded at this rate and everything about 60 Hertz is moot. Yet Apple now sells you 120 Hertz iPad and the difference is perceptible for many. It turns out that peripheral factors like head movement, haptic interaction with the screen and so forth, modify certain thresholds of visibility. In the VR context, this created big headaches and anybody who follows John Carmich's Twitter feed will know that he continuously is surprised that some of his subjects are very sensitive towards visual artifacts that normal users don't notice. So it's very easy to make fun about audio files into $2,000 Ethernet cables. This is one of these and notice the small error there. That's probably there because packets otherwise would get confused which way to go. So these kinds of gadgets and their many, many examples of those, of course, give ample ground to make fun of audio files. But that's not what we want to do here today. Instead, we want to look at master quality authenticated this new format as a development in the audiophile world that deals with these aforementioned limits of perceptibility and in a clever way exploits them to create and legitimize a new distribution channel for audio content. And we furthermore believe that these developments have possible repercussions outside of the audiophile world and might affect music consumers and producers and that is you. And labels and the way music is distributed in profound ways. So again, MQA stands for Master Quality Authenticated on Promises Studio Sound at Home. This is the MQA website or this is some page on the website. Join the revolution and recorded music. And there's a lot of, a lot to unpack in this statement master and quality and authenticated. To give you a little bit an idea why I'm talking about this, as I already said, I'm an audiophile, but I also sometimes spend time on the other side of music, namely in music production. I used to manage a metal band or co-manage a metal band called President Evil. We call it the Stiles Donut Trash. This is a great record. You just shouldn't look at the waveforms because it clips all the way through and has a DR of three for those people who know what that means. We toured with ministry and so forth. On the other side of things, I'm also interested in acoustic music. This is me at the Hansa Studio in Berlin, where among others David Bowie, Depeche Mode, and you too have recorded albums. And we recorded this album by Beatrix Becker, which is piano and clarinet music and cello music and a wonderful record. So coming back to MQA, you will understand that as an audiophile and as somebody who likes to be in studios, I'm somewhat familiar with music production distribution and their marketing claims of studio sound and take me into the studio and so forth, of course raise an eyebrow with me. And furthermore, somebody tells me or promises that they want to authenticate a digital file. It not only raises my eyebrows, my professional curiosity is peaked. I have given talks here about the history of authentication media and I publish on this issues on the questions of media of authentication. So all things authenticating of interest to me and MQA promises something like that. So what they say is it sound like what the engineers heard and they're going to authenticate that on your playback device. Everybody with basic knowledge of intellectual property enforcement and digital rights management by now should be a little bit skeptic about their claims. As we're going to show, this is rightly so, but the devil's in the details and MQA offers an interesting and clever design of how to bring digital rights management back into music and the music distribution context as some of you may know the music industry has tried to do that several times and always failed to implement and sell digital rights management to consumers. And most distribution channels these days and most files come without DRM. So this again is the website of MQA and you can see how they explain authentication and they have this interesting term provenance. And there's the important thing is here the blue and green light that will light that will go on on your device. If an MQA file is played on this device and authenticated and they promise sound like in the studio. So in order to understand what they mean by quality in the acronym, it's important to go back a little bit in history and give you a background about digital audio. So what is quality? In the 1980s digital audio was introduced that means for the introduced to the consumer. Here you see the CD as advertised by Sony. I think this ad is from 1984. The underlying format of the CD is Red Book Audio where you have a signal that's sampled at 16 bit worth of 44k, 44,000 times a second. And with 16 bit we have a dynamic range of 65,000 values to represent the difference between the most silent and loudest signal in the format. And with 44k we can represent frequencies up to 22 kilohertz. This is about 10% higher than what humans was very good and that usually means very young ears can hear. Most music today is recorded and distributed in this format. The Audio World Standardized in 44,1k, the Video Broadcasting World Standardized in 48k and those of you who have surround systems DVD or Blu-ray based will be familiar with the 48k sampling rate. So in the 1990s we see another step. We see music production becoming fully digital with digital audio workstations. The first 20 and 24 bit converters arrive in the studios. We see higher sampling rates. So this is one of these converters from the 90s. The little 24 indicates that it can do 24 bits. And there was also an attempt to sell these higher resolutions to the consumer in the late 90s with a little bit different format. This is the SACD introduced by Sony in 99. It's a different encoding scheme called GSD. It offers 120 decibels of dynamic range and in its initial form went up to 30 kilohertz, so way above human hearing range. But as many of you know, this failed in the market. And there's of course an interesting backstory to this too because the SACD was introduced in the very moment the patterns for the CD expired. And Sony tried to push a new format into the market and have a new pattern and licensing regime to generate revenue. So why did the SACD and the supposedly higher quality fail? Not only because people couldn't really hear it. On the turn of the 2000s we had MP3 and MP3 pretty much owned the market. And Steve Jobs put it best when he said convenience beats quality. And as you know, he turned around Apple basically with the iPod, which could hold in its initial form, I don't know, 5000 songs or something. And Steve Jobs knew what he was talking about. He was actually an audiophile with a system worth north of 50K in his home and supposedly only listened to vinyl. This together with Snapstar basically killed the music market and especially the market for high resolution audio as these formats are called that exceed the red book standard. Audio files to come back to our storyline here, of course fought holy wars against MP3 and any form of lossy compression. They deemed the format as inferior and argued that it destroys a generation's hearing and even inside brain damage. And I was actually surprised that Douglas Rushkoff made this argument too MP3 is bad for you. And he asked down there will the next generation of music listeners be able to hear reality the same way we did. So the idea is that we somehow lose contact to reality if we are exposed to lossy formats. But however, audiophiles to abandoned physical media and things like these standalone digital audio converters found the way in their systems or streaming amplifiers that either via USB or Ethernet allow you to stream music from local networks or from the Internet into your system. And these converters you can see here it offers DSD which is at the SACD format and 1624 bits and higher sampling rates allowed access to high resolution, higher bit rate files. So this is important because we come back to this in the context of MQA. And of course you can ask about the benefit of these kinds of devices and the audibility of resolutions higher than 16 bit and 44 1K. Again, I think there might be individuals that actually hear differences. Let's jump to the current decade, 2010s. I think we are actually in a golden age for audiophiles because music comes basically into the house more or less at very low prices via streaming. And furthermore, they can actually buy lossless files in resolutions up to 24 bit and 192 KHz in online stores like this one is the French company Kubos. And you can see that some of the contemporary current records are actually offered in these formats. So you can buy the new Burke record in high-res audio. I actually haven't looked at what resolution I guess 2496 or something like that. And Kubos will even stream at that resolution over the Internet. There is this little website where you can compare the different offerings. You just enter some band or some music and it will list you the availability of these files at different resolutions. So again, in essence, the 2010s are kind of a golden decade for audiophiles. They got ducks and streamers in their homes and could play the files at the same resolution as they recorded mixed and mastered in the studios in their homes. And it is this context basically that the master quality authenticated format was launched and promoted in 2014. There was a big launch event at the chart at this high-rise in London. Looks a little bit like this tower in Middle-earth where Zauron resides. And they claimed a revolutionary advance in audio and to deliver studio sound at home. But not only that, and that's where the story gets interesting. The inventors credit as Peter Craven and Bob Stewart. You already saw Stewart on the picture. Where is it? It comes back later. They backed their claims over the paper that was published in the Proceedings of the Audio Engineering Society. And they say that MQA builds on advances in neuroscience. And they also sometimes say advances in the sampling theory beyond Shannon Nyquist. Which are strong claims to make. So if you go to the MQA website, you will find this explanation in the FAQ. How does it work? And it says MQA is based entirely on science. Specifically it's based on new findings in neuroscience that have told us that the resolution of timing information is critical to our hearing and our ear-brain interaction. So if you look in this paper published in the AAS Proceedings, and look specifically at the neuroscience works cited there, I think it's seven or something like that. There's only one paper that actually deals with humans. The other papers, Test Hearing with Mongolian Gabriel and Barn Olds. And it's probably another mistake to believe that these animals have a very different auditory system than we have. And that their hearing exceeds our hearing when it comes to high frequencies. So building on these neuroscientific advances, they claim to correct for the time domain errors in the digital audio chain. Namely, errors introduced in analog digital converters. So when you record an analog signal in the studio and convert it to digital, and they say that these would create temporal blur. This is what they here say. There's a problem with digital. It's called blurring. It's also from the website. And this temporal blur is basically the main marketing mantra of MQA. And their quality argument, the argument that it sounds better, rests to a large part on this claim not only to bring the studio sound to the home, but by reducing what they call temporal blur. Their format, their format, so they claim contains mechanisms to cope and correct these temporal inefficiencies present in all digital systems where an analog digital or digital analog conversion takes place and remove that temporal blur. And reviewers in the audiophile press called these innovations revolutionary and claimed them not only to be audible, but to lift music to a whole new level. This is from audiophile online publication, The Absolute Sound. They say let the revolution begin. And here they more or less compare Bob Stewart with Einstein. If you read this here, this caption is Lord Kelvin up there. There's nothing new to be discovered in physics now. All that remains is more and more precise measurement Lord Kelvin five years before Einstein's paper on relativity. And if you look deeper in the text, and I'm sorry for this, I know you can read that now. But they basically bring forward Thomas Kuhn revolution, the structure of scientific revolutions and talk about a paradigm shift in digital audio and even come up with the Copernican models of planting plotemies, worldview and so forth, right? So big things are happening audio right now. You probably haven't heard yet, but we're entering a new paradigm. And we are currently and the fact that we are standing here is actually an indicator of the scientific crisis because we belong to the old world view and think probably not, right? So there is of course a big debate happening in the audiophile world around claims like this. And much of this debate is happening on a website called computer audiophile. And there is specifically this threat, which was opened on January 2 this year. And it has a mass 250k views by now. It's really interesting because there's a couple of really reputable people publishing on this forum in this threat. They have many outspoken critics. One of them is mastering engineer Brian Lucy, who has six Grammys and mastered among others Depeche Mode and Mary Manson and so forth. And he basically said that the available MQR files that you can buy now or stream via the title service were not created in mastering studios but encoded in an automatic way somewhere probably on the cloud. So the claim of MQA that you get the sound that the artist and the mastering engineer signed off on is definitely false for the MQA files available now. And we must presume that it remains false because there is no commercial converters available that include MQA encoding. So nobody running a studio can actually use this format right now. You have to send the files often in an encoding house where they get encoded and signed. So part of this controversy is a couple of people online who did deeper research. One of them is Akimago, a blogger, an anonymous blogger who actually set up an internet blind test MQR, if you're interested in that please look it up. It's actually pretty complicated what he does because since the encoders are not available, we cannot properly encode music in MQA. And in the past four years the company has not allowed anybody to basically send them files in order to create a proper comparison between MQA encoded files and files in PCM or any other format that you would have. So Akimago invented a pretty interesting way to work around this limitation. I can't go further into that if you want to take this test, look at this website. Another hacker that had a look at this is Mans Rohlgaard who is working on the SOX resampling package. And he has a GitLab site up where you find a filter coefficients of MQA and tools to actually look at the control stream of MQA. We will also list this later in the presentation once we put it up. So to wrap up my part so far, in 2017, 24-bit files as they're created in the studio, worked in the studio, are available for audio files. So they can actually listen to what's happening in the studio in a way. Of course there's other, you know, like the listening room and the playback change and so forth are other parameters affecting that, what they hear. But does that actually matter? Because most music, as all of you will know, gets consumed in lossy formats via Spotify or any other streaming service and listened on speakers like this. Mono Bluetooth speakers. So in this context now MQA promises lossless studio sound at home that is authenticated and they also promise a sound improvement via fixing the temporal blur supposedly happening in the AD and DA stages. So Anton, who actually works on these matters, will now investigate these technical claims a little further. Okay, there you go. Hi. So as Christoph introduced himself, I do the same. My background in digital audio. Christoph, it's changing slides by itself. Is there a way to press? All right. So my background in digital audio started in 1993 when I bought this CD player together with an amplifier and two HAKO speakers. And looking back at that time it was the only point in my life where I actually carefully listened to audio quality as a listener. I repeatedly listened to my favorite records, which were mostly dance music at that time again and again. The player did well and still makes the job today. Later on in 2000 I started studying at the TU Ilmenau close to Leipzig. And I heard the lectures of Karl-Heinz Brandenburg who led in the 90s group that developed the MP3 algorithm. And with that invention I later in the last decade I became a very convenient listener. I downloaded continuously and then in the current decade I'm streaming. So nevertheless, or of course, Onkyu wants to sell a system again to me and I'm inclined to do a purchase soon. But not the one that comes with MQA and why that is like that I'm going to show now. So apart from the DRM I do not appreciate the audio quality of MQA. And before we have a quick look at the audio technology, let me state the following. There's not much information out there from MQA on the algorithm itself. There are, Christoph mentioned it already, there are mainly two papers, convention papers of the AES. There are patents which of course hide the crucial parts and increase their freedom to operate so it's vague. But there's luckily a lot of information from engineers that looked into the details and did a reverse engineering. So MQA is an encoder decoder system nested in a container such as flag, free lossless audio coding, or Apple audio coding, or Apple lossless audio coding, ALEC, or the Microsoft Wave container. And thereby can be read and decoded by standard audio players in hardened software. Of course MQA wants to replace the entire audio chain, the entire audio chain of AD converters and DA converters. And wants to get royalties also from the streaming services. And the main approach is to reduce the file size such that you send a 24 or 22 bandwidth file over the internet. But nevertheless by unfolding is it, you can still have the high-resolution audio. So the MQA claims are there, it's lossless and it shall improve the audio quality. MQA is scalable, the more you pay, the more bandwidth you get. So from the sampling rate, 48 kilohertz, you can unfold it four times. And download prices are currently at 2L higher than for the original high-resolution audio. The 24-bit and sampled at 192 kilohertz files. Important to know, the end user does not own the file. This has implications. For example, if you do a loudness normalization on the file itself or room compensation, you lose the MQA license, so to say, and it's reduced to CD quality or below. So one of those claims in the AES paper, which is written at the bottom, is we advocate lossless compression, lossless processing and hierarchical up-down sampling. We highlight the quality and efficiency gains possible if the encoder and decoder are mutually aware and each match to their analog converters. So this is a bold claim and not based on evidence. So we look into that, lossless compression in audio means a redundancy removal and the possibility to reconstruct the original recording bit by bit, including the noise, whether the noise belongs to the music or not. Lossy compression removes information, be it audible or not. So it's MQA lossless. Shannon's entropy allows us to assess the maximum redundancy in an audio file that can be removed without information loss. So on a data set of music, Stefan Horto, a link is given to his paper, not the link, but the paper itself is given in the bottom of this slide, did a entropy optimization and flag compression and reached for a certain high resolution data rate, 50% of file size. So this is quite a lot and nevertheless MQA however has a data reduction of 57%. And it's obvious that this reduction, lowering the bit depths and changing the amplitude and phase response is a lossy process. This is also highlighted in the paper that I give here by Stefan Horto. So slide six, the next claim, MQA approach of recoding results in superior sound and significant lower data rate when compared to unstructured encoding and playback and has been enthusiastically supported in listening trials with a number of recording and mastering engineers, artists and producers. So this is a truly scientific wording that may give you an impression of the AS publications that we have as the resource. And we analyzed that. I quickly take you through the sampling process. On the left side you see two columns. In the top you see a drawing of the sampling scheme. 15 minutes, yes. So we start with a band limited analog time function and this is then multiplied in the time domain with a series of pulses. Then in the third line we have the multiplication. The right column shows you the frequency equivalent and the time discrete signal gets quantized with certain bit depths for high resolution audio. It's 24 bit per sample. And then if you put an analog low pass filter on the signal you can reconstruct the original analog signal bit by bit and at 24 bit practically equal to the original signal. What MQA changes is for example the B-spline kernel for a sparse sampling scheme. And this shall improve the temporal resolution. It's the highlight, Christoph already mentioned it. They highlight the detection of humans, the ability to detect transients with 10 microseconds and the kernel also shall match to the song. So a second point is they change the low pass filter for avoiding pre-ringing and for improving together with the kernel the time resolution, they call it de-blurring and compensation for signal distortions from the analog digital conversion during the recording process. So is it actually improving sound quality? The temporal resolution, I should start with a listening test that Christoph already mentioned. There was no preference towards MQA compared to high resolution audio. In fact in some situations standard high resolution audio was preferred. So in terms of temporal resolution is it improving audio sound quality? If you look at their claim it shall allow for the detection of events that are as short as 10 microseconds which would be a bandwidth of 100 megahertz. They call it de-blurring from a 48 kilohertz sample signal. This is questionable, so I don't think that's actually the case. The next point is, is it tuned to the song? At first they post to change the balance between time and frequency representation of their filters. Both are equivalent representations of a filter but if they use an adapter filter we don't know actually what is tuned to the song so there's information missing. In terms of the minimum phase filters to suppress pre-ringing minimum phase filters have their shortcomings. They change the frequency response, the amplitude and the phase response that has been also shown by Stefan Hotto. And in the last point that we want to make on that slide is the appetization in the time domain to avoid the ringing of the filters. The minimum phase filter has a ringing after the center of the filter and to improve also temporal resolution and to compensate for the signal change that is also a point Stefan Hotto looked at and he clearly shows that appetization, so windowing in fact is a low pass filter and not improving sound quality. So do we need MQA enhancement based on today's evidence from neuroscience? We heard we need a time resolution in that paper of six microseconds and it was referred to the binaural hearing and in fact we are able to discriminate interval time differences with just noticeable differences of 10 microseconds. Those differences are however based on a central auditory comparison of sound waves between the ears and because of the spatial nyquist limit only the fine structure of the frequency up to 1.3 kHz is analyzed and if we remember the binaural for example has of course ears that are closer together and that's why they can even detect differences that are smaller than in humans. So and from the media production side there's no need for an enhancement of audio from a scientific point as well as from an artistic point of view the artistic concept with all variables composition, performance, practice, etc. is captured on multi-channel files and the transmission chain has always been kept free from influences. Quote the music is the message and finally interpreted by the auditory system so we don't need a channel that is pre-processing what's going on in our head eventually. A general question, do we need HR audio, high-resolution audio based on today's evidence from Neuroscience for listening to music? And I have here a graph from Zwicker and Fassel on psychoacoustics so the audible threshold in quiet is this solid line and it shows clearly we are not able to listen beyond let's say 20 kilohertz. Moreover music has an amplitude spectrum with 1 over F so beyond 20 kilohertz the sound power is very faint and it's from my point of view I can hardly imagine that we are able to listen beyond 20 kilohertz. Nevertheless on the production side we use filters that perform interpolation and for that case it's good to have high-resolution audio. Another point is conservation of recordings so if we store music digitally then of course it's good to have a lot of samples such that we can interpolate once there are errors arising. So conclusion there is no evidence-based research that supports audio quality claims of MQA MQA is lossy because it reduces the bit depths and the frequency response we see a degradation of audio quality which in its current version might not always be audible the algorithm of MQA folding scheme is only vaguely described and it's disadvantages for high-resolution audio are obviously disguised by marketing slogans like deblurring. Once an MQA infrastructure is established the audio quality can be easily scaled down MQA's description of kernel optimization for each track song alludes to a perceptual coding scheme which might eventually end up at mp3 quality. One thing to remember there are three high-resolution audio codecs that compress audio losslessly like FLAC. Internet bandwidth is not a big concern in the western world anymore it increases hence there's no need to fallossy audio coding and most of the music as Christoph already told us is still stored in CD quality if we upsample that information signal-to-noise ratio decreases. Okay, thanks. Alright, so thanks Anton. Okay, I showed that. Good, so the question is where is the DRM? Basically Anton already showed that MQA employs a clever encoding scheme and as far as we know because this is proprietary technology and we have to work with what we got and that is reverse engineering by some people online and some of the papers and other publications that MQA makes. This is again from the paper that Anton just mentioned from Stefan Hotto where he drew up the MQA encoding scheme and the important part, I don't have a laser pointer here but the DRM sample up there which you can see is 24 bits and as you can furthermore see it's divided into two parts the bassband is the audio bits and the subband is the data bits so basically what MQA does is using a 24-bit sample throwing away 8 bits of that more or less 8 bits of that audio information and bury other stuff basically in the noise floor that a 24-bit sample allows for give me a second so if you look at the history I just gave you on digital audio this is in a way back to square one because we had in 81 with the CD Redbook standard 16 bits at 441 kHz what we now would get with MQA is in their own words something to the extent of 15 to 17 bits of dynamic bandwidth sampled at 48 kHz everything above 48 kHz is encoded in a lossy compression scheme as Anton just showed this is this folding what they talk about how this actually is accomplished in detail we don't know it's I guess it's a pretty clever way because you have to basically get spectrum from 24 kHz to 96 kHz into just a few bits and bury that in the 6, 7, 8 bits that you get with MQA compression so in a way MQA acknowledges that the standard that Sony and Philips arrived at 16 bits 441 kHz is good enough for human hearing and all the information above 16 bit and above 44 or 48 kHz is more or less useless not audible we can do something else with that so what do you do what do you put there and it's a fairly complex sophisticated cryptographic scheme oh wait I forgot this one this is the blue light on the on one of these MQA decks that lights up when the file is authenticated an authentication to remind you of that that's a loaded term in the digital realm it promises to verify the origin of the data that so who is the sender and the integrity of the data under the conditions of open networks where data can be altered at any point in the transmission and these requirements can only be met via cryptographic means as the continuous reports on hacked cryptos indicate MQA has to come up with a workable and sufficiently secure system to make this possible and MQA partnered with the German data security company UTIMACO to create an authentication component for their format UTIMACO is a major purveyor of cryptographic products and well established in the banking sector in Europe and also in the healthcare sector and this is the website for UTIMACO it's a powerful interception data retention services which is nice to have and they offer these hardware security modules which is actually what MQA uses UTIMACO also published a white paper on MQA it's available on the website that's very sparse on details but it suggests that MQA entails a public key infrastructure and that hardware security modules like these ones shown here will be deployed in at least the MQA encoding houses and purportedly later on in the mastering studios so the converters in those mastering studios would need to have these hardware security modules in them or some sort of hardware security modules derived from these devices and UTIMACO was even so nice or is it to provide hardware security modules for dummies book on their site which apparently audio journalists so far haven't been able to find and these hardware security modules basically provide the means to embed the necessary keys in the MQA encoded files and the exact means of the scheme management schemes and the handling of the root of trust is unknown at this point MQA might outsource this to UTIMACO or might use this crypto service that do the key management okay I have to come to an end in-house so I like to ask the audience here is this DRM I don't hear anybody I guess that means yes okay so I'm asking this because MQA claims of course it's otherwise they say it's not DRM and here is from Ars Technica MQA has no DRM competent application outside of the studio we have no opinion on this beyond nothing noting that DRM is a futile exercise right and this is like a large part of the narrative and the online debate because MQA and all the people involved including the audio press keeps tuning this is not DRM because you can copy the files but copy protection is just one form of DRM the goal of DRM if you just look it up on Wikipedia for example this is the first thing you see is to manage the rights of a user with data and this means both in the restrictive and the permissive use cases so MQA permits copying but it restricts access to parts of the file namely the higher resolution parts so in DRM parlance that's conditional access and the access is only granted if you have an MQA capable digital audio converter which you have paid for of course so I guess what we can say is that MQA is something like a premium model right you get the file that you can play on any kind of converter or CD player you can burn on CD it actually would be playable but if you want access to the premium part of that file you have to buy a device also where it allows that good so why do they do that one more point and then I'm coming to end the firmware as it has been reverse engineer so far chose that they actually have provisions in place for copy protection there are provisions there to encrypt the files and with the SARTSA 20 cipher and there's also bridge transposition schemes in there why do they do this Jimmy Iovine just said Apple Music boss that the streaming services have a bad situation there's no margins they don't make any money so he's talking about titles, Spotify and so forth whereas Amazon sells Prime, Apple sells telephones and iPads and so forth so the problem is that you have to sell something else outside of your streaming service in order to make money which of course Apple is in good place Amazon is currently the biggest speaker company in the world with these kinds of devices and Google is catching up so our interpretation of what MQA is trying to do is it's trying to generate a new licensing stream for the music industry or specifically for MQA these elements of the music production distribution and consumption value chain would need MQA licenses the recording software and plug ins the recording mastering studios with new converters the artists would get their back catalogs remastered and would have to agree on that streaming services would have to sign up with them physical media manufacturers and if it comes to that and of course all the hyphen manufacturers and end customers also are looking at paying MQA okay I'll skip all this why is it clever and why is it stealth I think the encoding schemes that MQA employs make very clever use of the space afforded by 24-bit samples namely to stealthily introduce a DIM in the music production distribution and consumption value chain and they don't openly talk about it and they're actually very very difficult to find these elements and if you ask what to do Tuvat in this context of the conference here it would be worthwhile if people start to dig deeper on what MQA is actually doing okay I just stop here thank you so thank you very much for the talk we have now like five minutes for questioning and answering so please line up to the microphones and do we have a question from the internet no so microphone number one hello thanks for your talk you talked about the reverse engineering of the firmware so it somehow sounds that they wouldn't have these complicated hardware security modules in the DAX well it's a specific firmware in one product that has been reverse engineered from Blue Sound I don't know the security provisions of that but these hardware security modules won't be at this point part of the consumer products they will be part of the converters in the encoding house and the distribution context how the exact crypto works between the MQA files and the consumer DAX is not yet known there's some hints and indications you will find in these resources of how that works but the hardware security models would be employed in the production distribution part of the of the scheme okay thanks we have another question for microphone number four yeah thank you for your talk I have two questions one of them is do you think that this marketing concept will have any success at all and the second one is do you think that music encoding does evolve anywhere else than multi channel encoding instead of increasing sample rate and sample bits yeah thank you good questions I think this this is a good example of how not to market something because I cannot think of any product that has managed to create more doubt with both the consumers of this product and also hardware manufacturers and all kinds of other people who are involved in digital audio than MQA I think they went out with very bold claims tried to play the audio files with their easy easy suggestibility and were surprised by the competence in the audience so I don't think this will go anywhere in these terms what was your second question again about how the encoding of music evolves if we are going for more channels like 5 channels 8 channels maybe 16 channels at some point or do some people still try to increase the bit rate and frequency so it's probably a question for Anton I can only as an audio file answer this in the audio file market multi channel audio is dead it's stereo I mean for the consumers this is why I showed this bluetooth speaker everybody is happy with mono now and that doesn't seem to be a problem stereo files will have audio files will have stereo setups and multi channel setups create problems with rooms and calibration all kinds of things it might be different in the AV sector in more general terms I think audio is a solved problem 24 bits and 192 kHz should be enough the frontier now is room correction and face correction with speakers so that's where the real problems are and where there is actually real progress digital what do you call weichen crossovers yes digital crossovers these kinds of things thank you we can add micrometric coding is used in multi channel audio so you send one mono signal and then you can reproduce by parameters a room for a multi channel setup that's often used thanks I think we have time for another question do we another question for number one I went to three conferences on digital audio and computer music this year I didn't hear anybody mention this I think the story about the authenticity authentication what do you call it can you step a bit forward it's fabulous I wonder if do they make any claims that do like virtual analog models of the original converters this story is completely like crazy that actually is being mentioned in the MQA marketing model that they call it fingerprint the original analog digital converter and compensate for the deficiencies and errors this converter makes and that's one of the big claims that there is a device specific compensation embedded in the MQA file being called upon upon replay and hence a better quality experience for the listener but all the evidence so far shows that this is absolutely not true what we have is a pair song encoding just like you would have an mp3 or something else the encoder apparently analyzes the file the song or the track it makes certain provisions on how to fold the signal so how many bits get allocated on the dynamic range on a pair track basis so there is no authentication or fingerprinting or any kind of modeling of the analog part or the ad part in the chain thank you very much for the questioning and thanks again very much for the talk