MQA Tidal to launch MQA Hi-Res audio streaming in 2016

MQA is a lossy compression methodology that compresses higher resolution data in a “smarter” way by encoding the higher frequency info in lower frequency least significant bits - they call it “folding” the spectrum. They have said “no information gets lost” but obviously some does - not that I have a problem with that necessarily.

I went to an MQA demo in NYC. The tracks they played sounded markedly better in their MQA version. The telltale for me was a 24/192 source file, which was played as-is with the MQA-ed version played after. The latter sounded much better. This clearly indicates the MQA process is not only NOT lossless, but it is doing some form of euphonic processing on the data. I have no problem with this - just don’t tell me it is lossless!

For a digital data perspective I agree it is not technically lossless, however from an audio point of view it is lossless as none of the ‘analogue’ audio quality (‘instrument to ear’ one might say) is compromised by the compression. Plus the non-perfect characteristic of the ADCs / recording device can also be compensated for.

It is as your say a “smarter” way.

All along I thought MQA is a lossless compression for hi-res contents, now it doesn’t really looks good. Originally we are moving away from MP3 to FLAC and this guarantee that no information is lost, that’s the whole idea why I sign up for Tidal Hi-Fi. If MQA is not a lossless format for compressing hi-res contents that I’m a bit worry. What will happen if MQA is encoded across the board to non hi-res contents if devices does not support it? What will be the effect of the sound quality? What will happen to MQA encoded hi-res contents; anything above 20kHz is considered non lossless or the whole audio frequencies response? The whole idea is to get bit to bit perfect and non lossless format just can’t do it. These are questions we need to ask ourselves.

This article from the absolute sound which is the most detailed description I have found of MQA says that it is lossless.
Beyond High-Resolution
An Inside Look at Meridian’s Master Quality Authenticated (MQA)

In addition I understand that the file contains timing information that an mqa capable DAC can use to ensure timing is correct with the original ADC that made the recording and so this further improves fidelity to the original analogue.

At least this is my understanding of the claims for MQA.

The nomenclature has become confused by reviewers and journalists referring to MQA “encoding lossless hi res files” or words to that effect. Sure, the original file might be lossless, but if the encoding process includes noise-shaping and dithering then the output will not be bit-perfect or lossless.

Does it matter ? Quite possibly not. If all that is being lost is the equivalent of “blank space”.

Hi-fi supposedly means high fidelity, which I interpret as the quixotic attempt to make the sound waves impinging on the listeners ear as faithful as possible to the sound waves hitting the recording microphone (leaving aside music created after the microphone for the moment). If at some point in it’s journey that signal is digitised then it is probably desirable for it to be transmitted in a bit-perfect way. But if an encoding process results in alterations to bits in non-significant parts of the digitised signal and the result is a closer match between signal and output then fidelity could be increased, even if the processing is not lossless or bitperfect.

I cannot say whether MQA does perform in that way. I’m just noting that lossless may not necessarily mean lower fidelity in this context. I think that is what the Absolute Sound article is trying to say.

Calling the encoded MQA bit perfect in relation to the original input file, however, does appear to me to be incorrect. The encoded MQA file will not match the checksum in the FLAC.

If MQA is indeed not lossless or bit prefect with the original input file then it will definitely raise a lot of doubt on this format. Lossless is considered mathematically correct; no data is being thrown once encoding and decoding are done. While bit to bit perfect ensure data is not modified when it get converted in a DAC. In these two instances, FLAC and ASIO for example ensure lossless streaming while ASIO take care of bit to bit perfect to the DAC. If the format in question is not lossless then the effort of doing bit to bit perfect is utterly useless. We need to get the whole process right in the start before we can talk about fine tuning the ADCs and DACs and other stuffs like timings and jitter. In the days when high bit rate MP3 is thought to be subjectively indistinguishable from CD, they are wrong! We should never let our subjective manner to rule because nobody hear the same.

I don’t agree that because some processing is done which results in a file not being bit perfect that using methods that would otherwise ensure bit perfect delivery are utterly useless. Such methods will continue to not affect the music in the way they always didn’t. What does become harder (possibly impossible) is our capacity to verify that those methods are indeed bit-perfect.

After reading the article on the Absolute Sound on MQA I begin to wonder if it shared some similarities on HDCD (High Definition Compatible Digital) or Sony SBM (Super Bit Mapping) methods by using clever noise shaping techniques and some custom designed digital filters to achieve better overall response in the audio range. If this is the case I can accept that MQA is indeed lossless in the audio range and anything above 20kHz can be theoretically re-construct back if high rez content is being playback.

I want to set the record straight, because the misunderstanding and misinformation surrounding MQA is high.

MQA is only lossy in the sense that it doesn’t encode the entire rectangular Shannon space… because it doesn’t need to. Most of that Shannon space (e.g. 192k x 24-bits) is literally a complete waste of space and there is no information whatsoever in this discarded space which is relevant to human hearing.

MQA samples the audio three times, once for each “mode” of human hearing (environmental, animal noises, speech). This is revolutionary.

MQA makes use of non-linear sampling techniques for efficiency without compromising audible audio quality in the slightest. This is revolutionary.

MQA goes beyond digital to characterize flaws in the ADC itself, allowing for correction in a suitable (MQA-compliant) DAC. This is revolutionary.

MQA achieves an end-to-end smearing in the time domain of less than 10 microseconds, which it is now realized is critical for faithful reproduction of audio. This is an order of magnitude (10x) less than achieved by a simple, but state-of-the-art, studio system consisting of microphone, ADC, and DAC. This is revolutionary.

MQA was recently described as the greatest single advance in digital audio. Not only do I agree, but I’m basing all my future music plans around it.

I don’t work for Meridian, but am a fan and Meridian users forum moderator. I’ve heard MQA on three separate occasions and after blowing me away on the first, continued to leave me with a massive grin on my face after occasions two and three. It is the real deal, and Bob Stuart’s legacy to the world.

7 Likes

@VirusKiller, your first statement really sound lossy to me. When you said ‘Most of that Shannon space (e.g. 192k x 24-bits) is literally a complete waste of space and there is no information whatsoever in this discarded space which is relevant to human hearing.’ What justification has been made that there’s no information that can be perceive to the human hearing? This beginning to sound like the day when MP3 trying to discard information which is deemed not able to perceive by human hearing. Moreover it sounds more subjective rather than actual. Sorry Meridian if MQA turn out too lossy than you don’t have my vote on this.

Hi Guy,

Have a look at the Absolute Sound article again. It explains exactly why MQA is not like the old MP3 argument.

What MQA appear to be doing is using space under the noise floor to stow information. Instead of having lots of 0’s under the noise floor, they say to the decoder, assume it’s 0’s all the way down (which it is) and then decode what you actually find there (the folded back info) in this way …

There is no question regarding the bennefits of MQA.

The one thing still not crystal clear to me…

FACTS
MQA decodes into PCM. The decoding can be done in an MQA-decoder-fitted DAC or in the players software (Tidal, Roon or others). If software decoding is used, the sound out of a non-MQA-fitted DAC (presently 99,9 percent of DAC users) will be better than what we are used to right now (including 24bit 192).

Q: If the decoding is done via software and the PCM-signal is sent into a non-MQA-fitted DAC, will the sound quality out of a pair of speakers be similar, better or worse than if an MQA-fitted DAC did the decoding?

The question all present users of high quality DACs would like to know is, will I have to purchase a new MQA-DAC to take full advantage of MQA?

@MusicEar Guy, think of it this way: in the frequency domain there is a very well established curve which describes the human hearing system’s minimum audible threshold. What happens under that threshold doesn’t really matter. On the upper side, there are de facto real-world limits on the max volume of individual frequencies which are actually present in recordings (even though, theoretically, you could record certain frequencies at volumes which never occur in speech, music, or even special effects). What matters is a very small wedge in the rectangular Shannon space. What also matters is time domain resolution, which is why we will see MQA masterings up to 768kHz.

Within the “wedge of audibility”, it’s not lossy in the terms that you are familiar with. Arguably, it’s less lossy than conventional linear PCM, because [as far as I can tell] there will be more / more granular sampling where it really matters. Think “24 bits applied to a smaller space”, but also think of concepts like logarithmic sampling (which blow conventional thinking out of the water).

We live in a very cynical world and I understand the skepticism, but I for one believe that Bob Stuart cares more about the fidelity of the recording chain than most, and has the engineering brilliance to do something about it. Meridian has a history of great technical achievements and almost going bust through, dare I say it, more of an interest in the engineering and music than the selling; these guys are passionate audiophiles and music lovers in every sense of those words. The world of digital is about to be turned upside-down and there are many who stand to lose out - a certain well-followed blogger and purveyor of Hi-Res recordings springs to mind… What Bob and Meridian have done is to go beyond digital and to create a new and far superior recording chain paradigm; they are as many decades ahead of the game as the existing paradigm is decades old.

But don’t take my word for it. There is a reason that the head of Warner Music (Mike Jbara) was at the launch of MQA at The Shard in London last December: Meridian has been working on MQA for over 5 years now and been demoing it to the leading recording engineers all over the world; these are the guys who work with state-of-the-art every day, and with the best ears in the industry; and they are convinced: “how did you do that?!” is apparently not an atypical response on hearing MQA for the first time.

2 Likes

The information stated by the people who speaks to Meridian tells us that a software decoded MQA PCM file will sound better than a non MQA file using a non MQA DAC.

But I would like to know if the software decoded file sounds as good as an MQA file being decoded in an MQA fitted DAC?

Can we keep a great DAC and go for the software decoding or will the people that really would like to benefit from MQA recordings have to purchase a new DAC? (Firmware?)

My understanding is that how good it sounds will be entirely implementation-dependent. There is nothing to say a priori that one will sound better than another. Having said that, dedicated hardware solutions may well have an advantage in terms of noise/jitter at the DAC.

After reading more on MQA, my conclusion is it is not a lossless format given the fact that the bit-rate is going to be same as CD for transmitting high resolution files by Tidal(assuming Tidal decided to keep the streaming rate the same). I would have to listen for myself between a 24-bit/192kHz which has a bit-rate of 9216kbps vs a MQA at ‘1411kbps’. At the moment the maths just doesn’t add up for me. Who know, if MQA does indeed going to sound ‘like’ the native 24/192 then it going to make me very happy indeed.

Regarding the “bitrate do not add up”

Well the biggest improvement of MQA will not come from bitrate but rather from having ten times more aucracy in timing at the A/D-mastering. If a sound travels at 340 m/s present cd mastering bundles all sound every 150 cm, resulting in pre-ringing (unnatural for the ear as it do not exist in the real world) and also post-ringing.

With MQA remasterings ten times better timing, all sounds being bundled every 15 cm, a gigantic improvement without adding higher PCM bitrate will be the result. That is one reason MQA will sound better than 24/192 at CD’s 16/44.1.

This means that a voice or guitar in one MQA-tone will not sound like
1 1 1 1 1 but rather like
1111111111111111111111111111111 (A lousy way to describe what will fell like more deapth, clarity and emotion).

Wrong compairing MQA with today’s high rez
Please note that an increase in sampling rate from 96 to 192 is only doubled, meaning a better pre- and post ringing (and noice jitter) of only 10 percent of what MQAs ten times better audio timing will deliver (if recording remastered with MQA). On top of that, the sampling rate in traditional high rez doubles on a source material which a ready have got the 150 cm audio bundling, meaning that the MQA is A LOT better even at lesser sampling rates.

There are more reasons why MQA will sound better than 24/192, the above mentioned worth noting extra.

1 Like

@Niclas_Mardfelt, I disagree until I’ve a an opportunity to listen to a AB comparison. likewise we shouldn’t jump into conclusion that MQA is indeed going to sound ‘better’ than 24/192 or even 24/384. I believe a good mastering and less processing (DSP) will sound closer to the original performance.

The only way MQA will take off is if its free to stream vs normal services and there is no hardware to buy and decode software is free…ie cost the user nothing…

MQA is going to have a lot of hurdles to overcome. MQA is lossy compression and a DRM scheme. That is why music companies are on board with it. If you don’t have the right equipment then it will play at a “lower” resolution.

Roon will decode it and send on a PCM signal but does it have the same sonic characteristics of a MQA DAC? From my conversation with the Meridian guys at the demo, they were pushing hard for people to buy into a hardware only solution.

The MQA at the demo decoded to a 24/192 PCM audio stream. Did it sound good. Sure it did. But it was played back through DSP8000 Special Edition Meridian speakers (~80k) powered by equally expensive amplification. Any well done high res PCM stream would sound good in that environment.

If you read the MQA supporters they mention the time signature. I understand what MQA does but to get that there needs to be a NEW mastering done from the original tapes. Which means the industry will be hapy to be able to re-sell you the same content. MQA resmastered “Kind of Blue” anyone?