Steve , yes the interest is remarkable - including my own especially considering I have never heard it However it seems the reaction from many who do hear it is like Joelās and I think this is more than just marketing hype - hopefully we will be able to judge for ourselves before too long.
Iām struggling to get my head around the technology too but I donāt think the idea of MQA is to replace the master but to develop something that is close to the master at much lower storage sizes and bandwidths that can be distributed to the customer more readily. This is the Abstract from Stuart and Cravenās Audio Engineering Society paper on this.
[quote]ABSTRACT
When recording, the ideal is to capture a performance so that the highest possible sound quality can be recovered from the archive. While an archive has no hard limit on the quantity of data assignable to that information, in distribution the data deliverable depends on application-specific factors such as storage, bandwidth or legacy compatibility. Recent interest in high-resolution digital audio has been accompanied by a trend to higher and higher sampling rates and bit depths, yet the sound quality improvements show diminishing returns and so fail to reconcile human auditory capability with the information capacity of the channel. By bringing together advances in sampling theory with recent findings in human auditory science, our approach aims to deliver extremely high sound quality through a hierarchical distribution chain where sample rate and bit depth can vary at each link but where the overall system is managed from end-to-end, including the converters.[/quote]
From what I can understand MQA is as much (or maybe even more) about getting the timing right than about resolution (whatever resolution really is!)
There are still many questions that remain to be answered like the ones from Mystic and others here and there is still a long road for MQA to go before it is a real option for the consumer. Stuart is pretty realistic about the challenges MQA faces. For instance here are a few quotes from his lengthy interview in the Absolute Sound with Robert Harley:- āyou realize that you are tilting at windmills and weāre dismantling sacred cowsāā¦ āwe knew we had a terrible uphill battle because of the established thinking.ā He seems to be getting plenty of traction at the moment and it will be even more interesting to see where this goes over the next few years and if its a real mainstream success or just another SACD or even worse!
On encoding personal music - it would not be surprising to have the ability to do that.
On TIDALās implementation - After reading what Bob Stuart has said, it seems to me that MQA encoding can either work with ādefaultā settings or can be fine tuned to the recording chain if the provenance of the chain is known (which I imagine is rare).
I would expect TIDAL to simply be using the compression component of MQA, ie would encode with default settings and stream that.
So are we reckoning this is likely to be looked back on as a key milestone in music media, like; vinyl, tape, optical, āearly digital filesā, MQA?
It does often amaze me that decades after CD, with all the advances in science and technology, with computing power that can literally simulate the dawn of time, people often still refer to CD as sounding better than a lot of current computer playback, and both of them considered a lot less realistic than vinyl or tape before it. Iāve never really had the luxury to compare such high-end systems but it seems common enough to hold some truth. I mean, Iām not really a classic audiophile (donāt even think I am one at all I just like a nice hifi and good music) but Iāve got a Ā£5 grand DAC/amplifier and decent speakers and itās not like I think the bands playing in my loungeā¦ Perhaps the foundations digital was built on were just incorrect or incomplete, and it needed a re-think?
So is the assumption with MQA that computer playback will be more realistic using modest sources, or will uber-expensive linear power supplied physics laboratory temperature controlled clocks and jitter eliminators still be needed? Was there a Ā£45k worth tower of a source behind the curtain feeding the Meridian explorer?
that is great news towards that end. Atlantic obviously has some huge records and artists and Warner is on of the biggest labels. I would be interested to see what Sony says. With their history of formats iām not especially hopeful. Personally i would love to see subpop and 4adās logo added to the partners list.
Here is an except on the hardware versus software decoding aspect:
Q: The decoder can be software thatās built into a DAC or other decoding chips across a wide variety of devices.
A: Exactly. The decoder can be soft or hard or a mixture. Weāre talking to chipmakers about building part of MQA decoding into the DAC. Making a DAC that understands how to render MQA natively is easier than making a DAC that understands how to render 48kHz natively. We imagine and hope that over time some of the pragmatic partsāwhich have to do with how to optimize the performance of any DACācan be done ever better.
One other thing i noticed in the interview and somewhat repeated in the AES paper linked earlier is this:
Q: The MQA file contains information about the A-to-D converter so that the decoder can correct some of the A-to-D converterās problems.
A: Thatās right. Part of the encoding process adds ancillary data such as date code and copyright owner.The encoder has information about the A-to-D, which tells the decoder how it was encapsulated so the decoder at the other end can use the best decapsulation formula; producing the shortest temporal blur possible for that content. It will vary according to the musical content.
From the Paper:
Conceptually, we are trying to connect the A/D and D/A modulators together with a signal that encapsulates the entire sound of the original but without artefacts that imply lack of resolution, and to package it for efficient distribution. The authors have used lossless buried-data signaling within the channel to carry instructions, metadata and authentication.
Does the phrase Copyright Owner and word Authentication sound like DRM to anyone else? I really hope not. I lived through that nightmare once and would be just as happy to keep buying cds off amazon or discogs rather than deal with it again.
[quote=āAgDev01, post:107, topic:5408ā]
Does the phrase Copyright Owner and word Authentication sound like DRM to anyone else?
[/quote]Hi,
DRM was considered very early on with MQA but it was dropped from the specification.
MQA Authentication has nothing to do with DRM, it is solely concerned with ensuring that whatās in the stream is exactly what the artist / label intended ā¦ Think of it as a seal of approval for the provenance of the music.
Weāve all heard the stories of Redbook files being upsampled and then sold as genuine HD, authentication is designed to prevent this unethical practice.
There is nothing I can read in Bobās responses which would mean that hardware decoding has features not available in a software decoder.
Encoding the ADC information into the MQA file definitely sounds like we will need to buy new versions of our recordings. Why would any record company not climb on board ?
Thanks Kevin, I had previously posted a link to this interview on the MQA āmusicischangingā website but it disappeared in their revamp so glad it is now elsewhere.
For point 1: 'MQA will NOT stream at ā16/44.1" bit rate, itās only similar to that bit rateā
What is actual bit rate in kbps output from the MQA encoder? I understand MQA can be further encoded in FLAC during streaming.
Point 3: If -168dB may refer to the internal working of MQA coding, I find it hard to correlate in term of the bit rate output from the MQA encoder and the noise floor is supposing to give? i.e. Low bit rate and very low noise floor?
I have read somewhere that the bit rate is around 1,5 Mbps, and as far as I know the MQA decoder will output a hi-res FLAC file 192/24.
Iām not sure if I understand your question. Iām no deeper expert either.
Assuming you are wondering how can MQA output 28 bit? Well, I believe it will not. The least significant bits are only used to store the additional information (of the frequency spectrum between 24 and 96 kHz) into the LSBs not used in the 20hz - 24khz spectrum, and the decoder will reverse the encoding process and technically just output a FLAC file at the end, again 192/24.
I think that @MusicEar is also on to something and that to encode things below the noise floor the MQA files are 24 bit. Bob Stuart states this in the Absolute Sound interview. [quote]The hierarchical download file that we end up with is typically 1x, which means itās 44 or 48kHz and 24 bit.[/quote]
This begs the question of how this works for CDs and it seems there is some additional processing needed to put the files on CDs at 16 bit. I also found this on the musicischanging.com site.
Not sure how this then maintains all the extra information so that MQA capable CD players can decode it and is still compatible with legacy players. I imagine there are some compromises in this otherwise they would be doing everything in 16 bit files and eliminating this additional step for CDs.
@philr, thanks for the details information, that really helps.
I was trying to understand why an increase of input information, i.e increase in sampling frequency to the MQA encoder will not increase the output bit rate; in this case it is capped at 1.5Mbps. How could such a vast input information can be contained in a limited 1.5Mbps bandwidth? Iāve two possible explanations:
According to @NOV point 2, [quote=āNOA, post:93, topic:5408ā]
MQA encoding process will evaluate how much of the lastest bits can be used for folding depending on the dynamic range the song/album needs itself. So it can be something between 16 to 20 bit for the content between 20hz-48khz.
[/quote]
Am I right to say thereās some psychoacoustics processing to eliminate redundant data before doing the fold back?
If not, then all the redundant information must be fold back completely into region āAā so that it makes sense that it can be ācontainedā into a limited 1.5Mbps bandwidth. However, what puzzle me is region A noise floor can be extended all the way to -168dB(In @NOA explanation, referred to internal processing of MQA), but eventually this information need to be output over on a limited 1.5Mbps bandwidth, how can this be achieved?
I believe that this is because the final MQA file is always 48 kHz (or 44.1 kHz) / 24 bit, no matter what the original bit rate of the recording was. The bits in the file which are located above the noise floor preserve all the audio signals below 24 kHz, while the audio which was present above that frequency gets sliced up into frequency segments and each is āfoldedā below the noise floor of the next. The high frequency data is then compressed and stored below the noise floor of the final 24 bit file, and the entire thing can then be losslessly compressed into a file which gives 1.5 Mbps.
Iām guessing that this is achieved by using the 24 bit depth and adding noise-shaping.
@MusicEar I think it is akin to 1 but not quite as simple as just eliminating the redundant data in HiRes files. It is worth noting though that in the AES paper Stuart and Craven point out that the audio information in a 192/24 file is equivalent to that of a stream having a peak data rate of only 960 kbp!
If I am understanding right MQA uses different processing ākernelsā (different combinations of sample rate, filters and maybe even bit depth) depending on the octave (ie. frequency) of the signal and the sample rate of the original source to capture all the musical information more much more efficiently (especially the very high frequency noise). By this process all the musical information is captured using much less data than 192,000x24. This reduced data amount of data in the ranges above 48 Hz is then captured below the noise floor in 44.1 or 48 kHz 24 bit files.
If the final MQA file(after decoded) is going to be 44.1/24 or 48/24 regardless of the original bit rate recording(176.4/24 or 192/24 or higher) then it begins to make sense that file can be compressed using a fixed bit rate of 1.5Mbps. A 24 bit depth with added noise shaping to extend the noise floor all the way to -168dB sound more like it.
that shows MQA file in āOriginal Resolutionā while the rest are quoted with their respective bit depth and sample. The original recording is mastered on a DXD 352.8/24. It is a bit strange why 2L quoted as āOriginal Resolutionā and not 352.8/24 or 44.1/24? If someone here have MQA playback device can confirm what is final output file resolution displayed on it?