Porting/building MQA decoding software?

I’ve no idea if this is practical or of interest, or even sensible. However thought I’d at least ask…

I’ve continued to investigate ‘MQA’ which you can expect to come to you in audio files and on CDs, like it or not. Companies like Warner and Tidal have adopted it, and are promoting its use.

Given this I’d like to do some more checks on it, but ideally need a ‘software’ MQA decoder to do this so I can run sample-by-sample analysis compared with test files put into an encoder.

One apparently exists, and is here:

https://code.videolan.org/mansr/mqa/

However it is way beyond my skills (sic) as a programmer to use this to make a working program – on Linux or RO. So it’s existence doesn’t help me much in getting a version for use in tests.

Question is: Anyone else here think they could help, and – ideally – make a working version for RO, or failing that, help me to get it done on Linux? Fancy a challenge? 8-]

Doesn’t matter if it is slower on RO as that would be more convenient as all my following analysis software and graphics plotting is done on RO.

But is this essentially impossible? Well, for me it is, but for someone who can program, maybe…

Oh great. That’s what the world needs – yet another lossy codec, this one perverted to be a DRM fetishists wet dream.

This is interesting: https://en.wikipedia.org/wiki/Master_Quality_Authenticated#Criticism

FWIW I’m not a fan of MQA, and have been particularly concerned by the likelyhood that we will get Audio CDs released that have been MQA encoded. I’ve already done a number of investigations, etc, recently
http://www.audiomisc.co.uk/MQA/investigated/MostlyQuiteHarmless.html

But a problem I face is exampled by that. I needed to use an MQA DAC and then grab the analog output for comparisons/analysis. That means some aspects are difficult to check. I need a sample-by-sample comparison between the encoder input and decoder output. (The files are from 2L who also provide DXD source files.) So I need a software decoder, not just an ADC and DAC.

Rick, are you in Japan? IIUC You can get MQA CDs there already fairly easily.

One apparently exists, and is here:

https://code.videolan.org/mansr/mqa/

Even if you could compile it, the file bluos_ssc.c dynamically loads libbluos_ssc.so so a no go on RISC OS and also on Linux unless you can get hold of libbluos_ssc.so.

I think I know how to contact the person who put up the code I URL’d. I’ll ask him about the libbluos_ssc.so

In case anyone wants to know some of the gory details, the recent exminations I’ve done of the effects of MQA on LPCM are shown here

http://www.audiomisc.co.uk/MQA/GoldenOne/ChallengeAndResponse.html
http://www.audiomisc.co.uk/MQA/questions/QuriousAssumptions.html
http://www.audiomisc.co.uk/MQA/OnImpulse/RingingInArrears.html

To go further I’d need a software decoder to give me sample-by-sample values I can compare on that basis with what gets shoved into MQA encoding. That would let me uncovered any distortions or losses in forensic detail. As it is I’m using a lab/studio quality ADC to capture the output from a DAC. Close, but not ideal.

If you listen to music, this matters, because MQA will come your way…

Rick, are you in Japan?

Uh… <looks out window> Nope.

You can get MQA CDs there already fairly easily.

That doesn’t surprise me. When it comes to tech, Japan is not so much another world as a completely alternate reality.

by the likelyhood that we will get Audio CDs released that have been MQA encoded

An audio CD that won’t play on standard equipment (and thus rippable by Streamripper) would go back to the shop as “unfit for purpose” (unless it clearly said “this crap won’t work on an ancient CD player”). I choose to interpret “compatible” in a strict sense.

It is rather interesting that morons are trying to introduce a DRM blighted audio codec, given the long and hard fought battle to get music in DRM-free formats. I think the likes of Amazon have a good idea – you can purchase MP3s and they are completely open, but the customer ID is stenographically encoded into the audio so you can be tracked if you pass it on (plus being in the audio rather than in the file, it ought to survive most transcoding to other formats).
What would be nice if if they offered the option of WAV (raw uncompressed), AAC, and FLAC for those who think that MP3 isn’t “good enough”.

But a lossy DRM codec? What’s the point except a blatant attempt to control what users can do after purchase?

If you listen to music, this matters, because MQA will come your way…

As will the consumer fightback. :-)

I’ve read some of the stuff that you have written. I won’t claim to understand much any of the technical bits, but my takeaways appear to be:

• The encoding process appears to splatter high frequency rubbish into the audio, which could have consequences even if not directly within the range of human hearing.

• “Compatible” with standard CDs? I feel like I ought to point at PALplus. You see, PALplus was a widescreen version of PAL (letterboxed) that encoded extra information into the black bars used for the letterboxing, so PALplus compatible televisions could show the widescreen content with a higher quality than a normal broadcast TV.
  Those who were sensitive, or had the brightness up, could see this information as random blue sparkles in the black bars.
  I don’t believe that it is possible to stuff additional information into an existing signal without either a lot of compromises or affecting the original signal (I’ll point you at colour analogue television again – this time the colour was added in such a way as to not affect the monochrome, which led to several decades of weird colour effects, especially with high frequency components; not to mention the lower resolution of the colour information).

• You probably won’t really have a better idea unless you can get hold of an MQA CD and a regular CD of the same thing; and while audio differences could be down to mastering or source material used, you might be able to spot distortion introduced by MQA?

• Tidal know their encoding is rubbish and won’t accept anything they deem as not being music, specifically things like regular sine waves that can be analysed to better understand the behaviour of the encoding. I’m sorry, but their protective attitude and claims of “master quality” all sound like snake oil. Or, as you put it, “trust me, I’m a doctor”. Doctor Shipman, perhaps?

There are a few companies in the industry not too pleased with MQA but it does seem like it is gaining traction for streaming services and now maybe CDs too. I have seen several reviews of MQA that from a sound quality point of view do say it works but it’s likely extremely marginal. Also worth pointing out the RISC OS link with MQA being a Meridian product and Alan Boothroyd being a co-founder and the designer of Meridians best products in the past.

Linn’s take https://www.linn.co.uk/uk/blog/mqa-is-bad-for-music-heres-why

Audio CDs with MQA-encoded content will ‘play’ on normal CD players and similarly can be ripped to files that also play. However any non-MQA DAC or player treats the content as plain LPCM. The result is some added garbage, particularly up around 20kHz. The nature and audibility of this are aspects I’ve already explored on the webpages I’ve written.

Frankly, most often it probably won’t matter to me as an individual as I’m old and haven’t been able to hear c 20kHz for years! But it may matter for other people. Also, MQA have ‘defined’ music in terms of an explicit assertion about having a ‘1/f spectrum’ (sic). I suspect some music will break that presumption and get released on CD, etc.

I’ve now heard from mansr wrt the bluos code. He has said:

“Yes, it loads libbluos_ssc.so which contains the MQA decoder. This library is built for ARM Linux, so that’s what you have to run it on. There’s a script (get-bluos-ssc) in the git repo that fetches the library since redistributing it would (likely) be a copyright violation.”

Alas, that’s problematic for me at present. All my Linux boxes are non-ARM. And the only ARM machine I use is my ARMX6 running RO. Unless it is a very safe doddle I’d not want to try installing linux on this ARMX6 just for this purpose. So something of a hurdle… :-/

Meridian have, indeed, had a history of producing some very good digital items. I used their DACs for some years and think they are good. But having looked at the theory and claims for MQA I fear it is simply a needless project based on some ideas and claims that are questionable in information theory terms. And that the aims could be dealt with more effectively by other means that are ‘open’ and better-founded on Information Theory.

Fortunately, I’ve also concluded that the MQA encoding may be ‘mostly harmless’ for LPCM replay. But by the same token, I don’t want to find I have to pay for it because the music I want is MQA encoded and thus priced into the cost of the CD or file.

BTW There is a certain irony in Linn complaining. Their CD players included HDCD decoding, and the Linn CDs routinely are HDCD.

This gives their players an advantage because when you play their CDs on a non-Linn player the peak compression of HDCD isn’t expanded. Happy day. :-)

‘defined’ music in terms of an explicit assertion about having a ‘1/f spectrum’ (sic)

What does that actually mean in simple terms? Because I can think of a massive difference in the various parts of Holst’s “Jupiter” ¹, a bit of Nightwish or Epica ², and then if we should throw in Penderecki’s Threnody (which must be interesting to play) ³… all are music.

may be ‘mostly harmless’ for LPCM replay

I hope so.

I don’t want to find I have to pay for it

Well, there’s the question. What is the point? Is it a DRM-happy attempt at providing higher quality audio than regular CDs?

Isn’t… well… getting away from the gold-plated plugs crowd, isn’t 16 bit 44.1kHz more than adequate for domestic use? The sampling rate allows capture of more frequencies than the average person can hear (especially as they age), and the signed 16 bit supposedly giving thirty two thousand unique levels either side of 0v, within a signal of around +/- 1.4V, which is probably decoded back into analogue using a delta-sigma 1 bit DAC… so it may well afford more values than the hardware is capable of reproducing. Not that this stopped a lot of CD mastering being awful, because in the ’90s all that mattered was that your CD was louder than the previous. God forbid that the listener might just, you know, turn up the volume……. :-/

I think it would be much more advantagious to have more channels rather than more samples. DVD audio is routinely 5.1 these days. Stereo seems so… old fashioned. How come there’s no 5.1 style audio format?

¹ Very quiet passages around full orchestral parts.

² Exercise your speakers, especially the bass (guitars) and treble (Annette, Floor, and Simone in particular).

³ A lot of high frequencies.

MQA is the biggest audio marketing bullshit since the gold plated oxygen free HDMI lead. It completely messes up the high end frequencies which some children/dogs can hear, to generate some approximate mess in the range only dog/bats could hear, even if your equipment could produce it and isn’t set to filter it all out anyway.

Luckily my hearing tops out about 8kHz now, so no amount of MQA nonsense will make any difference to me.

(Unfortunately neither does most of the features my very expensive Dolby Atmos 9.4.1 system, including any height effects. I prefer my old 5.1 system because you can feel more thumping coming out of the sub woofer).

I prefer my old 5.1 system because you can feel more thumping coming out of the sub woofer)

Thats because your old age hearing is now so poor the volume needs to be turned up so high and hence you feel the vibrations :-)

Pity your next door neighbours or the rest of the family who now look like Marty from Back to the future after he tries a pumped up amp.

The “1/f” claim by MQA is vague. If they were clearer it would mean saying “There is very little high HF anyway, so we can probably find some room to hide that away without people noticing.”

IIRC They only started saying this after the “Golden One” episode covered by one of my webpages as the ‘reason’ his file gave the output it did. His test file generated a LOT of ‘process noise’. Much more than in the 2L examples I’ve examined. But he dared to put a fair amount above c 20kHz into the file he sent to Tidal to encode. They encoded it and released it. The “1/f” argument was deployed when MQA found out what he’d done. Tidal didn’t object, they just processed it.

So at some point I do plan to trawl examples of music and determine the real stats of the tendency for the level to fall as the frequency rises.

Thats because your old age hearing is now so poor the volume needs to be turned up so high and hence you feel the vibrations :-)

Eh, pardon? Unfortunately true, too much listening to loud low quality stuff when I was young, so now I can afford decent stuff, I can’t hear it.

Pity your next door neighbours or the rest of the family who now look like Marty from Back to the future after he tries a pumped up amp.

The children do keep complaining they can’t hear their TV!

Going back to my actual reason for starting this thread:

Since it looks like we will – like it or not – be getting examples of MQA content to play, does anyone think they could help with the task of being able to decode MQA using the blusound code? I’m simply not capable of doing it by myself as it is way above my ‘pay grade’ as a programmer. But it would help a great deal with being able to uncover any further, erm, foibles, of MQA so people will know what is inside the tin!

I’ve managed to find out some of the effects it has, but can’t get the precison need for a full ‘diff’ that way.