New codec MQA on the horizon...

[brunok]

Tidal streaming HQ audio likely with new codec MQA.  See this link
https://www.youtube.com/watch?v=IxFafCO9khE  Interesting and balanced.

Regards
brunok

[PeterSt]

Hi Bruno and all,

(having finished this post, I see it is quite strange or vague - but I guess it shows my mood about this ?)

Personally I think MQA isn't going to make it, for a number of reasons;

First of all it targets Hires. But there is no Hires to begin with and it will never be there as well. Maybe for our grand grand children, might they still like the music produced on their path to that era.

Then, some toko like Tidal is supposed to support it for ... Hires again. So, still useless because Tidal too won't have Hires (apart from the relative few of course).

Next, indeed it is possible that such a codec is going to incorporate "DRM" (Digital Rights Management) and if that happens all stops before it even started.

Then our DACs must decode it. Yeah, sure.
OK, then the computer software can do it. It won't be for free, I'm sure, and so we are going to pay all $10 or whatever, for what ? For something which already has been covered for (call it FLAC though that will consume more space).

So you see, this can only benefit the streaming bandwidth which now will be less and from there it is back to square one because there is nothing to stream (no Hires).

Quite now :

In my view these kind of things are all hyped like this Hans Beekhuyzen is trying to hype something. But you know, guys like that, these days, go for the Youtube Likes. This is a different story, but once there was a time that this person was read (also by me) as a highly respected audio (gear) reviewer, while now he thinks he knows a lot about "streaming" as he also calls the normal audio files. But I don't think any real Dutch digital geek exists amongst this old gang. They try hard, but they are stuck anyway.
Possibly I am biased because this strange view this person puts on to us and while under way in the video I already wondered what the heck his subject was, at the end all became clear (Likes). So I don't know what these ever back fine people these days want to earn money over, but it does not work for me.

MQA is a bit similar. Some (Meridian) company sees a bit of light at the end of a tunnel and many many icenses sold because something like Qobuz or Tidal is going to offer streaming of audio which will be the future. Yes, but too bad that the target will not even be Meridian's high end audio users because no high end audio is going to be that with streaming. But who in the world know that ?
And this is part of the point - such a guy or Meridian itself have grown way out of the modern world with a hand ful (not even more) of audio players on computers. They use Audirvana because it is convenient and there it stops.

I could also say that not so many people in this audio world know what can be achieved these days, and what it takes to get there.

Another thing :
This Hans *is* coloured because he doesn't like DSD. Now he has a good reason not to mention DSD as a "codec" which is way smaller to begin with (say half). But now all runs stuck because DSD can not compress by this sort of (MQA) means.

Done ? almost, because we can also think how it is possible that I can do 256x PCM. Yes, that's different sh*t than 16x, right ? And it has one reason only : use the same available bandwidth as is available now (which would be the 480Mbps of USB). So what I did was making a codec which is 16 times smaller than current PCM and the theoretical frequency would be 5.8MHz (which again is a bit different sh*t than 192KHz).
So now I have the potential to again knock on Tidal's doors and overrule Meridian ?

No, of course not. My feet are both on the ground I hope and I don't see a whole world applying this nice feature into their DACs. They can buy ours though and I don't even have pretentions about that.

Sorry for the vagueness !
And thank you Bruno !
Peter

[Hans_Beekhuyzen]

Hi Peter,

Thanks for your kind judgement on my youtube video. Indeed, I'm trying to hype MQA and I love to have Likes. But there are only a few things you got not completely right:
- I was streaming audio when the rest of the world still thought it could never equal cd-players. BTW I also published an ibook about it (File Based Audio aka Streaming Audio).
- I do love DSD (as you could have found out easily if you had visited my site). I was amongst the first to have a DAC that supports DoP (Chord QDB76HDSD).
- I do know what can be achieved, I was the editor in chief of the Dutch Pro Audio Video magazine for 21 years, visited many studios in the Netherlands, Belgium Singapore, Japan and the UK.
- I did read all technical documentation, including the AES papers and patents. have you?

It's easy to judge on other people, it's hard to give decent comments. Q.E.D.

Best regards
Hans Beekhuyzen

[PeterSt]

Dear Hans,

Can I tell : I love it that you found this (so quickly) and are so kind to respond. And I am dead-serious. Also, I knew in advance that I could challenge for this, but did not expect it to happen.

This is only a first response, about good intentions and such.
I'll try to have a next response as well, maybe this time showing a little more respect.
Otherwise feel free what you want to say really, with the notice that we are "Dutch" anyway.

Sincere thanks ...
Peter

[PeterSt]

So were it about persons and such ...

I was the most surprised to see a Dutch "guy" doing such a video in the first place. Along with that, a name of fame (to me). Hats off only, actually.

But what I don't like about such YouTubes is nowadays general approach with one means only : Likes. The forest and the trees thing. A million of them around from how to iron a shirt to how to water plants. And how to reboot a PC of course.

You will not know it, I suppose, but quite regularly people come up with similar and I can not recall one I did not debunk. Not because I like to debunk everything and all, but because it is a tad too easy to make the videos with content which really is about something; they are not meant to. But it can bring a living (I heard).
However ...
Wat I see happening just the same is the by now sheer impossibility for someone like you, being an artist in itself, to create a nice "readable" piece of audio-story in a magazine (it doesn't go too well in that line of business). So what to do. Create a YouTube (and a few more).

My first post became that "vague" as I described it, because I can't stand the "mood" springing from such videos;
People do their best to point them out when they feel it could be important (just like Bruno did in this case) and the people creating them, also do their best for a felt good cause.
But only the best videos are the best and worth while. And now so much more is to it, if it isn't to come across as the necessity of Likes only. And in this field ... too much of them.

I made the connection with DSD, only because you did not mention it, but IMHO should have. Not that *I* personally like DSD, but if it is about the space requirement anyway ...
I just looked back but my remark about you not liking DSD was a plain mistake; I read a text on a page from PvW where he quotes you (the Audirvana thing) to next express about DSD in the next sentence which is *his* thinking.
Apologies !

Best regards,
Peter

[PeterSt]

Hans,

With the getting acquainted now behind us, what about the real content ?

Btw, if we talk about Streaming as such ... that is what Tidal offers. Or the same thing Squeezebox ever back invented.
Do we agree ? ... this is how I judge videos when I watch them anyway.
So possibly you were not Streaming back in the days, but just playing music from a computer (and then especially from the one where the playback software runs too).

So Yes, MQA will be offering a better bandwidth for that, so to speak. Now what about my comments on that ?


Meanwhile, when you have read all the AES papers and patents, maybe you are *the* person to tell us - and especially me - why DSD would be better one over PCM. And I said "better" and I do not talk about that you (or anyone) likes it better. Thus, technically.
On this forum there's hardly talk about it, which does not mean I personally don't like to (and I did so in 1000s of posts elsewhere). Also, if I work on a "best ever" DSD D/A converter (for years)- only to find out myself with apples and apples comparison, this is a quite stupid task and I like the good theories better. But not many around to lay them out, unless biased for whatever reason.

Thanks,
Peter

[Hans_Beekhuyzen]

Hi peter,

Excuses accepted. Good of you to respond this way. You must know my track record; I rather had my magazines go broke than sell my soul to the devil. It was therefore that I needed to respond. But enough about that.

The difference between PCM and DSD? The jury is still out. DSD is extremely difficult to use in production, almost always post production is done in PCM at 352,8 kHz/32 bit. The main advantage is that DSD64 has equal resolution to PCM at 352.8 kHz sampling at 32 bit depth while having a bitrate equal to 192 kHz at 24 bit depth. Another advantage is that - even when doing post in PCM - the anti aliasing and reconstruction filters cause less time smearing since they only have to filter at very high frequencies. The nice thing about MQA is that, while filtering at 192 kHz, the time smearing artifacts reportedly are compensated for while the file size comes close to a PCM at 44,1 kHz and 16 bit. Meridian (ie Craven and Stewart) had already developed a reconstruction filter that  to a degree compensated for time smearing (the apodizing filter) that sounds very 'right'. Hence my enthusiasm for MQA.

Hans

[PeterSt]

Hey Hans,

What it comes down to, in my opinion, is how any normal 16/44.1 RBCD can be transferred to a format which does the same as you described for the filtering, but now thus with the multi millions of CDs available and the thousands we like. So to me this is the only useful approach, with the knowledge that any available Hires that did not fail for any of the many reasons can be counted on two hands only and that amongst those there's maybe one I like.
Do notice that "failing" as such, for me already starts at *any* modern remaster, and all the so called Hires is amongst that to begin with. So people behind desks molesting things.
Thus : any remastered CD (just 16/44.1) already sounds way poor compared to its original counter part.
Maybe my standard is too high.

What I feel is that DSD in the end is the better format, BUT when used as upsampled means of 16/44.1 and when done right. So indeed, nothing like "only when DSD remains unmolested right from the recording" it is right. It doesn't exist anyway (when analog mixing was used Yes, but what to say ...).

So just for your information, our D/A converter is 24/768 input-capable for this filtering reason (choose any filtering means you like in-PC). Still this is not enough in my view, which is why I meantioned the 24/11.28 MHz PCM input in near future (PCM x256).
That this is also DSD x256 capable is another matter, but crucial for apples and apples comparison; now all goes through the exact same electrical means and speed and current draw etc.

What I'm also saying and "claiming" thus far (I mean, as long as the real physical thing is not operational yet) is that DSD will be the better format than PCM, when upsampled/filtered from 16/44.1. But I also said (above) "when done right". The difference ?
No noise shaping into the HF regions, no matter many will say it is harmless. So DSD as we are used to, but now without all the HF noise.

I am not saying that this will thus be better than MQA, because there's nothing to compare anyway. I mean, all what MQA will do (if all is right) is providing a lower bandwidth requiring means for Hires PCM. Remember, that Hires which does not exist in the first place.

So what this world should head for is how Redbook CD can be improved upon. At least I feel that this my job here and I think that togehther with all the helping ears (community) we do very well (in the 8-9 years spent on that by now).
Maybe this is also how I mentioned in between a few lines that most have no clue and that especially no 21 years of review etc. experience is going to help much. It is about last week's changes, and yesterday's and what pops up tomorrow. Seriously. Also, the improvements get faster and faster because we get closer and closer to the real (instrument etc.) thing and it gets more and more easy to hear the differences in objective fashion;
All we need to do is compare with the real thing, which is not much subjective ...

Hopefully this sheds a better light on my, say, bashing on MQA; it is just not the we to go forward. In my view of course, but there's some strategy behind that view.

Thank you again,
Peter

[Hans_Beekhuyzen]

Recordings made at 44.1/16 are what they are. There is no way to improve the resolution. Upsampling might give improvements when your DAC's reconstruction filter is poor (it's easier to build a filter for 192 kHz). Buying a better DAC would be a better option.

What I don't understand is your statement that there is no hires music. Many studios record at 96 or 192 kHz and when digitizing analogue recordings 192 kan be selected too.

MQA might improve 44.1 recordings since it seems to have a way to compensate for artifacts of the anti aliasing filter.

[PeterSt]

Hi again Hans,

Recordings made at 44.1/16 are what they are. There is no way to improve the resolution.

I think we are getting somewhere now. But now I don't know whether I must tell you to be 3-4 years behind, or just since "I" am doing this (make that 9 years).
Hopefully this one comes across better to you as well :

It's easy to judge on other people, it's hard to give decent comments. Q.E.D.

Notice that you said "resolution" but it is not about that. It is about the filtering.

Upsampling might give improvements when your DAC's reconstruction filter is poor (it's easier to build a filter for 192 kHz).

So Yes and No. But I easily claim : you don't know everything about it. Uh-oh ... take with a grain of salt since you're not the only one. Told that in my first post (right ? - QED and such).

So maybe FYI :
This nice filtering thing is what I started (on this globe) and it is more hot than DSD I tell you (this is already because DSD requires "super" filtering).
So I am not saying you need to learn something (and especially not from me) but the trend is applying the filtering in-PC. Investigate a bit and btw, use Audirvana for it (but has one (too) weak means only - iZotope).

Investigate how "we" overrule the in-DAC filtering, or better, how a DAC these days should not contain any filtering at all.
And now "we" must buy a better DAC ? some people from this forum will be smiling at you - or at me. But who cares as long as we keep smiling.

What I don't understand is your statement that there is no hires music. Many studios record at 96 or 192 kHz

Oh. Such as ?
But I beg you, name those who record the music I like. Ah, you might be able to come up with one or two albums (I like) but what must I do with those ? play them every night because there's nothing else ?
I still don't have *any*, but my vision could be limitit (still I have over 40K of albums here and over 1K are this nice Hires).

And thus it is still so that that a lot of hype is going on for no reason, or no strategy if you want. Like I said it (first post) ... maybe our grand grand children can benefit from today's recordings but I'm afraid the music at that time will be even worse than today's Billsh*t top 100. Now is this the US, but I don't think it makes much of a difference.

So yeah, I'm afraid you are directing me to 2L and some Beethoven or a nice choir. But not my music. At least that is Hires done (very) well.

MQA might improve 44.1 recordings since it seems to have a way to compensate for artifacts of the anti aliasing filter.

I'm all ears Hans.
But I'm afraid this won't be related to MQA other than I create nice filters without the "hassle" of a compression scheme. Remember, "we" don't need that. And btw, we also don't need the Tidal's. Not since they are not open to stream their music in well fashion so that "our" fitering can be used. And yes, I tried (see topic elsewhere).

So there's a few miles of distance between you and me (and you and all others on this forum). So what. It is only that I told this already. And I wish it were not so. Why ? well, because I too some times like to learn from others/elsewhere. Really so.

Snobbish me.

[Hans_Beekhuyzen]

Re: upsampling: I know many people who do that and with success. But they are only compensating for poor reconstruction filters and thus not improving the recording. They are merely working around a hardware problem. Which is ok btw.

Re: resolution. I was talking about time resolution and thus filtering

Re: hires recordings. I own about 500 SACD's that I play 'ripped'. The most part contains information above 20 kHz and thus must be hires. Classics like Jacques Brel (tape transcript) but also Herbert Grönemeyer which definitely is recorded in hires (I know the mastering engineer Ronald Prent personally and heard the recording in the mastering suite).

RE MQA and time smearing: here you're absolutely wrong. There yet is no other 'codec' (essentially a wrong moniker here) that does contain metadata on the impulse respons.

[CoenP]

RE MQA and time smearing: here you're absolutely wrong. There yet is no other 'codec' (essentially a wrong moniker here) that does contain metadata on the impulse respons.

Hello Hans,

IIRC: The HDCD coding contains explicit information to select the "best" filter and as such modifies the impulse response constantly. Are you suggesting that the HF information in MQA is there to make the filter reproduce the time domain correctly?

Regards, Coen

[Hans_Beekhuyzen]

IIRC: The HDCD coding contains explicit information to select the "best" filter and as such modifies the impulse response constantly. Are you suggesting that the HF information in MQA is there to make the filter reproduce the time domain correctly?


No, it's not in the HF information. For as far as I understand it now, they have developed a way to store the impulse respons of the complete recording chain in metadata. They are rather secretive about it and questions I posted to Meridian remained unanswered. But Steward and Craven are extremely knowledgable and have been following research  on spycho acoustics and neuro science intensely. If they managed as they claim, it really will set new standards on audio quality.

BTW how do you make these 'quotes'

[Hans_Beekhuyzen]

BTW how do you make these 'quotes'

....found it

[PeterSt]

With the notice that it doesn't happen often that I'm wrong ...
I'll bet you that MQA is doing this :
(just my own idea, taken from nowhere)

Notice : On the "meta data" thing, so unrelated to the compression technique.

An Analog to Digital converter, being a Sigma Delta Modulator, works with a filter. Actually such a filter is similar to the ones in D/A converters, or as we know, in computer software. However, it is only "similar" and not the same. It can be the same, but it would be sheer coincidence (and the chance for it to be the same is regarded to be zero).

Let's name the parameters to such an ADC filter the "Loop Filter Parameters"; they define how the filter works, how many of (cascaded) filters are in order, etc. - all parameters to an actually normal filter we could use in our playback means as well (D/A converter, Software).

Let's say that when filtering is in order (it always is for 16/44.1 unless someone likes the distortion of NOS (Non OverSampling) / Filterless) that such a filter describes the sample to sample transition; The oversampling (or upsampling - does not matter although technically it is not the same) places the samples at different positions in the time domain (simply put : at more positions because the sampling rate gets higher) and at different amplitude levels (which is called the "reconstruction" as such) so we get a more close resemblance of the original signal.

The "original signal" would be the analogue signal and the ADC involved at the recording always "under samples" that signal (sampling is finite, like e.g. 192KHz is way far from analogue, which would require infinite sampling).

The Loop Filter Parameters describe how the ADC came to the amplitude levels of the samples, which are in fixed time positions (like evenly spread 192000 times over one second).
If we apply these same Loop Filter Parameters to the D/A process, we get super close to the signal as how the ADC saw it during the recording of the signal (the music).

Now easy (once you thougt of the method) : Incorporate a few bytes of "meta data" about those used Loop Filter Parameters in the music file itself, and let the D/A converter use those when the file is played.

In my view this is the only possibility to vastly improve the played back quality of the music, and how it can get mighty close to what the recording "saw".
Do notice that what people report from it is, to me, a bit beyond what can be normal, unless the file is processed as such and a large degree of subjectivity will be in order. So, this is NOT what I think and I thus think that something real is going on. See description above.


So assumed I am right (and of course I am ) then too bad this has been taken from a Philips patent. Here :
http://www.google.com.ar/patents/WO2006129215A2?cl=en
Notice : I have been trying to find back this one I already knew about, but I don't think it is the very same because the one in my mind incorporated hardware, or explicitly referenced to that as required as well (at the playback end). I didn't read this one throughout, but I don't see the hardware part at a glance. So I think there is another one as well.

This one, summaraized, uses PCM to encapsulate DSD. Yes, laugh. But it uses PCM as an almost lossless format to compress DSD. Think 16/44.1 files which represent DSD files which latter are twice as large and "Hires", and where the Loop Filter Parameters are used for the again way under sampled PCM samples and how the one sample transitions (originally transitioned) to the next one.
Smart eh ?

What also would be smart is that possibly this patent is not violated by Meridian because they (most probably) won't take DSD as a base. So while the patent functionally describes a DSD compression method (while it would work just the same for PCM but doesn't describe that), the real functionality possibly is not patented : the Loop Filter Parameters meta data.

Still assumed I am correct, if I were Meridian I would not talk about this too much; May win may lose. On the other hand, and obvious to me, *I* would need some description of what's happening, instead of the blah blah that our audio world will change. I mean, this is too much of commercial blah and that without one single word on the technique used. Why ?
I think I know.


To make my story complete :

When we look at the (few) raving reports of auditions, this is NOT about new recordings. I also take into account the "remastering" of analogue tapes etc. I see mentioned and how that can be regarded a new recording as such, because of the known digitising device (ADC) and thus its parameters. But also, when the ADC is known which was used for the recording or the digitising in the more recent past (say after 1983) then too the parameters can be known (just look in the datasheet of the ADC chip for the most easy interpretable ADC situations (say not the Pacific Microsonics);

What's to be done is that all the already available digital files are to be re-encoded, while all what really needs to happen is embed that meta data. Well, then someone will be so smart to do that himself just the same without even re-encoding (just store the meta data next to the file), say the playback software developers. But needs some work on finding data about the ADC used and what's in there for technology. So still not easy.

That's all.
Peter