MQA is Vaporware

[ipeverywhere]

11 hours ago, Paul R said:

The same type of compression that MQA uses has been used in telephony for ages. It is common in all VoIP that I know of.  Where do you think MQA got the idea from anyway? It seems to be pretty common knowledge. After reading your post I

 

Not exactly. I believe you're associating "VoIP" with MQA because there are specific CODECs which can be negotiated for a voice call which are ADPCM based CODECs. However, ADPCM is not a CODEC itself. ADPCM defines how to store the data between adjacent PCM samples as well as how the far end then needs to reconstruct the two adjacent samples with only having about half the bits (kind of like in video if you have pixels 1, 2, 3, 4, and 5 but only send pixels 2 and 4 you can predict or extrapolate what pixels 1, 3, and 5 should be). Theoretically, this should result in half the bandwidth or data store needed for the audio file.

 

But... we've not defined what we're sampling, any filters applied to that sample, or any compression being applied yet and that's the more important parts when it comes to audio quality. Most VoIP calls are sampled at a 8khz rate. The most popular, because it's royalty free, takes about 64kbit/sec to encode the 8khz samples with no compression/filters/etc. There are royalty based CODECs which still sample at 8khz but only need 16kbit/sec or greater, but still less than 64kbit/sec, to sample that same 8khz because those CODECs use ADPCM when sampling the audio. The difference between the two, at least for voice, is not perceivable by the listener. Most (or all really) HD cellular CODECs use ADPCM to reduce the bandwidth required per call while still sampling at 16khz or higher. In fact most cellular CODECs sample 16khz (twice the freq of your copper telephone line), use ADPCM, and apply compression to remove various things that are not important to a voice call which yields an actual bandwidth requirement of 24, 32, 42kbit/sec or somewhere in there. Much better sound quality at much less bandwidth. But, don't forget, my example of "better" starts with the "standard" being an 8khz sample rate. It's easy to make 8khz samples "better".

 

Lots of royalty based CODECs use ADPCM at all kinds of different sampling rates (AptX is one designed to fit audio in the specific bandwidth constraints of BT). However, it is always throwing away some amount of the data being sampled and that leaves the far-end responsible for reconstructing the missing bits. For a lot of things this works without any perceived audio loss.

 

OK, back on the topic of MQA. If the idea here is to get "hi-rez", or greater than a 44.1khz sampling rate, in the same amount of bandwidth as the original 44.1khz PCM sample then MQA is a success. The comparison/goal of MQA should be to deliver higher than redbook quality within the same bandwidth as redbook. As much as MQA might want us to buy-in to the fact they are the same or better than the hi-rez PCM/DSD? Um, no. Will a > 44.1khz MQA file sound the same as the "raw" PCM file without MQA applied? It might... but probably not (strong on the not here) since that is dependent on the playback device accurately reproducing the missing samples that got thrown away during the ADPCM process. MQA has some magic sauce as to what is being sampled for ADPCM and exactly which samples are being thrown away during the ADPCM process. This may make the reconstruction at the listening end far more accurate. But, it's still reconstructing bits and samples at the listener that don't exist in the file/stream and that's going to be inaccurate when compared to the bit-perfect master.

 

I like MQA for what it is. But, I'd rather have the raw, lossless, PCM/DSD stream because I don't have the bandwidth problem MQA is trying to solve. I'm not even going to comment on the DRM bits.

 

Anyway, hope that helps.

 

Could you set-up a VoIP call with MQA? As long as the encoding was fast enough to be "real-time" and it's probably not. Most "streaming" CODECs are specifically designed to be slow and complicated encoders but very fast and simple decoders. This makes sense as all the heavy processing of what to compress happens at encode and then the act of decode can be fast with less processor power. All VoIP CODECs are basically equal on their encode/decode times to keep things in sync and this process needs to be as low latency as possible (or in the case of cellular have some predictive/fixed latency within the chipset of your handset).

 

 

 

 

 

[ipeverywhere]

26 minutes ago, Sonicularity said:

 

MQA taking over the world is the only real concern when you boil everything down. When would be a more appropriate time to take a stance in opposition to MQA's adaptation?

 

I'll reckon that 90% of music streamers (those individuals who stream music) have no idea what they are streaming. If you asked those who switched from Pandora and Spotify to Apple Music if they knew or cared that they also switched codecs they would say "no". FLAC to MQA to MQA2 to DSD or whatever the streaming space is going to do next will be a blip on any streaming services subscription numbers because there just are not that many people paying attention. The number of people following a codec from service to service are just too few to matter*.

 

When people buy 0 MQA files maybe they will get the hint. But, no one is, should be, expecting an increase on the sale of music files. It's all about streaming revenue and maximizing streaming revenue. If MQA thinks their primary revenue stream is from selling copies of "masters" they are going to fail so I cannot imagine this is their growth strategy. And, I'll make another guess into the future...  The number of people who don't know what a codec is _and_ buy files will drop to 0 as those individuals will all shift to streaming exclusively. The number of people who do know what a codec is/file format is and buy music will go flat or maybe even rise <shrug>. Those people will (should) put formats like MQA out of the business of "selling files". But, again, no one is focused on selling files. It's a dead future.

 

The future of MQA is streaming and there are just not enough people who have any idea what that means to voice an opinion on it. (my opinion of course) 

 

*The only exception here is when you cannot stream over cellular because the file size is too big or you need a massive data plan. Then  masses will get vocal. This is why all streaming services default their mobile apps to a lower bitrate normal/standard. Listeners will hear music skip, jump, and stutter, well before they notice a higher resolution stream. 

 

 

 

 

[ipeverywhere]

15 hours ago, Paul R said:

transmission, so very little, if anything, will be saved by MQA vs ALAC or FLAC that it is completely a non-issue. Might feel a bit non-intuitive, but the cold hard engineering facts are immutable. 

 

 

Paul this is well said and I'll back this up (we may be in similar industry). Only thing I will add is that there is significantly more cost savings with regards to bandwidth by tweaking the UI to default back to the "normal" setting every now and then. When most users cannot hear the difference because they have earbuds turned to 11 on a noisy bus they won't flip that setting back to "high quality" and the whole argument of how to store/stream the "high quality" is mute. In fact...   Those of us who can hear the difference will be a tad frustrated but I'd rather have to flip a switch back to high quality after every upgrade than be reminded every time high quality equates to "MQA" coming out of my system. Anyway, there are plenty of ways to reduce the bandwidth bill and MQA should not be part of that discussion if anyone is serious about truly reducing the bandwidth. MQA is the solution you use when the non-technical/non-engineering people need to say they did something. 

 

A couple months ago I would have absolutely believed the MQA bandwidth / storage saving "hype" but there are numerous points within this thread that now totally debunk any advantage MQA provides with regard to this. Additionally, now that I have a much better understanding of what MQA is doing it's pretty obvious the savings simply cannot be there. It seems like a selling point that should have already been discredited and removed from the marketing glossy. In fact...  It would almost make sense to me if we saw a "my high quality is better" wars when Spotify and the like open their high quality service. This would be a great marketing argument against MQA and a differentiator to those services. 

 

But... as I browse other audio forums that are not this one...   I am seeing a pretty significant following for MQA. To the extreme that when someone says they "dislike" MQA or say it doesn't sound as good as some other format the respondents are all over the original poster on how to "fix" their system because "it's just not possible you didn't find MQA to be the best". 

[ipeverywhere]

41 minutes ago, Rt66indierock said:

 

Those thoughts mirror a couple of journalist's in my rogues gallery of stupid MQA comments. If you need "a highly resolving system" you have no market just like high resolution.

 

I'm actually reading comments that are worse than this. It's one thing for someone to say a system needs upgrades because the individual cannot hear the difference between two formats; people have been doing this forever. What I'm actually reading are comments saying that there is _something wrong_ with a system because MQA doesn't sound better. <rolls eyes> 

 

[ipeverywhere]

6 minutes ago, Paul R said:

 Absolutely - I should have said "seems to be better"  - because it actually might not be. But people get that impression when they listen to it. Even me.  

-Paul 

 

 

Play 45 seconds of a song someone loves. Then hit the "loudness" button and play the same 45 seconds. I'd bet money the person prefers the 45 seconds with loudness on. Now make them listen to the whole album on repeat for 3+ hours. After being totally fatigued from the EQ'ing / other crap loudness generally does I would also bet that person prefers loudness off for the rest of their listening session. 

 

This was my audio journey with MQA (decoded/first unfold in software, I do not have MQA hardware). Initial there was a "WOW! That's alive and fun and different." But then after a few hours with it I found it to be really fatiguing to listen to. It just started to sound processed, digitized, emphasized in the wrong places, and generally overdone. Like those early HDR photos. Some people will love the MQA filters over their current redbook set-up and I totally get that. Heck, I don't even use the DAC in my CD player anymore because I prefer what my external DAC is doing when I use the player as transport. Doesn't mean my external DAC is "better". I just prefer it. 

 

There are fundamental reasons to stop MQA from polluting what gets streamed / downloaded into our homes. But I don't put as much value in debating if sound quality is one of those things. MQA, for some people, will sound better than what they are used to. Doesn't matter if bits are missing. Heck, an EQ'd/room corrected 128kbps MP3 that better matches someones listening environment might very well be preferred over a 24/192 file. Who am I to tell that person their preference is wrong even if I can scientifically document why I'm right?

 

 

 

[ipeverywhere]

Question as it's unclear to me (I may have missed this page somewhere):

Is MQA really "compressing" the higher frequencies or using something like ADPCM to sample them?

 

If it's ADPCM I wouldn't call that "compression" in the way most people think of lossy compression. When I think of lossy compression I define that as a filter literally throwing away specific parts of the waveform so that those parts don't need to be encoded digitally. 

 

ADPCM is very different. ADPCM doesn't have any concept of what's in the waveform. ADPCM is literally throwing away entire samples. It does this in a way that those missing samples can be reconstructed on the other side. This generally gives you an _extremely_ close duplication of the waveform with only half the bits required (or however many bits equate to how many samples you're willing to throw away). That's an oversimplification of how ADPCM works but.... I'm trying to understand if reconstructing the MQA "compressed" part is an ADPCM function to totally reconstruct the missing samples or if it is truly lossy compression where some parts of the waveform were never samples to begin with. 

 

It may be a combination of both. 

[ipeverywhere]

21 minutes ago, new_media said:

 

I have to wonder how much of a problem piracy even is anymore. Why bother downloading pirated files when you can listen to anything and everything on the free Spotify tier? I'm sure the RIAA still tracks data on it but I haven't seen them publicize it recently.

 

 

 

Filed just this past week: https://musically.com/2019/03/26/major-labels-sue-american-isp-charter-communications/

 

Of the people I know who were heavily into this back in the Napster days those still doing it are more interested in the number of files they own than actually listening to anything they download. Most people I know who used to pirate music are now focused on pulling down movies and have pretty extensive Plex libraries of pirated stuff. You don't hear much about the RIAA going after "people" anymore because they had to stop doing that. Suing individuals was destroying their reputation and souring the entire industry. It backfired. They are now going after the ISPs for inaction when they identify the source of an illegal repository and the ISP doesn't close the account. Not sure they will be successful going after the ISPs though. So, yes, piracy is still alive and well if for no other reason than people can. I'm continuously amazed where people find the freetime to inconvenience themselves over just paying the $9.99 a month but, I guess, there are stranger hobbies. 

 

 

[ipeverywhere]

28 minutes ago, Jud said:

I'd put money on the proposition that it's lower than the number of AV receivers sold in Los Angeles in a comparable period.

 

Lower yes but I don't think MQA cares about high-end DACS so much. Remember that AV receiver _is_ a DAC and can probably directly stream at least Spotify. Slap a MQA logo alongside the other 10 or so formats from DTS and Dolby and others already on the box and the owner of that AV receiver _will_ tell you that they have never heard their streaming service sound better. MQA just won another cheerleader. 

 

 

[ipeverywhere]

14 minutes ago, new_media said:

 

I would guess Netflix is the #1 reason customers want high speed internet.

 

I believe the order is this:

1) The number is bigger and bigger numbers are gooder. 

2) Netflix.

3) I actually work from home and need to get stuff done. 

 

I'll add one more in all seriousness though...

0) My Wifi network is horrendously mis-configured and poorly implemented so I'm getting less than 10% of my Internet speed on all my devices but I decided to blame my ISP for this and so I'm going to upgrade to even higher speed Internet I cannot use. Will repeat this process when my contract runs out. 

[ipeverywhere]

There are a lot of things people do in this world that I will simply not understand and that puts me in the _minority_ of people not doing those things. Copying streamed music is certainly one of them. But, I guess, this is no different than making mix tapes off the radio which is something I did do back when that was a thing. I'm sure the kids these days appreciate not having a DJ talking for the first third of every song in your collection. 

 

 

[ipeverywhere]

I'm sure there are plenty of these examples but I found one for the first time yesterday: An EP released in 24/44.1.... 

Qobuz - 24/44.1 PCM

Tidal - 24/44.1 MQA

Release is from this year so I assume these are from the same master. 

 

The album does sound different (by different I mean worse / better with/without MQA) and even track to track they sound different. Is this the purist way to hear what MQA is doing before getting lossy in the "beyond redbook" frequencies? I assume that the MQA version is just applying the MQA filtering and debluring?