Meridian's mysterious MQA site.

[Lighthouse]

Meridian presents MQA

 

 

My quick guess is either a new music format or something. Anyone knows the details?

 

It seems it will be revealed tomorrow.

[Audio_ELF]

Not sure - perhaps a portable media player?

 

There is going to be a Explorer v2 DAC on the day after (6th December).

[monteverdi]

Seems to be a new file format Meridian Audio MQA paves way for high-res streaming | What Hi-Fi?

as far as i can understand from that little blurb the encoding is diiferent from existing formats also involving somehow a time component. May be quite interesting -or only an other confusion of the audio market.

[firedog]

Robert Harley Listens to Meridian MQA | The Absolute Sound

[Boris75]

Seems to be a new file format Meridian Audio MQA paves way for high-res streaming | What Hi-Fi?

as far as i can understand from that little blurb the encoding is diiferent from existing formats also involving somehow a time component. May be quite interesting -or only an other confusion of the audio market.

 

Proprietary, lossy "high-res" is the last thing we need, really.

[flatmap]

Proprietary, lossy "high-res" is the last thing we need, really.

 

Boris75, you're probably reading this more deeply than I, but I'm not seeing anything that says it

will be lossy. In fact, the What hi-fi? article says that MQA can use "any lossless container."

 

Did I miss something?

[YashN]

Robert Harley Listens to Meridian MQA | The Absolute Sound

 

That was a very nice article and comment thread, Firedog. Thanks for sharing.

 

I am curious to know the details of the three axes at recording and how the format works.

 

Bob Stuart is very knowledgeable when it comes to sound.

 

I wonder if starting with the premise that the ears have low bandwidth is a good one though.

 

Looking forward to listening to this and seeing if it can best DSD 2x on a native DSD DAC.

[Boris75]

Boris75, you're probably reading this more deeply than I, but I'm not seeing anything that says it

will be lossy. In fact, the What hi-fi? article says that MQA can use "any lossless container."

 

Did I miss something?

 

Here is what Mark Waldrep, who has been exchanging e-mails with Meridian about the technology, writes on his blog:

The innovation made possible by this new technology is the ability to deliver the benefits of high-resolution audio (96 or 192 kHz/24-bit PCM audio) in a container that is roughly the same size and has the same bandwidth of a CD

 

Delivering the benefits of 24/192-resolution audio in a container that has the same bandwith of a CD can only be done by eliminating part of the information. Aggressive lossless encoding can at most reduce bandwith by a factor of a bit more than 2, but going from 24/192 to 16/44.1 requires a factor of 6.5. The idea behind the scheme is to throw away high-frequency information but to retain tight time-domain information:

 

Meridian’s MQA does this by creating metadata about the time resolution (and other aspects of the analog signal) into a very low level signal (below the level of the noise) in a PCM file with a much lower sample rate.

 

This is clever, yes, but definitely looks lossy.

 

On top of that, it is proprietary.

 

I doubt high-res audio can really get a lift from a proprietary format for which content and hardware producers would have to pay royalties...

[YashN]

Here is what Mark Waldrep, who has been exchanging e-mails with Meridian about the technology, writes on his blog.

 

 

This is clever, yes, but definitely looks lossy.

 

 

It isn't claimed to be lossy, but the precise opposite, lossless.

 

It seems you aren't really understanding that the format is using PCM as a wrapper of sorts. There's no automatic loss here. If there were, such a thing as DOP couldn't exist (DSD over PCM).

 

On the other hand, we would need more information to now for sure how they do the recording and encoding process.

 

Because the initial premise is that the ear is a low bandwidth sense, it is probable that at capture of at least one of the three axes of recording, some resolution is lost or discarded [as is inevitably the case when going from the analogue domain (infinite resolution) to the digital one] but more so in this specific case on MQA, because the technology presumably relies on psycho-acoustic perception and its models and thresholds.

 

The question then will remain whether the information encoded and decoded provide a resolution high enough than our ears perceive the sound as better than existing conventional means.

 

Waldrep is continuously sceptical and critical of anything other than PCM. I wouldn't use his information as reference for anything at all.

[audiventory]

Proprietary, lossy "high-res" is the last thing we need, really.

Theoretically, if format will give great improvement for compressing (as example) possibly paid license is justified.

But DSD some time ago almost died due same reason.

[Boris75]

It isn't claimed to be lossy, but the precise opposite, lossless.

 

It seems you aren't really understanding that the format is using PCM as a wrapper of sorts. There's no automatic loss here. If there were, such a thing as DOP couldn't exist (DSD over PCM).

 

On the other hand, we would need more information to now for sure how they do the recording and encoding process.

 

Because the initial premise is that the ear is a low bandwidth sense, it is probable that at capture of at least one of the three axes of recording, some resolution is lost or discarded [as is inevitably the case when going from the analogue domain (infinite resolution) to the digital one] but more so in this specific case on MQA, because the technology presumably relies on psycho-acoustic perception and its models and thresholds.

 

The question then will remain whether the information encoded and decoded provide a resolution high enough than our ears perceive the sound as better than existing conventional means.

 

Waldrep is continuously sceptical and critical of anything other than PCM. I wouldn't use his information as reference for anything at all.

 

You did not understand my point, so I'll try to be clearer. Yes, putting something in a PCM container does not make it lossy. Your example of DoP is right on point. But putting 24/192 information in a 16/44.1 has to be lossy, because the implied compression ratio is mathematically too high for lossless compression. Then, as you say, clever psycho-acoustic thresholds can reduce the audible impact of the compression. However, this "underlying premise", as you put it, is also the one that underpins mp3s, so I am not too thrilled by it.

[Boris75]

 

Looking forward to listening to this and seeing if it can best DSD 2x on a native DSD DAC.

 

I very much doubt it as I generally don't believe in miracles: DSD 2x has much more bandwidth.

[jhwalker]

I very much doubt it as I generally don't believe in miracles: DSD 2x has much more bandwidth.

 

But their contention is we don't *need* all that bandwidth to fully and completely encode the music; i.e., that both PCM and DSD are using lots of data to describe inaudible, ultrasonic "sound" at a very low level in the mix.

 

From what I'm reading, it's like the difference between a bitmap (i.e., a one-to-one map of every single pixel on a screen) vs. encoding the content of a screen using metadata - you'd say an *awful* lot of data that way, particularly if large parts of the screen contained no meaningful content. So long as a reasonable definition of meaningful content is used (e.g., "anything potentially audible" - between 5-25kHz and higher than -100db below peak volume, say), I'd be very interested in giving it a listen and potentially saving a lot of storage space and bandwidth.

[audiventory]

I very much doubt it as I generally don't believe in miracles: DSD 2x has much more bandwidth.

Audio quality mathematically we can consider as (is not exact math of course, only illustration):

 

Quality = [bit depth] x [sample rate]

 

For keeping quality we can:

a) increase bit depth but decrease sample rate

b) increase sample rate but decrease bit depth

 

For lowest bit depth (1 bit) is not enough made band wide as possible. Formula don't work. Need apply dithering. Dithering noise we shift out audible range.

 

Thus wide band used in DSD not for improving quality. Used for storing unwanted noise.

 

I made measurements and comparing different formats PCM16/24/32 DSD64/128. There software work about theoretical limits.

How Impact to Audio Quality of PCM to DSF Conversion. 1-bit DSF vs. PCM

 

Quality of Audio Formats by Results These Tests

(first place is best)

 

1. WAV PCM 32-bit float

 

2. DSF D128

 

3. WAV PCM 24-bit

 

4. DSF D64

 

5. WAV PCM 16-bit

[firedog]

Instead of arguing over whether this will sound good or not, let's wait and see. Maybe the concept is very good and the "lost" material is actually irrrelevant; maybe not. The people at Meridian certainly have a track record for which they should be given the benefit of the doubt for now.

 

I'm intrigued/excited by this. It could be quite a useful innovation for audiophile downloads.

[Maldur]

I made measurements and comparing different formats PCM16/24/32 DSD64/128. There software work about theoretical limits.

How Impact to Audio Quality of PCM to DSF Conversion. 1-bit DSF vs. PCM

 

 

Nice test... but again not for human ears, for measuring programs instead. So, we humans can ignore the conclusions.

[esldude]

Audio quality mathematically we can consider as (is not exact math of course, only illustration):

 

Quality = [bit depth] x [sample rate]

 

For keeping quality we can:

a) increase bit depth but decrease sample rate

b) increase sample rate but decrease bit depth

 

For lowest bit depth (1 bit) is not enough made band wide as possible. Formula don't work. Need apply dithering. Dithering noise we shift out audible range.

 

Thus wide band used in DSD not for improving quality. Used for storing unwanted noise.

 

I made measurements and comparing different formats PCM16/24/32 DSD64/128. There software work about theoretical limits.

How Impact to Audio Quality of PCM to DSF Conversion. 1-bit DSF vs. PCM

 

Quality of Audio Formats by Results These Tests

(first place is best)

 

1. WAV PCM 32-bit float

 

2. DSF D128

 

3. WAV PCM 24-bit

 

4. DSF D64

 

5. WAV PCM 16-bit

 

That is all well and good, but just for instance FLAC can be around half the total number of bits and expand losslessly into a full file that loses not one single bit. This MQA looks to do even more also without loosing one bit. MQA looks to be 4 to 1 or better compression while losing nothing.

[Boris75]

That is all well and good, but just for instance FLAC can be around half the total number of bits and expand losslessly into a full file that loses not one single bit. This MQA looks to do even more also without loosing one bit. MQA looks to be 4 to 1 or better compression while losing nothing.

 

My understanding of MQA is that, unlike FLAC or ALAC, MQA files lose bits, but Meridian assures us that these bits do not matter - as the mp3 promoters did.

[YashN]

But putting 24/192 information in a 16/44.1 has to be lossy, because the implied compression ratio is mathematically too high for lossless compression.

 

No, it still isn't automatic since PCM is only used as a wrapper...

[YashN]

I'll wait till I hear it in a revealing system before making any conclusion myself. I will not dismiss it simply because not enough is known about the method and implementation.

 

Probably Stuart did give some relevant info in his recent papers but I haven't had access to them.

 

I very much doubt it as I generally don't believe in miracles: DSD 2x has much more bandwidth.

[YashN]

That's what I think as well.

 

I am certainly in agreement with giving more importance to the accuracy of the time domain nowadays insofar as, firstly, I consider frequency reproduction a solved problem (of course there are still better solutions and worse solutions here too), but now I consider getting fast and accurate transient response a must for dynamics, timbre-recognition and soundstage and reverb tails, etc..., so secondly, if more information in the time-domain axis for MQA allows better transients, then it will be very interesting to hear.

 

 

Instead of arguing over whether this will sound good or not, let's wait and see. Maybe the concept is very good and the "lost" material is actually irrrelevant; maybe not. The people at Meridian certainly have a track record for which they should be given the benefit of the doubt for now.

 

I'm intrigued/excited by this. It could be quite a useful innovation for audiophile downloads.

[YashN]

Yes, they mention not having to code unnecessary data, like silent parts.

 

This makes it interesting to think about. For example, perhaps the time axis is used to pinpoint where the audio in the file has to be played and the audio portion is coded contiguously without silence.

 

So let's say a track has 2 seconds of silence at 1:33, I could code a contiguous audio portion without those 2 seconds of silence (arguably something very near all zeroes at 16-bit, 24-bit, 32-bit resolution, but still taking a lot of sample storage space), and then in a separate time axis, mention that actual audio stops at 1:33, and only starts again at 1:35.

 

On playback, the necessary bits for silence will be filled in in real-time at the player.

 

If the additional coding of the time axis this way doesn't exceed the sample size saved, the format is overall smaller.

 

There could be many more happening. I am eager to learn more about what they did and how, and listen to it too.

 

But their contention is we don't *need* all that bandwidth to fully and completely encode the music; i.e., that both PCM and DSD are using lots of data to describe inaudible, ultrasonic "sound" at a very low level in the mix.

 

From what I'm reading, it's like the difference between a bitmap (i.e., a one-to-one map of every single pixel on a screen) vs. encoding the content of a screen using metadata - you'd say an *awful* lot of data that way, particularly if large parts of the screen contained no meaningful content. So long as a reasonable definition of meaningful content is used (e.g., "anything potentially audible" - between 5-25kHz and higher than -100db below peak volume, say), I'd be very interested in giving it a listen and potentially saving a lot of storage space and bandwidth.

[Matias]

If I understand correctly, this is a new format to use the "empty file space" from unused frequencies in audio files to encode more data from the frequencies being used. A trick to store more music data in a smaller package. But as far as sound quality, it is the same as an "uncompressed" high-res PCM file. MQA only makes it more compact, like a super-FLAC or something, which is good for streaming or selling download using less bandwidth.

Did I miss something?

[monteverdi]

If I understand correctly, this is a new format to use the "empty file space" from unused frequencies in audio files to encode more data from the frequencies being used. A trick to store more music data in a smaller package. But as far as sound quality, it is the same as an "uncompressed" high-res PCM file. MQA only makes it more compact, like a super-FLAC or something, which is good for streaming or selling download using less bandwidth.

Did I miss something?

 

If the empty file space is used to compress just higher rate/depth PCM data than it is only a more efficient than FLAC or ALAC etc. But that empty file space could also contain time related information and that would be completely different from standard encoding (less or no problem with jitter).

[spdif-usb]

If I understand correctly, this is a new format to use the "empty file space" from unused frequencies in audio files to encode more data from the frequencies being used. A trick to store more music data in a smaller package. But as far as sound quality, it is the same as an "uncompressed" high-res PCM file. MQA only makes it more compact, like a super-FLAC or something, which is good for streaming or selling download using less bandwidth.

Did I miss something?

No, it is not really a new format (i.e., it is compatible with existing formats!), and no, it does not use the "empty file space" from unused frequencies in audio files (i.e., it uses metadata, or "tags" to store additional data that can then be read by the MQA decoder app, software player, or hardware). As far as sound quality, no, it is not the same as an "uncompressed" high-res PCM file, but the audible difference, if any, is so small that one could peruse that it is either equivalent in terms of human hearing, or, in a worst case scenario, a VERY extremely hair-splittingly close approximation. Reason why this is possible is because it is la crème de la crème result of bringing together advances in sampling theory with recent findings in human auditory science.

 

In a (small) nutshell,

Meridian Launch MQA

( Also worth reading is AES Convention Paper 9178—A Hierarchical Approach to Archiving and Distribution by Bob Stuart & Peter Craven.)