MQA Off-Topic Spinoff

[fas42]

3 hours ago, PeterSt said:

 

Nah ...

It will hurt ears, irrespective of it being a great piece. Or I just don't understand a few things (which always is the case ).

 

Maybe I'm different from many, but I find, within reason, that there is no such thing as "too loud" - if the sound is ugly at higher volumes then it's because the playback is distorting too much. Clean replay can be done at volumes ranging from a barely audible whisper to as loud as the system, and you, can sustain - and it always "sounds right".

 

If the volumes are excessive for too long, then the ears will start ringing - but it should never hurt, IME.

[fas42]

54 minutes ago, Brian Lucey said:

There is nothing about 2 speakers that even remotely equals a live event.  I'm sorry that is a joke. Go to a live event, and listen to a record.  Nothing alike.   Even surround fails badly ... and forces people to sit JUST HERE in he middle, more dogma and control.  Recordings are FAKE, all of them.  The aim is an emotional response, and connection to the artist, not DR for you.

 

Sorry, 2 speakers can sound like a live event - I did this decades ago, and found going to live events afterwards sometimes to be a bit of a letdown - the impact, and sense of connection with the performance was of a lesser order, in the live, acoustic environment. That this can happen makes sense, because recording microphones will typically in the absolutely optimum position for picking up the sound; a poor paying member of the public takes his chances with where he sits.

 

The emotional response can happen with recordings with a huge range of DR - just get the playback to do its job well enough.

[fas42]

8 hours ago, Brian Lucey said:

Bottom line ...  2 speakers is never TRULY recreating a live event.  Sorry.   It's just a degree of illusion that you find acceptable.  A line in the sand you can live with based on DR and recording quality.  I get it.

 

Yet your own words are self contradictory.  2 speakers can make live a let down?   Then they are not the same.

 

Yes, always an illusion - but far more than just "acceptable", and something I "can live with". As you appear to be saying, the emotional 'hit' is all important - and high quality replay delivers that, consistently. Poor DR is the worst offender in making it harder to deliver the emotion - because the lack of light and shade, without letup, is disturbing to listen to; unkempt recording quality OTOH can be "listened through", to quite an amazing degree.

 

"Sameness" is not the issue - what is relevant is whether reproduction delivers the equivalent, or better, kick of the real thing.

[fas42]

1 hour ago, Don Hills said:

 

I realise you're speaking partly in jest, but it's a serious problem. You can slam music, but you can't un-slam it again. Remastering for more dynamic range assumes that the original 2-track is available and hasn't been slammed by the mix engineer before being sent to mastering. If that isn't the case, you have to hope that the original multi-tracks are available for re-mixing. I fear there will be a significant amount of current music that will be unrecoverable if the fashion changes back to dynamic music.

 

 

I disagree. I have done enough experiments in reversing what compression does - limiting is, will be, harder though - to see a path through. If smart enough algorithms are used, then the subjective over-cooking can be subtracted from the track, and it will be more than acceptable. No, it won't be a "perfect" undoing, to match precisely what the laid down tracks contained - but it will be damn well good enough ...

[fas42]

2 hours ago, Samuel T Cogley said:

 "HDR audio" is only really convenient to listen to in a very quiet room with expensive gear (unless you enjoy fidgeting with the volume knob constantly).  And really, the obvious source material for "HDR audio" would be classical.  That's grandpa music for sure.

 

 

Nope. "HDR" playback works at all volumes - something that many people may not realise is that competent replay still works beautifully at very low SPLs - the sound never degrades into background filler, but retains its subjective impact and vitality. When you hear a live musical event from quite some distance away the measurable SPLs are very low, but you have no trouble discerning the "liveness" of what you're hearing. And exactly the same behaviour occurs with capable audio reproduction.

[fas42]

1 minute ago, Shadders said:

Hi,

With regards to reversing the compression - is this an issue if the unadulterated/uncompressed recordings are still available ? They should be - as it is just data, on a hard disk somewhere ?

Regards,

Shadders.

 

If "unadulterated/uncompressed" tracks are available then it becomes very easy - if not, iterative smart guessing by software algorithms can untangle what is available. It all depends on whether someone decided to archive the original material in a safe place - which will highly variable.

 

I can see a future trend where albums are released as a set of "raw" tracks of the sound elements - and the consumer can mix them to suit, via some hardware that combines them on the fly when playing - everyone is now their own mastering engineer! Presets for those who like that, complete manual control for others who want total command. Yes, far more data being shipped around and stored, but that's where the future is heading, right now.

[fas42]

Just now, Shadders said:

Hi,

Fas42 indicates that he has reversed compression. I recall that in telecommunications that companding is used to transmit a higher dynamic range signal across a lower dynamic range channel.

It is possible - but i assume in PCM terms, information is lost - not sure of the details.

Regards,

Shadders.

 

If the settings of a simple compression pass are known then the inverse parameters can be fed in - and the information is fully recovered. Technically, tiny, tiny bits of accuracy may be lost if not enough bits are used in the exercise, but this is essentially inaudible.

 

It's harder if the inverse parameters are not available - then careful inspection, and iterative attempts using combinations of likely parameters, and evaluating the various outputs - all done by software algoirthms - will lead to an optimum strategy. Again, it won't be "perfect" - but subjectively it will be close enough - the gains will easily outweigh any losses.

[fas42]

 

Just now, semente said:

 

It might be possible but the results will only satisfy someone with really low expectations.

 

 

Why do you think this? No loss of fine detail occurs - it is the equivalent of having a very, very smart volume control following the waveform as it goes through, and instantaneously amplifying, or attenuating the sound, far faster than you can hear such happening. If done with exactly the right amounts of gain at each point then what you hear is now largely "correct".

[fas42]

9 minutes ago, esldude said:

You don't appear to know the parts of compression.  Nor how compression really is done in recordings.  You have at least the attack, release, ratio and knee of the compressor settings plus any low or high pass settings.  This is about as simple as any of the compression gets. 

 

You have several settings, and no you aren't going to intelligently guess or figure out what those were.  Next is the issue where you can have different amounts and settings of compression covering various bands of the audible band.  Next is all of these may have been done in more than one stage.  You are not going to be able to undo all of that or undo enough of it you are getting anywhere close to reversing the process.  This is true of music going back 50 years, and gets more complex the closer to current day the music was recorded, mixed and mastered.  This ignores that some newer compression devices are FFT based.  Or that they oversample during their application.  

 

Now all of this is ignoring how EQ, or delay or reverb would cloud the matter.  Compression may have been done differently to different parts of the mix before it was put into a stereo track (which means there is no fixing it).  Then mastering will do a bit more and do limiting and you'll never disambiguate all of this.  Not going to happen. 

 

People have this idea that 80 db of dynamic range is squeezed into 20 db.  But it is a process not nearly so linear as most people are picturing.  The various settings give various sounds and are nothing like a straightforward process.  

 

I appreciate all this. I spent quite some weeks playing with the Reaper audio software, which has very good effects modules which allow one to play with all this. And one can manually guess, and get very close just by viewing the waveform - there are visual characteristics that mark the compression imposed, and one uses those as clues. Of course there are going to be complex applications of compression, but that is part of the sound of the track, in itself - you don't want to undo the "art" of the producution, you're trying to restore some sanity to the amount of squashing that's been done. In fact, my experiments started because I had a sampler pop track which was ridiculous in its presentation - surely something can be done about this? And, it turned out so ... I ended up getting pretty close to what a garage band - the group in question - would have sounded like, live.

[fas42]

Just now, esldude said:

Here is a simple video of compression at its very simplest. 

 

We haven't talked about side chaining compression (where one track sort of controls compression of another track) or the use of parallel compression of tracks or any number of techniques commonly in use for decades.   It is nothing like riding a gain knob up and down to go back to the pre-compressed state.  

 

 

Yes, it can get very complicated! But we're not trying to uncomplicate it - we're engineering the sound to be more reasonable to listen to - without introducing audible artifacts.

 

Sorry, have got to head out, right now - won't be around to debate this, for a while ...

[fas42]

5 hours ago, Samuel T Cogley said:

 

While I freely admit my knowledge of these things is something like the high side of layman or the low side of amateur, there is a significant different between compression used on studio tracks, and a "mastering limiter" used on the final product.  The latter is where the evil happens.

 

 

 

 

 

Indeed that can be the case. Limiters are more 'sneaky' in their action - when viewing the waveform at a particular spot everything looks hunky dory, because of the look ahead by the algorithms, to make sure the transitions are smooth.

 

I will most assuredly assert that the developing of a reasonable, mature process for undoing the worst damage wrought on the final mix will not be trivial, straightforward to do. Which is not the same thing as saying that it can't be done, full stop. Again, I don't care one iota whether I've secured a perfect reversal - if 99% of the listeners can't pick it, and are fully happy with the listening, then ... job's done.

[fas42]

4 hours ago, esldude said:

Okay Frank, how about some parallel compression. 

 

I have a track, and make a duplicate track.  I do some compression on the duplicate.  I then mix it back into the original.  This gives me a track with higher loudness, higher average level, but leaves any spiky transients intact without being blunted.  Given the result of that how can you work backwards to get close to the original track?  At the very least this is a non-trivial problem.  

 

Well, we're not trying to do code breaking of the Enigma machine here! Though the process will in part be similar to solving such problems ...

 

At all times one has to keep in mind that various manipulations should be considered to be part of the Art of the track - and if those are reversed then one is most assuredly wiping the oil painting with turps. Again, the point of the exercise is undo the more extreme squashing of the sound, so that it becomes possible to listen to these sort of recordings over a significant time period on a high peformance playback setup, without overloading on the experience. This is the reason I decided to investigate doing such - and I believe given enough time and motivation just about anything could be achieved.

[fas42]

22 minutes ago, Samuel T Cogley said:

 

In practice, I can't see how this works without having to constantly control the volume.  Up for the quiet parts, down for the loud parts.  I'm not against the idea of "HDR audio", but I can't see the practicality of it.  With headphones or IEMs, perhaps.  But not with speakers unless you're in a quiet room with good gear.

 

In practice, very well. "Good gear" is the first priority, quiet room is very far down on the list - put it this way: you have a grand piano in your listening room; a highly accomplished pianist comes in, and commences to play a wide array of pieces in style, from the most subtle dream state conjurings, to almighty, thunderous Chopin or Liszt extravaganzas - is the fact that the room is "quiet" or not have the slightest bearing on the experience?

[fas42]

2 hours ago, Brian Lucey said:

 

False

 

In stereo playback the room is 50% + of the experience.

 

Put your best system in a gymnasium, and get back to us.

 

Put your musicians in the same spot, in the same gymnasium - do we only get 50% of the musical value from them?

[fas42]

Talking of good mastering, this is an album that took my breath away - every song on this is a beauty ... I've often used this for a solid workout when visiting shops, etc.

 

 

[fas42]

Awww, what the heck! ... I'm in the groove, now ...

 

 

[fas42]

2 hours ago, PeterSt said:

 

Ehm, Yes. I testified of that a few months ago, including a recording of it, which I put up. Without having seen the artists, I thought it was a very poor reproduction system I was listening to (it was in a large hotel lobby).

 

Well, we can't really expect reproduction to do better than being 100% accurate to what was recorded. If someone chooses some incredibly unsuitable place for playing it back, then that's their choice  - my pleasure is sensing the energy and vitality of the music making; that the environment it's in causes it to have a less than perfect FR or reverberation properties is not particularly interesting to me - rather, what really bothers me is hearing obvious clues that it's "fake" .

[fas42]

8 hours ago, Samuel T Cogley said:

 

A recording of classical piano?  That's a DR range that's pretty well defined from a recording perspective.  Is that the example of "HDR audio" in this case?

 

We're talking about recordings, right? 

 

Here's a title from my collection.  Track 1 has an Loudness range of 21.1 LRA.  Is this a sufficient example of the DR you're talking about?  Because this exists today and anyone can buy it.

 

And yes, if you listen to this in the car, you're going to have to ride the volume knob.  Unless your car is so quiet you don't have to turn up the volume for the quiet parts.  It would have to be one quiet car.

 

 

 

Yes, recordings. If no compression or fiddling is done then the DR should be true to the acoustic experience that people would have had at the recording site - do we want more dynamic range than the real thing has?

 

Hmmm, you know people who manage to squash a grand piano into a vehicle, and as a bonus they get a pianist who is able to play it as they wander through the traffic? ... I think we started with a "quiet room" - so perhaps we should stick with that basic scenario. IOW, like for like - you listen to the recording in a similar place to where the live experience could/would occur - if "quietness" is a factor.

 

IME I have never felt that the playback needed more "quiet" - what happens is that one is drawn into the performance, and irrelevant, extraneous noises vanish from one's consciousness - the "strength" of the illusion dominates your awareness of what's going on - it's a primary focus of the highest order. As an example, I can have the phone ring while something is playing - and I miss it completely. The sound of the phone is so out of context that my mind dismisses it, the ringing is "invisible".

[fas42]

19 minutes ago, Samuel T Cogley said:

 

I'm not sure we were ever talking about the same thing.  I commented on the notion of "HDR audio".  I assumed this hypothetical new format would have a dynamic range that was higher than the highest dynamic range found in recorded music today.  And in my experience, that would almost certainly be classical music.  I was simply proposing that any DR higher than the highest DR found in currently available content was a non-starter, because it would be so impractical from a playback standpoint.

 

 

My immediate response: what's the point? For any reason, whatsoever ...

 

16/44.1 is perfectly capable of capturing the full range of the most extreme orchestral exercises - I have a recording which starts with an climactic fanfare - if set at the right level, and people were unaware, then a few changes of underwear may be called for - the piece then continues, with ppp passages galore, and no irritating anomalies.

 

Decades ago I did an exercise where I listened to a classical piece attenuated by 60dB, in 16/44.1 format. The volume had to be set on maximum, and I could just barely hear it with my ear against the speakers - and there was the piece, fully intact; very noisy, as if it was raining quite hard - but none of the sense and impact of the piece was lost, at all. So, I was listening to something with 60dB DR, and, it worked - it was impractical because the gain of the amplifying chain needed to be so much greater to even hear it.

[fas42]

15 hours ago, PeterSt said:

 

It was this. Get your plugs :

2017_08_26_09_56_32.wav.mp3

 

 

 

A valuable recording. What one hears are two clear sound elements, the rattle of voices and activity in the lobby, and mixed within, the live performance. One can "hear" that the music is live, because there are no distinguishing markers from, say, a PA system with its typically overcooked sound - the fact that the music is fighting the noise in the area is neither here nor there - what counts is that the sense of realness of the instrument playing is not undermined  by anything - at least to my ears.

[fas42]

1 hour ago, PeterSt said:

You are correct. You can't. I can't either. However, there's theory behind this all that should make one frown. In positive sense that is. This is all related to the announcement (was that from today ? if not then yesterday) : "The master is the master and it can not be improved upon" (and then referring to something which was put out to us ever back). This is plainly wrong and it is already wrong per the existence of this thread : BL. He can destroy everything in your eyes (ears). He could also improve on everything, if he only owned the technique MQA is proposing (but hiding). So MQA is doing right that.

For our comfort we can think of some smart decompressor which @fas42was trying to use which is impossible and can't exist, but which MQA to some extent still developed.

 

Digital recordings are data. And computer algorithms which manipulate data can be very simple in concept or operation, or highly sophisticated, complex in nature - to take things way beyond relatively straightforward manipulation of DR, there is active research in unmixing of audio tracks - that is, extract the sound elements from a mix, and generate a multi-track from the master; having done that, it's open slather on what one can generate, for the listening.

 

To my mind MQA is just another Dynagroove tweak - helps to make the technology of the day sound better because it 'compensates' for typical weaknesses in playback systems - a crutch, IOW.

[fas42]

7 hours ago, PeterSt said:

 

It is a bit too bad that you have this example from a 2L based album, but the explanation normally is that any (random) means of better playback/reproduction implies more "hall". Remember what I said yesterday about the Metallica and "now it is stereo" ? it is an example of such an improvement. In this example it was about less squashed music, but we must realize how much waves in air interact and how the sharper they are the more 3d things become (including a more vibrant stereo image).

And all improvement is about "sharper". Sharper but in a good way and thus easily bearable.

 

Thrash metal and such has always been totally off my radar - it may as well never happened, in terms of my exposure to it. But seeing how Metallica is being focused on, I grabbed an audio segment of Kill Em All from a YouTube clip, of the early 80's mastering - and, must say, excellent stress test for playback. Huge cymbals, reverb, soundstaging, vocals deeply embedded, frantically driving - it's got all the ingredients to bring non-optimum systems to their knees, effortlessly.

 

This is the sort of material I enjoy working with, because the transfomation when a setup gets it right is truly amazing ...

[fas42]

5 hours ago, PeterSt said:

 

Hey Shadders - Everything. The whole chain. Playback software, D/A converter, Dedicated Audio PC (with LPS), Amplifiers, Loudspeakers. Interlinks, USB cables. The only think I can think of what we do not design/make is power cords. Maybe next year.

Playback software explicitly includes Operating System leanness (beyond AudioPhil and WindowX together), OS running completely from RAM (no HDD/SSD) and of course Tidal support and MQA; MQA hardware purposely not for the reason of not being able to apply my own filtering easily - but I have it for myself.

 

This is the approach that works. I always, always, always think of the system as a whole as being the beast that has to be brought under control - any part of it not pulling its weight correctly is what has to be addressed, and doing that yields the worthwhile returns.

 

I am truly amazed at the amount of talk, discussion being expended on MQA - this is worth, maybe, 0.01% of what's worth doing to get optimum sound.

[fas42]

12 minutes ago, Cornan said:

 

As far as I see it the same album with FLAC with MQA is a step-up in SQ with more HF revealed. Not a lot, but still worth it IMHO. It is similar to a bad recording on FLAC to a good one IMO.

 

The "more HF revealed" type of thing really gets me - audiophile systems generically do their utmost to dump, discard the real HF content - the giveaway is that cymbals typically sound blahh!! on them. Then they try and "recover" what they've just thrown out ...

 

The pattern is, when more HF is heard, it's obviously distorted, and sounds unpleasant. So, just cut it off, with a pair of scissors - so much nicer now!! But ... the playback now lacks life, sparkle - what do I do to recover that ... and round and round and round we go - and never are the real issues addressed ...

[fas42]

All this fuss for "music content" above 20KHz, that the mics may have picked up . Having done simple analysis of some 2L recordings, I have yet to find a single spot in a recording, any recording, where information above 20k has any significance, that could possibly be audible.

 

If someone could point me to a recording, apart from any silly "prove a point!" efforts, where it may have some relevance - I would be interested in knowing.