Originally posted by john33
Beatles

If bitrate is of no concern to you why are you looking at lossy as opposed to lossless formats? Or, is it just an academic exercise to evaluate the current state of the art?

john33

Originally posted by JohnV

Interesting. Btw what kind of commandlines you used for MPC, I guess you also disabled mid/side-stereo? I could give you some hints for very high bitrate/quality commandline.
I don't think there's much use to go over 600-700kbps because then lossless coding becomes a viable choise. 

Yeah, I wouldn't save masters in lossy format either...

Originally posted by Dibrom
Beatles,

While it is possible that you hear the differences that you describe, and that you say -insane with MPC is "not good enough" so to speak (I assume you mean non-transparent), of course people are going to want abx test results and specific clips to back this up.  Just as VeryBlur's comment about "lifelessness".

No it is not "possible" it is what I hear. Now it sounds fine on my computer with Klipsch speakers but on my home rig which is let's just say insanely expensive the shortcomings can be heard. And I KNOW what they're supposed to sound like because I did the majority of the sessions and I know what was put on tape...I did it.
For the shortcomings I find please see my other post. I think transparency is an overused term here.


In general I tend to find these sorts of comments usually do not lead to anything absolute or objective because they are never proven, and often the person making the claims has no desire from the beginning to make any attempt to even try to prove them.  However, if this is not the case with what you are testing, then it would be a nice surprise...


Firstly I have not made any concrete statements just very preliminary findings which will be more fully tested. And should I come up with what I think I will I'll be more than pleased to share even though you and I listen to decidedly different music. My testing is based exclusively on acoustic instruments recorded in real space.

About your MPC lines, if you find that mid-side stereo and temporal masking are failing, I would like very specific examples of this.

This I will do. I am as yet unsure about the temporal masking but the mid/side I have no doubts about. Further testing is required.

Forgive me for my skepticism, but I do find it a bit hard to believe that even being an audio engineer, you would find a format so non-transparent... one that has been tuned so highly by people with some of the most sensitive ears around all ready..


Skepticism is healthy. My ears are how I earn my living and have never let me down yet.


So again, as with VeryBlur, I'd ask for both abx results and test clips for all claims regarding something not being adequate, especially in regards to MPC.  Anything is possible, but without this information it does the discussion little good IMO.

Let me get further on and I will send samples as I suspect the music I'm using will be a little hard to find. MPC is quite a good format yards beyong MP3, of this there is no doubt but AAC has tremendous potential and I am focusing moreso on it right now.

Oh, and yes the IRC channel is a good place to discuss things.

Look forward to seeing you there.

No it is not "possible" it is what I hear.

Now it sounds fine on my computer with Klipsch speakers but on my home rig which is let's just say insanely expensive the shortcomings can be heard.

For the shortcomings I find please see my other post. I think transparency is an overused term here.

Firstly I have not made any concrete statements just very preliminary findings which will be more fully tested.

And should I come up with what I think I will I'll be more than pleased to share even though you and I listen to decidedly different music.

My testing is based exclusively on acoustic instruments recorded in real space.

This I will do. I am as yet unsure about the temporal masking but the mid/side I have no doubts about. Further testing is required.

MPC is quite a good format yards beyong MP3, of this there is no doubt but AAC has tremendous potential and I am focusing moreso on it right now.

Look forward to seeing you there.

Originally posted by Beatles
One more I can't recommend highly enough to give your system an all around great workout is Sally Oldfield- Water Bearer preferrably vinyl or the domestic CD. The Japanese mastering is nothing short of music killing.

Originally posted by Volcano
Hi Beatles,

this is great, I'm enjoying your posts very much! A recording engineer has always been missing on these boards


Hmm, I just dug that out of my CD collection - it sounds quite ordinary to me, but maybe I have got the wrong recording (it's from the "The Sun In My Eyes" Best Of album, released in Germany in 1997 by Delta Music - never heard of them).

I like very much for "richness" of sound (and therefore for testing codecs or speakers): Maggie Reilly's "Follow The Midnight Sun" ("commercial pop" with hardly intelligent lyrics, but IMO great music) and Dire Straits' "On Every Street". If you could get your hands on either of them (especially the first one), could you tell me if you would also consider these good for testing?

(If this request is maybe a bit cheeky and over the top, then sorry )

I'll certainly try to get hold of those other 3 songs that you mentioned.

Thanks again for your interesting posts!

Originally posted by JohnV
Beatles, I would be insterested what do you think about Ogg Vorbis GT2 350kbps (RC2 Garf tuned 2).

http://sjeng.sourceforge.net/ftp/vorbis/oggdropgt2.exe

Originally posted by Beatles

This depends on what system you are A-Bing through. If it's incapable of proper resolution or producing a credible soundstage then people will of course say there's no difference at a much lower level of resolution....doesn't mean it wouldn't be heard on a better system.  The threshold tends to be limited by the resolving power of the sound chain. As to feel the points I make above are VERY valid in music reproduction.

Yes I hear flaws and remember I know INTIMATELY what was layed down at the sessions I engineered. I know every sound on those sessions...sometimes too well.

No I don't wish to get into that...I've done many electronic sessions and then spent many hours recreating an ARTIFICIAL soundstage. Electronic instruments certainly have "sounds" that must be reproduced accurately. It CANNOT be used however for imaging or soundstage testing.

I have not stated a preference for one format over the other yet...you might be surprised what a high resolution system might reveal. As of this point I'd say AAC has much POTENTIAL.

Originally posted by 2Bdecided
john33,

While some of us accept that lossy codecs aren't 100% transparent, others claim that mpc is exactly this.

We'll ignore the torture clips which are artificially generated, since nothing lossy can code these (as yet), so let's just talk about "real" music (whatever that may be). In fact let's narrow it down even further - let's limit ourselves to sounds than can be captured by a microphone in an environment where no electrically powered things are producing sound - i.e. a purely acoustic event, captured by two microphones - this is as basic as recording gets!






This is the only way to properly discern the damaging effects of lossy encoding in my opinion. The only criterion MUST be real instruments in real space. It is truly the "absolute sound".








Dibrom believes that mpc can code such a recording transparently - that no one will be able to detect the effect of mpc encoding/decoding reliably in a blind ABX or ABC test. Am I right, or am I putting words into your mouth Dibrom?

Beatles clearly disagrees - he hears a difference. His recordings are probably not so basic as I've described, but they're not so artificial or special or whatever that they're codec killers. They're just acoustic recordings, and mpc audibly changes them.







I do have some recordings of acoustic performances in churches and halls with exceptional sound but due to copyright and ethics I cannot share them. Yes MPC audibly changes these recordings. As I've said to Dibrom there is obvious soundstage truncation and some changes in fundamentals ie. the woody sound of a properly recorded cello.




This topic fascinates me, because I have had a hunch for several years now that people such as Beatles are right - but I've never found one who can prove that they are right. The reason I think that there are people who can detect mpc encoding is because I know of people who can detect when a recording has been re-dithered. That's a change to the last bit. mpc (and every other lossy codec) is doing a LOT more than this, so surely someone must be able to detect it? Even in the absence of (what we conventionally call) artefacts.






It depends what the definition of artifacts is.........









Beatles,

So, do you have the patience (and time!) to do a blind test? Can you try to burn the original .wav back to a CD-R with the decoded mpc version as well. (Make sure you didn't use -scale on the encoder, because that would ruin things.) You just need the two files on your hard drive, but you need to create maybe 9 or 12 tracks on your CD-R, in groups of three.

In each group of three tracks, you include the mpc version once, and the original version twice. You (or preferably someone else!) randomise the position of the mpc version within the three.

Your task: play the CD-R, and pick the odd one out from each group of three. Give your answers to the person who radnmoised the tracks in the first place, and collect your score.

If you get them all correct, Dibrom will eat his hat (now I'm putting words into your mouth!). If you find the differences suddenly vanish, and you can't hear any difference between the three tracks, then you've been victim to the placeabo effect.


I hope you can find the time to do this test. The results will be fascinating.









I do intend to do further testing but not for the purpose of making Dibrom eat his words...I have found him to be quite agreeable even if we do disagree..........

http://www.David.Robinson.org/

Originally posted by Dibrom


One thing to keep in mind, is that many of the people who have been advocating and testing various codecs, and whom use "transparent" to describe the sound, usually have tested on at least a somewhat decent system, but what is more important is that they usually have excellent hearing and certainly have a feel for the sound and how it will behave across  many situations (both in encoding and the resulting quality).

Listening on a higher end system may be more revealing, but I don't believe that it will be so much so that the sound would go from being transparent, to completely non-transparent (as in only decent now).



Here's my point. The main differences I hear are in soundstaging and in certain fundamentals. Now if the system which is being used for auditioning is incapable of rendering the soundstage or the fundamentals properly then you won't be able to hear a difference because the system is the limiting factor ergo it will sound perfectly fine.





So is this something that anyone else here is going to be able to reproduce at all?  I'm just curious.  The general tone of this discussion seems to be moving into the realm where proving any of this is going to be difficult to near impossible.  First you have the extraordinarily expensive equipment factor (which the majority wont), and second a recording that you seem to know very well, but I assume others won't.  Add that to the fact that you are talking about encoding MPC at 500kbps and it just seems like this will be a case of "I can hear it, even if you can't" and that you will be the only one able to say that.  Again I could be wrong.. but the fact that I don't know anyone who seriously encodes MPC to 500kbps really makes me wonder...






I t can be heard on my real world system as well as recordings I didn't do but am intimately familiar with after years of using them for auditioning purposes. Even at 500BR there are differences. Such is the nature of lossy compression.








I agree with this partially, but not entirely.  Yes, acoustic instruments recorded with natural reverb and the like will create nuances that synthesized music may not have but the issue is more whether or not a codec will reproduce the soundstage, whatever it may be, period.  This includes an artificial soundstage or any imaging effects created from synthesized music.  In short, what applies to one should apply to the other though perhaps not in the same obvious manner.  I also believe that what you are listening for in the soundstage plays just as much of a role.  There certainly is a different effect in this regard from non-synthetic music. 







While I feel that acoustic instruments in a natural setting is far more appropriate your point is well taken. As well please see my above comments regarding the auditioning equipment.








So basically, the temporal resolution matter is not going to change, and if you favor that (as would seem the case by further increasing this by turning off temporal masking, and discussing blurring of the sounds and decays.. both of which are temporal matters), then I cannot see how you might come to the conclusion that AAC was as good or possibly better.



Turning off temporal masking was merely part of the testing procedure. That happened to be the last command line I had used.  At this point I prefer to listen rather than possibly be swayed by any technical arguments. Prefer to remain as open as possible.













I realize you haven't made your decision yet at any rate, and I'm also not implying that AAC isn't also a very fine format, because it definitely is, I'm just stating some points.

Originally posted by 2Bdecided
Don't get me wrong - the purpose isn't to get Dibrom to eat his words. My reason for asking you is to satisfy my own curiosity!

I hope your planned tests are blind.

Cheers,
David.
P.S. I assume you know much more about copyright than me, but it appears that you can distribute a 30 second extract of any recording for educational use under the 10% rule. 30 seconds may not be enough for us to soak up the atmosphere and acoustics, but it might help.

Originally posted by Volcano
Beatles,

could you please use the quotes properly? It is very hard to read you replies with quotes, because one can't distinguish what you say from what the person you are replying to said.

It's easy (note that the < and > have to be replaced with square brackets, [ and ]):

<quote>The person you are replying to said this.</quote>
This is what you are replying.

You must not include your reply within the <quote>...</quote>, if you want to quote multiple passages, you have to use the <quote>...</quote> manually.

Sorry for being off topic.

CU

Dominic

Originally posted by 2Bdecided
While some of us accept that lossy codecs aren't 100% transparent, others claim that mpc is exactly this.

We'll ignore the torture clips which are artificially generated, since nothing lossy can code these (as yet), so let's just talk about "real" music (whatever that may be).

Dibrom believes that mpc can code such a recording transparently - that no one will be able to detect the effect of mpc encoding/decoding reliably in a blind ABX or ABC test. Am I right, or am I putting words into your mouth Dibrom?

Beatles clearly disagrees - he hears a difference.

They're just acoustic recordings, and mpc audibly changes them.

That's a change to the last bit. mpc (and every other lossy codec) is doing a LOT more than this, so surely someone must be able to detect it? Even in the absence of (what we conventionally call) artefacts.

If you get them all correct, Dibrom will eat his hat (now I'm putting words into your mouth!).

This is a quote

This is a quote

Beatles clearly disagrees - he hears a difference.
--------------------------------------------------------------------------------
Correction. He claims he hears a difference. 

Originally posted by 2Bdecided
I wrote it like that to give Beatles the benefit of the doubt. He's much more likely to come back and give us the results of any test he does if we approach him like that, rather that saying "you're probably wrong, but go and prove it anyway".