Announcement

Collapse

HUG - here for all audio enthusiasts

Since its inception ten years ago, the Harbeth User Group's ambition has been to create a lasting knowledge archive. Knowledge is based on facts and observations. Knowledge is timeless. Knowledge is human independent and replicatable. However, we live in new world where thanks to social media, 'facts' have become flexible and personal. HUG operates in that real world.

HUG has two approaches to contributor's Posts. If you have, like us, a scientific mind and are curious about how the ear works, how it can lead us to make the right - and wrong - decisions, and about the technical ins and outs of audio equipment, how it's designed and what choices the designer makes, then the factual area of HUG is for you. The objective methods of comparing audio equipment under controlled conditions has been thoroughly examined here on HUG and elsewhere and can be easily understood and tried with negligible technical knowledge.

Alternatively, if you just like chatting about audio and subjectivity rules for you, then the Subjective Soundings sub-forum is you. If upon examination we think that Posts are better suited to one sub-forum than than the other, they will be redirected during Moderation, which is applied throughout the site.

Questions and Posts about, for example, 'does amplifier A sounds better than amplifier B' or 'which speaker stands or cables are best' are suitable for the Subjective Soundings area.

The Moderators' decision is final in all matters regarding what appears here. That said, very few Posts are rejected. HUG Moderation individually spell and layout checks Posts for clarity but due to the workload, Posts in the Subjective Soundings area, from Oct. 2016 will not be. We regret that but we are unable to accept Posts that present what we consider to be free advertising for products that Harbeth does not make.

That's it! Enjoy!

{Updated Nov. 2016A}
See more
See less

Our brains do not need High Resolution! (UNPROVEN ... READ ON)

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Mono?

    Ummm. Found the buried audio samples eventually.

    But look at the attached. The orchestral recording was mono! 'I think the classical recording was an old analogue one'. So it has limited resolution to start with, both dynamically and in frequency content compared with a digital recording.

    That has a huge influence on the compressability of the audio. Frankly, if the other (pop) tracks are stereo and the classical track is mono, the entire experiment is unreliable and any conclusions drawn from it meaningless.

    The devil is always in the detail.
    Attached Files
    Alan A. Shaw
    Designer, owner
    Harbeth Audio UK

    Comment


    • #17
      Meaningless conclusions from poor choice of audio clips

      I must say now that I've downloaded the three clips (Beethoven, Kylie, Idol) that I'm disappointed that this thread was started (and passed through Moderation.)

      The Beethoven track is quite appalling. It sounds like it is a cassette recorded off a 78. It has no detail, lots of constant hiss (which masks low level detail) and there is absolutely nothing musical between the notes. If you were hunting for an example of an obsolete classical recording which typifies state of the art recording in the 1930s, this is it. Nobody here can glean anything useful from the classical track. Nothing.

      In future, can we please make an effort to check the source audio before we wind readers up with a dramatic new thread? Better still, take the concept of some theory and construct our own tests, within the defined limits of our ability and test equipment, and lay them out for others to comment on. I've even said this here before. The bottom line is that there are few if any new miraculous discoveries in audiology - and we should use our common sense before falling for wild claims.

      I will Flash encode these clips and put them here for you to decide later. The classical track is complete garbage.

      Golden rule: Always ask yourself what possible motive has the writer for making the claims he does.
      Alan A. Shaw
      Designer, owner
      Harbeth Audio UK

      Comment


      • #18
        Sample audio quality

        I've downloaded and listened to the Beethoven and the Kylie and Alan is right.

        But I'd also like to suggest that the writer is unlikely to have any competing motives for his study. He works in a non-music or electronics area (livestock industries). He declares absence of other motives in the paper.

        But back on topic, yes it's true that any conclusion based on these samples is invalid. Pity.
        Ben from UK. Harbeth Super HL5 owner.

        Comment


        • #19
          Nothing new to discuss?

          Originally posted by A.S. View Post
          I must say now that I've downloaded the three clips (Beethoven, Kylie, Idol) that I'm disappointed that this thread was started (and passed through Moderation.)..The bottom line is that there are few if any new miraculous discoveries in audiology - and we should use our common sense before falling for wild claims.....
          Golden rule: Always ask yourself what possible motive has the writer for making the claims he does.[/I]
          I am disappointed too. It is clear there is nothing new a member can contribute for discussion in Harbeth forum.

          ST

          {Moderator's comment: you can take that attitude if you like, but that's misrepresenting the situation. Threads that generate needless churn just waste everyones time.}

          Comment


          • #20
            The audio clips used in the Report

            Here they are. I present them as-is.

            The 'reference' classical music track - an acoustic disaster. This should have been disallowed on grounds of totally unacceptable audio quality:



            Kylie 'the pop track':



            Billy Idol 'the rock track':



            It's blindingly obvious to me that the frequency range, tempo, loudness, spectral energy, mono/stereo width and dynamic range of these recordings are wildly different. Not only that, but the classical track has a persistent flutter (pitch instability) throughout which would again have an impact on comressability. This makes drawing valid conclusions about data reduction impossible. To use the classical track with its appalling audio quality as the 'control' is daft.
            Alan A. Shaw
            Designer, owner
            Harbeth Audio UK

            Comment


            • #21
              The challenge is more enjoyment, not fewer bits.

              Originally posted by STHLS5 View Post
              I am disappointed too. It is clear there is nothing new a member can contribute for discussion in Harbeth forum. ST
              No hard feelings, but few people 7 to 8 years ago could foresee that storage costs could fall so dramatically and that technology could augment data "transport" needs through network storage, wireless and better faster internet access. Personally think the challenge today is not about getting better sound, but how to facilitate the enjoyment of music. It’s less so about amps, cd players etc

              Comment


              • #22
                A technical look at the examples presented

                This should have been done by the author of the research paper. It took me about 15 minutes with standard software (Adobe Audition) and completely confirms what common sense and our own ears tell us: the classical music track is radically different in its energy and spectral content to the pop track. Consequently the entire theory proposed is built on unsafe foundations, in my opinion.

                Attached is the technical review of the classical (Beethoven) and pop (Kylie) tracks as supplied. I hope I've annotated the images to make sense of them. As you may be able to see, as far as this Beethoven example is concerned, it is clearly mono as the L & R traces (top chart) are identical. As the audition suggested, in this classical example, there is almost no energy above about 7kHz. Contast that with the Kylie which shows high level energy (red streaks) right out to the CD cut-off at 22kHz.

                All data compression systems have to walk a fine line between discarding sound to reduce bit-rate and the consequent potential loss of fidelity. These two examples would stress a coding system in very different ways: the classical piece would compress easily as it has limited loudness and limited frequency range: the pop track would be more difficult to compress because there is so much high-level energy at high frequencies ... a nightmare scenario for a compressor because to trace the high frequency waveform accurately needs more data, and that is the very thing that a compression system doesn't want to have to do!

                NOTE: This does not imply that all classical music or all Beethoven recordings have these characteristics - we have only looked at one old mono recording, the one the author used.

                >
                Attached Files
                Alan A. Shaw
                Designer, owner
                Harbeth Audio UK

                Comment


                • #23
                  Designing a valid test

                  It appears that the author may have invalidated his own hypothesis (i.e. rendered it incapable of meaningful proof) but offering poorly-chosen samples.

                  But it does beg an interesting question. How would one choose valid samples to test whether there was a difference in compressibility between pop and classical music? Both recordings should be stereo, obviously, and both should be considered to be "good quality" recordings. Beyond that, however, what would the criteria be? With a small sample size, could not any differences (assuming there were any) be ascribed to the spectral and other differences between individual recordings? Would not one have to use a fairly large number of recordings before drawing any meaningful conclusions, even assuming other variables could be equalized?

                  Comment


                  • #24
                    "Hypothesis" means still unproven! Peer review failure ....

                    Originally posted by kittykat View Post
                    No hard feelings,...
                    I have no hard feelings but bemused by some of the comments here. Surely, all of us are busy with no time to spare if we choose to be so.

                    A cursory reading would clearly show that Robin Whittle was aware of the Beethoven audio clip was in mono. On further readings, you will find he did comparisons with all tracks in mono. He also mentioned that the original audio files were missing. Whatever it is, if one pauses for a moment and wonders why would one take Whittle statics to form a basis of Dr Hudson's hypothesis, then I wouldn't dismiss it outright.

                    Whittle's compression data was the basis for another thesis by Tatung University (yes. It is ranked 1450 in the world) and also used by Nanyang University (yes, again not the best in the world). Probably there are more since this is also part of Wikipedia but I have no time to search and knowing this post is very unlikely to pass the mod.

                    And how could BioMedCenter approve unsound research papers, which were peers reviewed? So we have statistics published in the 90s used by Universities and researchers like Hudson in 2010 but all of them somehow overlooked the mono track which came under scrutiny within half an hour of me posting the link here! Amazing!

                    For a truly educational forum as one aspires to be then the least we can do is at least honour Hudson's request, i.e. I quote "As with all generalisations, a frank discussion of the presence of both supporting examples and counter-examples will illuminate where and why the musical information compression hypothesis breaks down".

                    This article is not as much misleading when comparing the Hifi News article on Festival of sounds, which only received a mere 22 response out of the 3000 attended and yet a conclusion was drawn that all couldn't tell the difference between live and recorded.

                    ST

                    Comment


                    • #25
                      Forum member confirms compression ratio

                      Originally posted by EricW View Post
                      ...I have noticed the same phenomenon using iTunes and Apple Lossless compression. Starting from the fixed CD data rate of 1,411 kbps, I often find that pop or rock music often compresses down to file sizes of only 800 to 1,000 kpbs, while classical music (especially with simpler instrumentation, e.g. solo piano recordings) often yields much smaller file sizes. It's an interesting phenomenom, but I suspect it's more a mathematical issue than a perceptual one.
                      Originally posted by EricW View Post
                      It appears that the author may have invalidated his own hypothesis (i.e. rendered it incapable of meaningful proof) but offering poorly-chosen samples......?
                      In your own words classical music compresses in to smaller files than pop and rock. Isn't that good enough? Whittle spent many years to study about compression and if you go throughout the website you will find xls sheets and many other lossless classical music comparison data.

                      ST

                      Comment


                      • #26
                        The beauty of old fashioned (BBC) research for the common man ...

                        I think the issue here is not so much the compressability of one type of music versus another but the rather romantic hypothesis associated with the research. To quote the paper ...

                        The adaptive solution to this problem of scale is information compression, thought to have evolved to better handle, interpret and store sensory data. In modern humans highly sophisticated information compression is clearly manifest in philosophical, mathematical and scientific insights. For example, the Laws of Physics explain apparently complex observations with simple rules. Deep cognitive insights are reported as intrinsically satisfying, implying that at some point in evolution, the practice of successful information compression became linked to the physiological reward system. I hypothesise that the establishment of this “compression and pleasure” connection paved the way for musical appreciation, which subsequently became free (perhaps even inevitable) to emerge once audio compression had become intrinsically pleasurable in its own right.
                        etc. etc. etc.. Contrast that with the opening Summary of the BBC report ... (attached). A very different style of writing indeed.

                        It reminds me of the beauty and simplicity of the BBC R&D papers from thirty years ago - attached one covering the spectral content of pop v. classical music. No romantic talk there. Just hard facts. Had I not fallen under the spell of the sublimely deceptive simplicity of these BBC papers when I was a young man, we, Harbeth, wouldn't be here today. And you wouldn't be enjoying the fruits of that BBC legacy.

                        All that's required to really get to the heart of any published paper (assuming it's not smothered in obscure mathematics) is to read it from the conclusion backwards - and apply a really big dollop of good old fashioned common sense asking yourself "does that agree with my own personal experience?" If it doesn't, and you trust your ears, then you need to hunt for clues as to why the reporter - or you - are wrong.
                        Attached Files
                        Alan A. Shaw
                        Designer, owner
                        Harbeth Audio UK

                        Comment

                        Working...
                        X