Hi Amir,
below are my questions... Hi Amirm,
Put another way, as a scientist, are you 100% satisfied that we know all the right measurements that specifically measure the elements from sound reproduction that are particularly important to human hearing (as opposed to mechanical hearing)? I suspect they are different in the way that human vision is clearly different than mechanical vision, as evidenced by the myriad different optical illusions that fool human eyes (and would almost never fool a camera or mechanical measurement tool).
View attachment 26630
Hi Lloyd. First an apology for not answering you earlier. Your post came at a heated moment in this thread and throttled back some to keep it from boiling over.
Answering your question, yes, there are 70 years of research into human perception of sound. The field as I am sure you have heard is psychoacoustics. You can get a degree in it and it is a far more well researched field in audio than all the others. The reason is that there is strong interest from other industry such as Labor and Industry (L&I) related to worker hearing loss, of course medical field to diagnose disease and hearing loss again, in education with respect to having student comprehension improve in classrooms, and of course audio.
As with many other sciences there is a bible and that is the book by Dr. Fastl and Zwicker, called Psychoacoustics: Facts and Models:
http://www.amazon.com/Psychoacousti...stl zwicker&qid=1461465921&ref_=sr_1_1&sr=8-1
I will quote the bios from Amazon so that you see these are serious folks:
"Hugo Fastl is Professor of Technical Acoustics in the Faculty of Electrical Engineering and Information Technology at the Technical University München, Germany. He graduated 1969 in Music from the Academy of Music München, and 1970, 1974, and 1981 he earned at the Technical University München the degrees of Dipl.-Ing., Dr.-Ing., and Dr.-Ing. habil., respectively. His research interests are basic psychoacoustics and its applications in fields like audio-communication, noise control, sound quality design, audiology, or music. In 1987 he was elected Guest Professor of Osaka University, Japan, and in 1990 he became a Fellow of the Acoustical Society of America. He is head of the committee "Auditory Acoustics" of the Society for Information Technology (ITG), and with the German Acoustical Society (DEGA) he is member of the Board of Directors, and Treasurer. In 1983 he won the Award of the Society for Information Technology (ITG), in 1991 the Research Award in Audiology of the Forschungsgemeinschaft Deutscher Hörgeräte-Akustiker, in 1998 the Research Award of the Japan Society for the Promotion of Science (JSPS), and in 2003 the Rayleigh Medal of the Institute of Acoustics (UK).
Eberhard Zwicker was one of the worlds top authorities in psychoacoustics. In his labs in Stuttgart and München he educated scientists and engineers who hold now key positions in basics and applications of (psycho-)acoustics. From his many honors, the Silver Medal of the Acoustical Society of America and the degree of Honorary Member of the Audio Engineering Society are among the more outstanding."
If you read the book (not saying this literally) you see that they have tested human perception from every which way. This is NOT the mechanics of hearing but conducting listening tests for thresholds of detection for everything under the sun.
On top of this book are hundreds if not thousands of research papers published at Journal of Acoustic Society of America (ASA), Journal of AES, Acta Acustica/Acustica, IEEE Spectrum, and probably a number of medical journals I don't read.
Now, your reaction may be that you are not going to and buy and read that paper so maybe I am just bluffing. Turns out I have literally billions of proof points that we know incredible amount about the perception of sound and music by humans. And that you can test this right now for yourself. Namely
lossy audio compression.
Every lossy audio compressor such as MP3, WMA, AAC, Ogg have a perceptual model of human hearing system. They attempt to compress/reduce the bit rate in a way that is least audible.
Take 128 kbps MP3. CD data rate is 1,400 kbps. That means 128 kbps represents just 9% of the original size of the music file. Despite being decades old at 128 kbps MP3 fools huge swath of the population into thinking nothing is taken out. Let me repeat: 91% of the original size is gone, and every PCM sample reproduced by MP3 is different, yet vast majority of people cannot tell.
Why is that? Because our hearing at the same time is very precise and very deaf. In a rock concert that is playing at high SPL, you can scream and the person 10 foot away can't hear you. The louder sound of the concert will mask that. Given those two combination of sounds, i.e. the music at high SPL and you screaming at much lower SPL, we can throw out your scream and the result would be the same.
I know what you are going to say. "But MP3 is not hifi" First that is not the point. You asked if we know how our hearing works and we obviously do or we wouldn't be able to fool almost everyone into thinking 9% of the music is as good as all of it.
But let's go with the objection. Go ahead and compress your music using variable bit rate (VBR) with quality set to 10 or 100. In this mode, you are telling the encoder to only compress music to the limits of audibility but no more. Should more bits be required, the encoder will use it. If not, it will shrink the data rate. This mode on the average will give 2 to 3 times compression ratio. So a typical song will be at 500 to 600 kbits/sec depending on the codec. In that sense, we are still throwing out half or more of the original bits.
When I was at Microsoft, we conducted a large scale test of our audiophile community with 24 bit/96 Khz compressed this way to about 700 kbits/sec. All the audiophiles failed to hear the difference between the original and that much reduced bit rate. I took the test and out of the half a dozen track, I could barely tell the difference in one clip. That issue was looked at and resolved.
Even stepping down to fixed bit rate of 320 kbps, almost every audiophile will fail to tell the difference between it and the original file even though 75% of the data is thrown away and the encoder knows that at times it has exceeded the threshold of hearing. It gets away with that because as a rule, despite our constant bragging on forums, audiophiles seem to have no special ability to hear these dynamic distortions.
I am able to pass such tests frequently but they require strong effort on my part. At AXPONA I walked into a room that was playing delightful music. As soon as I sat down the presenter said, "by the way, this is a 256 kbps MP3 that I bought from a music store!" That track sounded better than vast majority of LP playback I heard at the show!
This is a complicated field and for that reason, there is little to no exposure or understanding of it among the audiophiles. As such, people just assume that we don't know this domain but we absolutely do. Show me a distortion graph and we can overlay psychoacoustics on it and determine the likelihood of audibility. Take total harmonic distortion. By itself, the one number is useless. But give the spectrum and we can use masking thresholds and determine if that distortion is audible or not.
Unfortunately there is no instrument that pops out audibility of distortions. I complained to Audio Precision for not having overlays for distortions that do that. I should not have to manually do it. Yes, it won't be comprehensive but a lot can be done to add intelligence. Until then, you just need to rely on experts in the fields.
So take comfort that we know this science. While there is further research, answers to whether two DACs sound different to humans, etc. are all very well known. Those answers make us upset as they go against the experiences we have. But they are the answers. And there are explanations as to why our experiences are in conflict with them.
