PEQdB Diamond Target Curve - New Stanford white paper

[RogerSmith]

He is also uploaded few hours ago a video with the title "Responding to AudioScienceReview" where he responds to this thread, of course doesn't make sense to waste any second on it.

One of his problems is that like usually with people shouting loud he misses even the basics, for example 9 days ago he uploade a video with the title "Speaker Measurements Don't Matter". There he argues that stereo playback creates acoustic comb filtering (interference dips in frequency response) because sound from the right speaker hits the left ear slightly later than sound from the left speaker. He concludes that because of this interference, speaker measurements are pointless and mono listening is the only true "high-res" experience. This argument is fundamentally flawed for two primary reasons related to human hearing. The video treats human hearing like a single static measurement microphone. In reality, the human brain and auditory system utilize binaural hearing. Our brains are highly adapted to process acoustic differences between ears to localize sound in 3D space (known as Head-Related Transfer Functions, or HRTF). The human auditory system does not perceive acoustic comb filtering from stereo speakers as a distorted, low-fidelity mess. Instead, our brain "smooths" these comb-filtering notches out psychoacoustically because the two ears sample the sound field at two different points in space. This prevents us from hearing the interference as a tonal coloration, allowing us to instead perceive a spacious stereo image (the phantom center and lateral soundstage). Therefore, speaker measurements (like on-axis response and directivity) absolutely do matter because a speaker with smooth directivity ensures that the reflections bouncing around the room match the direct sound, allowing our binaural hearing system to accurately separate the speakers' true tonality from the room's acoustics.

This pattern unfortunately is been repeated on many of his bold claims which tend to impress people who are new in the topics.

True, but it is still a fact, as even confirmed by Toole himself, that preference differences that were determined in mono, although prevailing, are much less pronounced in stereo. See for example here: https://www.reddit.com/media?url=ht...bp&s=c98a7823878c244e6bafb7b8ca97ba26e10c36a8

 

[totti1965]

And before you give him credit for filling in the hole. He stole that from the endgame target that did well in shootouts
View attachment 543484

What??? You obviously didn't realize that the "HiFi Endgame" target was also developed by peqdb/Sharur in an online shootout.The peqdb HiFi Endgame target curve was an intermediate stage.

I can certainly understand the criticism (also @amirm´s criticism) that Sharur essentially made only a relatively small improvement over the Harman 2019 target curve.
However, while Sean Olive's test group for the Harman IEM target curve consisted of only about 14 people (albeit "trained listeners"), Sharur's study placed the whole thing on a much stronger statistical foundation with around 400 participants.
Fun fact: He also ensured the "trained listener" aspect by giving more weight to results from users who were able to reasonably recognize their own previously selected preferences in randomized, blind tests!
Why would Sharur fake his results? It was to be expected that there would be an evolution—however slight—compared to Harman 2019,
if only due to the control group being 30 times larger!!!

 

[chinsu]

What??? You obviously didn't realize that the "HiFi Endgame" target was also developed by peqdb/Sharur in an online shootout.The peqdb HiFi Endgame target curve was an intermediate stage.

I can certainly understand the criticism (also @amirm´s criticism) that Sharur essentially made only a relatively small improvement over the Harman 2019 target curve.
However, while Sean Olive's test group for the Harman IEM target curve consisted of only about 14 people (albeit "trained listeners"), Sharur's study placed the whole thing on a much stronger statistical foundation with around 400 participants.
Fun fact: He also ensured the "trained listener" aspect by giving more weight to results from users who were able to reasonably recognize their own previously selected preferences in randomized, blind tests!
Why would Sharur fake his results? It was to be expected that there would be an evolution—however slight—compared to Harman 2019,
if only due to the control group being 30 times larger!!!

He was a Harman trained listener too.

 

[chinsu]

What??? You obviously didn't realize that the "HiFi Endgame" target was also developed by peqdb/Sharur in an online shootout.The peqdb HiFi Endgame target curve was an intermediate stage.

I can certainly understand the criticism (also @amirm´s criticism) that Sharur essentially made only a relatively small improvement over the Harman 2019 target curve.
However, while Sean Olive's test group for the Harman IEM target curve consisted of only about 14 people (albeit "trained listeners"), Sharur's study placed the whole thing on a much stronger statistical foundation with around 400 participants.
Fun fact: He also ensured the "trained listener" aspect by giving more weight to results from users who were able to reasonably recognize their own previously selected preferences in randomized, blind tests!
Why would Sharur fake his results? It was to be expected that there would be an evolution—however slight—compared to Harman 2019,
if only due to the control group being 30 times larger!!!

Even Listener (Headphoneshow) admitted that the JM-1 target lacked data from listener surveys. Sharur is right on this point, and unlike JM-1, his Diamond target has been verified by listeners.

 

[Mocie]

spouting off death threats and disgusting racist outbursts is rather more than mean words, and then to double down and laugh about says quite enough

What does any of this have to do with audio science? You may be shocked to learn the people behind most of the scientific innovation in the USA aren't particularly great people either, are you suggesting they should have abandoned rocket research for similar reasons?

 

[Mocie]

We can debate Sharur’s ethics, but I think we should also recognize that he is trying to approach audio from a scientific perspective.

Pretty much. I'm not saying you should be giving the guy $100 (or whatever the fuck he wants) to calibrate your buds, or that you should buy Verum 2's. There's a really big difference between discussing the science and voting with your wallet on cases like these and it's distressing that a forum, and creators, dedicated to audio science seem to be unable to distinguish the difference.

It's like the audio equivalent of the 'Breakfast Question'.

 

[kemmler3D]

This isn't relevant. A persons opinion on speakers do not change their target curves sound.

Also yeah our brain tries to smooth it out. But it still obviously isn't optimal. Compare a true center channel with a phantom center in a well treated room. I guarantee the center channel will fix the comb filtering (as long as there isn't strong reflections).

If you have a true center channel with a left and right speaker aswell, you can also get wider imaging by being able to put the left and right speaker further apart, and still having the center playing without comb filtering.

Maybe he really does misunderstand comb filtering, if so it speaks to someone that is pretty thoughtless about audio and hearing. Speakers suffer from comb filtering... but to a varying extent, so does any sound that gets reflected inside a room, including live music. In that sense, if speaker measurements don't matter then nothing matters... But hmm, that can't be right, can it?

Instead of realising that the entire industry isn't vastly stupid and missing super basic insights for... Oh, 70+ years, he posts a clickbait video.

As a "scientist" he makes Sabine hossenfelder look as even keeled and thoughtful as the Nobel committee.

Aside from the unrelated video, I have to ask a dumb question about the conceptual validity of a preference test here.

I believe he got a good statistical result that says people prefer this target, sure.

What is less clear is what the preference means. Is it more accurate to reality or is it just a version of loudness / "showroom" enhancement?

What's the science beyond "this one got the most votes?'

 

[ruben1]

Nobody is saying physical comb filtering doesn't exist. If you put a single omnidirectional microphone where your head is, you will absolutely measure a nasty comb-filtering notch around 2 kHz caused by the crosstalk. The point is that a microphone is not a human brain. If you want a specific study, look up Dr. Floyd Toole’s Sound Reproduction (specifically his chapters on the stereo illusion and the "phantom center"). He explains that while the physical 2 kHz acoustic crosstalk dip is real, human listening tests consistently show we don't perceive it as a massive tonal coloration. Why? Because your two ears are spatially separated and sample two completely different parts of the interference pattern. Your brain performs what psychoacousticians call binaural integration—it essentially combines the distinct inputs from both ears, masks the sharp notches, and extracts a tonally stable phantom image. So yes, stereo is a psychoacoustic illusion. But saying "measurements don't matter because of comb filtering" is like saying "high-resolution TV screens don't matter because the human eye has a blind spot." The brain effortlessly handles the limitation, which is exactly why a well-measuring speaker still sounds objectively better in stereo.

It seems that again that you don't understand that the comb filtering is not a real issue for loudspeaker design and sound reproduction (psychoacoustics) while getting smooth FR and directivities matter upmost.

Watch this video Sharur made for you when he saw your post!
Floyd toole literally said the opposite of what you were saying

 

[ruben1]

I've seen more than enough measurements and used eqalizers since I was kid, so yeah I do pretty much know how that will sound like just by looking at it.

If that was the case you would probably understand how good peqdb diamond is

 

[oratory1990]

his Diamond target has been verified by listeners.

Listening tests with uncontrolled nuissance variables (uncontrolled level, uncontrolled background noise, uncontrolled leakage, ...) are not really useful for ... anything.
That's a major issue with doing online (not in-person) listening tests.

 

[Somafunk]

The End

 

[Cbdb2]

Watch this video Sharur made for you when he saw your post!
Floyd toole literally said the opposite of what you were saying

He said speaker measurements dont matter?

 

[kyuu]

Pretty much. I'm not saying you should be giving the guy $100 (or whatever the fuck he wants) to calibrate your buds, or that you should buy Verum 2's. There's a really big difference between discussing the science and voting with your wallet on cases like these and it's distressing that a forum, and creators, dedicated to audio science seem to be unable to distinguish the difference.

Kindly shove off with this. There's no issue with being suspect of someone who has a well-known history of spouting bullshit and being an awful person in general. Contrary to what awful people like to tell themselves, there is no correlation between being an awful person and doing good science. Quite the opposite, actually. For example, much of Nazi "science" was, in addition to being morally reprehensible, of poor quality.

In any case, this is quite plainly just more of this man's insatiable need to self-promote while doing shoddy work, then spend a lot of time being hostile to anyone who dares to question him. He has not performed any useful science, this is just some poorly-controlled listening tests. Then we get some random forum-goers telling us that this totally upends the work of actual scientists in the field because their totally unbiased uncontrolled listening says it's good. Yawn.

 

[Cbdb2]

What does any of this have to do with audio science? You may be shocked to learn the people behind most of the scientific innovation in the USA aren't particularly great people either, are you suggesting they should have abandoned rocket research for similar reasons?

So put him in a lab where he can keep his mouth shut and maybe he can actually come up with some scientific inovation. What hes done is taken previous work and adjusted it a bit. With very little scientific control. Was everyone listening at the same level? I doubt it. How consistant are $13 IEMs. IEMs dont fit everyone the same and this will make a huge difference.

 

[thewas]

He said speaker measurements dont matter?

Also if he wants a serious discussion he can respond here and not use his monologue castle, but maybe he already does?

 

[RickS]

Due to complaints, thread locked for review.