Discussion in 'Blind Testing and Psychoacoustics' started by ChaChaRealSmooth, Nov 9, 2020.
What you're probably looking for is this:
That was a bad idea and instantly hurts the science behind it. They just discarded 99% of people who would call them on their bullshit: the musicians and sound guys.
Almost all of them have some hearing loss and tinnitus whether they admit it or not. They are the group most likely to pick this stuff out. The distortion is not psychoacoustically masked anymore by the physical ear. We are the guys who will not only hear 1 db of compression, but bad dithers, noise shaping, and different tubes and op amps.
The internet does not reveal this because most of the people on the big music and audio sites have never worked on audio electronics, played real instruments live, or worked with artists playing real instruments at real volumes. The hive mind ASR and Reddit mobs have infested them.
You are thinking way too hard about this. Last time I checked, Neuman or JBL didn't make specific tunings depending upon country. While the monitors from Neuman, JBL, Yamaha, Mackie, Presonus, Tannoy do differ somewhat in FR, they are probably a closer together than any two samples of the the HiFiMan Deva headphone. And trust me, none of these below sound anywhere Harmanized. People with hearing loss can adjust, especially if they are pros in sound mixing and mastering.
I'm thinking slide 10, where the green dashed line is the flat in-room curve as measured at the HATS. We already know what a flat curve at the listening position in-room curve is going to sound bright.
Ah, you're right - I was thinking the green line on slide 10 was the speaker target on slide 5 applied on top of the in-room response, so pretty extreme. Still maybe an issue though, as that preferred correction is a bit more than B&K and we don't know if what's responsible is the room being a bit different from a 'good B&K room' or if it's listener preferences.
I'd like to see how speakers that measure flat in an anechoic chamber measure in that room. They must have done that, but I haven't found it online.
Going back to Marv's comment about handing headphones to someone at work. My experience doing the same thing was that the first thing the coworker did was turn down the volume, then as for more bass. That got me to wondering if Harman wasn't controlling for consumer playback levels and adding in a loudness curve to make it sound more neutral at a lower listening level.
If you guys are like me, you prefer a balanced or neutral frequency response, but you play at louder levels to hear the details.
They have a group of trained listeners in the Harman research center, which I think is based South California. It's the same group that Sean Olive used in his research in speakers, which I believe Revel uses for their tuning.
What's the definition of "trained" listener? Someone who achieved level 4 in the Harman online trained listener course?
This can be done without a meet. A decent small condenser microphone recording of human voice, guitar, car exhaust, random sounds. Play back on HD650 tweaked for SBAF curve. Then apply Harman adjustments per Oratory on HD650.
Some internal version of that, but there needs to be minimal variance in their collective ability to notice changes, since Harman uses few trained listeners instead of many untrained listeners, but they reportedly have equivalent statistical power. Olive stated each trained listener was equivalent to 8 untrained listeners.
Oh, sort of like how each long-time active SBAF member is equivalent to 1000 untrained listeners? I see.
If you could get each long-term active SBAF member to grade things the same way consistently, you could probably sell their services to Harman and eventually make a new Harman curve.
Na. Harman's got Sean Olive. AES papers, and subsequent adoption by Internets (Oratory, Rtings, and now ASR, etc.) have a lot of pull. The idea behind this is a sound profile that consumers will like and presumably buy. What Olive did was establish this with formal research and investigation. (I'm surprised Jude hasn't gotten in on Harman train yet.)
A better way to increase sales to get some famous rapper or a Kardashian to sell headphones. Or advertise like hell and buy shelf space at the airports like Bose.
We need to check out the AKG K371 since it's supposed to be the closest to the Harman target. Get that on the loaner program.
I can see it now. Raycon tuned by SBAF.
How convenient their preferred sound profile for headphones correlates with their profile for speakers. No where is any of this compared to live sound.
Did he actually say this and is this actually part of his research methodology into his consumer curve? Im not even sure what to make of this. This stuff makes it into peer reviewed audio journals? I will acknowledge that I dont now all of the tiny details here but this lessens my opinion of the curve and it already was less than stellar.
Not him specifically, but he does state that's the reason trained listeners are used.
There notice the same changes, and grade the changes similarly, so you need fewer trained listeners to grade something and still have a high statistical power.
He does other studies that show Sean Olive version of neutral is considered the most desirable speaker tonality for large groups of normal people.
The Harman curve is an extrapolation of that to headphones.
This is a presentation of the second paper on the Harman Curve, where they test it with a combination of trained listeners and a bunch of other guys that work at Harman. They basically simulate the frequency response curve of multiple headphones.
https://www.dropbox.com/s/cty4728c8...phone Sound Quality Preferences .key.pdf?dl=0
I just wish it was one of several curves.
The existing Harman curve for the bassy average Joe.
"Neutral" (begin flame war) studio-oriented sound in the vein of Neumann and Genelec monitors. Or really, in the vein of the Neumann NDH20 with Sonarworks profile. Or hell, a slightly veiled Sennheiser if they can't do the treble as cleanly, but add a little oomph in the bass.
Trebley, presence-range boosted profile for speech (Youtube, movies, audiobooks, podcasts) or people with hearing loss (older folks in particular).
EQ is great but most people are never going to use it and you should be able to plug into anything without needing EQ to have a decent listening experience.
IMO, Harman could easily provide these three flavors. Doing so in hardware is a tall order, I know, but if they were targeting wireless (like their JBL cans) then you could just take a lower-fidelity can targeting "neutral" and provide the other two as EQ presets, right?
I'm sure I'm oversimplifying but I followed the development of this curve for years, from Tyll's posts onward. I follow Sean Olive on Facebook. I had Lauren Dragan of Wirecutter ask him a weird question one time (I don't remember what it was but he was mostly confused). Harman has the research facilities and technical chops to make whatever they want. I just wish they'd step back from the curve for a minute and try a few alternatives. Make something for the audio enthusiasts - not even us, but the online person who would drop ~$300 on Bose or Sony wireless cans or the Senn HD6xx - and see what happens. Tease the prototypes at trade shows, or by sending to strategic influencers (which would include SBAF in my mind). They could rake in a lot of cash.
They've started on that: see here and here. Obviously it needs more data on specific demographics (eg people who are prepared to spend to a price point, as you refer to).
And they'd hope to be able to automate it, using the process in this patent application.
All of which reinforces that the Harman curve isn't intended to be much to do with neutral, but rather with preference(s), in order to help sales. Fair enough, it's a business. As long as everybody understands it's a preference, which seems to be the problem over much of the internet. Kind of like with single-figure distortion numbers...
I’m not sure they would agree with you on this because it seems in their mind, this is the “correct” curve because it correlates with the in-room frequency response of what they deem as an “accurate” loudspeaker.
This leads me ask, accurate to what?
Actually no - their headphone curve is different from the "accurate" (flat in-room) loudspeaker as in slide 10 of that 2016 review; and the flat in-room loudspeaker wasn't perceived to be "neutral" in the room (red preferred correction on slide 5).
Yes. I take your earlier point about live sound vs speakers; but in this context I'd still like to see how a speaker that measures flat in an anechoic chamber measures in their experiment room, so we know exactly what their 'flat in-room' is.
Separate names with a comma.