Note - the wrong target curve was tested for the IE section due to a mislabeling issue. Shown here is the DFHRTF with Harman 2015 filters, but it should've been the Harman 2018 filters, which actually resembles the SoundGuys curve closely. I've since sent the correct one to Sean so that'll be tested out in the future. Remember that we calibrate to the DFHRTF and then represent that data relative lower and upper bounds, which Dr. Olive also discusses in this talk.

Oh thank goodness someone recorded this since it really sucked that I still couldn't make it this year

Harman's back baby

22:34 Crazy that the man himself had to take time out of his day to debunk people like sharur.

It is a BLESSING to be able to view such a large amount of data. Thanks to Dr. Sean Olive and the Headphone Show! 🥰 I think the most important point is not what kind of equipment you use, but how well it is calibrated appropriately for each equipment. I am curious to see what the results will be if the same test is performed with the HMS.II HEC with ITU Type 4.4 ear simulator.

Just finished it now. Again, thanks a lot for uploading the full presentation. It’s really nice for those of us who weren’t able to make it to CanJam

Great presentation! Thank you Dr. Olive and guys from Headphone Show!

Aww yiss, thank you so much! I've been waiting for this so eagerly ever since I saw the announcement on Olive's twitter.

Thank you Dr. Olive, and thanks HPS staff for making this accessible for all to watch

10:21 is too good 😂

glad to see this research continue to get hammered out and incorporate community feedback and ideas. I'm still not convinced FR is the end all be all of headphone quality but it's certainly a great starting point to have so that headphone research can continue to mature, where as it had been neglected for so long.

Definitely going to watch this later Nice, I really like this 😊 Thanks 👍🏼

Thank you for recording this!!! One day, I will go to a Canjam!!

P R E F E R E N C E B O U N D S

Yay, I figured someone would record this so I spent all my time demoing stuff.

This would be like going to church for you guys....

Thank you . Very enlightening. I guess if you’re going to make a headphone that auto the most people a tighter delta in the bass and keep an eye on the 4-6 K .

Also, that PLS regression done at 57m in is driven massively by that one outlier. There’s one cluster and then an anchor point. So, filling in others in the low preference range is really needed to make that work.

Thank you so much for the video post! 👍

What’s crazy is that the Soundguys and Harman 2019 were ranked the same preference wise and yet they differ massively in the FR. It’s pretty much the biggest difference above 1k with the lowest energy in the SoundGuys and the most (or tied for) in the IE2019. So, the most different ones are ranked the same?

Sean telling Blaine gets his hearing checked 😂 poor Blaine, Sean woke up and chose violence.

fascinating, thanks!

Thanks so much for posting this! I was really hoping I’d be able to see this

This looks like it was held at the same location B&H Video uses for their videos too.

The goat speaks again

Oh boi, the Harman guy is at it again

Timestamps plz

The message I get is that it is Important to let people have tone controls or eq to adjust the sound to how they like it.

To me moondrop's VDSF target looks very simillar to the Harman target 2024

I thought all that mattered was complete compliance with the one current Harman Target and no distortion at 115db. The website that says this says "science" in the name so you can't question it.

*points to blaine* lmao

I know EXACTLY what the Austrian headphone is that measured like the Alps, hahaha!

I like Harman but it's a result of the circle of confusion. In an ideal world, we would return to diffuse field

👍‍‍⭐👍‍‍

Why not measure these things on human heads as well as machines to establish a baseline variance that can be applied to machine measurements in future? It doesn't make sense to me if you don't do that.

🤨

Glad to see the state of the art being advanced. But there is something I think is really missing from the common objective measurements in audio science. It's ignoring the time domain way too much. Playback devices exist in the physical world. They have resonant properties based on their material, rigidity, size, shape, etc. (Besides the room they're in). These properties are stimulus-dependent. As well as time-dependent. Resonances die off based on absorption or flexion etc in nearby materials (speaker housing, headphone earcup housing, IEM body, etc.). We could expect these resonances (sustained sounds) to constructively and destructively interfere with other sounds being played back -- it changes the sound of the next thing you play based on what was played just before it. For example, it is widely known in music production that tweaking attack/decay of a compressor will make something sound further or nearer to the listener. Our ear and brain hears the rate and qualities of the decay of louder/quieter sounds and uses it to estimate the size of the space. It is also why humans can be trained to echo-locate to some extent -- because rate of sound decay can give us positional awareness and clues about distance, despite not touching or seeing our surroundings. Some blind people have done this to help them ride a bike more safely. This rate of decay should all be objectively measurable, e.g. on a spectrogram of frequency response chart. We have microphones and couplers and all that, after all. This rate of decay and resonance, the physical properties/biases of a speaker headphone or IEM (and to an extent the room you're in), as well as the "speed" or transient handling of the driver, should change "imaging", "sound stage" but also timbre, in a stimulus (music sample) dependent way. All objectively measurable. *Objective sound measurement really needs a test corpus of some royalty free music/speech samples to test frequency response against.* (Royalty-free for legality of redistribution, for reproducibility purposes.) e.g. a second or two of some drum fills, a second each of some vocalists, speech with no music, full band, loud reverberant sample, sample in an anechoic chamber, go nuts. Point being, it needs to be some real-world sounds. Not just frequency sweeps or performance at 1KHz sine wave at xyz dB level. Frequency response of the device under test, with these standard real-world sound samples, should reveal objectively so many of the currently labeled as "subjective" qualities of audio setups. Sure, the point of these differences could be down to subjective values. But a reviewer confidently can say "this device reverberates at around 1k with an unusually long sustain, potentially coloring other sounds and affecting sound stage." Just as one totally random made-up example. *The visual image compression world doesn't test lossy compression techniques against just pictures of full-spectrum rainbows. The music and audio world shouldn't settle for tests based on just frequency sweeps. And music and audio world shouldn't relegate more complex test stimuli / test corpuses to subjective-only preference tests.* Objective testing for audio needs to start incorporating real-world complexity, standardized (repeatable) audio samples, or else we will continue to relegate some of the most important aspects of the audio experience ("audio quality") to be purely subjective forever. It doesn't make sense. *We have the equipment and the methodology available to test better, I can't understand why people don't do it.* Thank you for coming to my TED talk.

Oluv's still not gonna like this target.

I tend to appreciate other characteristics about speakers and headphones more than frequency response. Detail, speed, soundstage and voicing have a larger impact on whether the speaker/headphone are enjoyable to me. It's always more desirable to have a piece of equipment more competent in these areas, but I believe there is no single best response curve for any individual. All of those characteristics being equal, most audio enthusiasts seek variety. Preferences depend on mood, genre, media, etc. This is why people collect different gear. For example, for one individual, (again, detail, soundstage, etc being equal), a "grail" (i) home theater might have much more bass than the Harman Target, while (ii) their preferred headphones for metal music have much less bass than the Harman Target, and (iii) preferred headphones for pop perfectly match Harman. It would have less to do with normalizing the different frequencies emphasized across these genres - but more that people look for different experiences in different scenarios. It's also dependent on setting. I bump the bass at the gym and in the car, but leave the bass on my Focal Radiance (pretty laid back response) when critical listening. I think people place too much value on the frequency response, and many people in the audiophile community subconsciously feel this way. So long as the tuning isn't so unbalanced that you lose detail/soundstage or experience pain, the community should stop overemphasizing this characteristic. Product reviewers should describe the tuning, but it should be through the foregoing lens. In this regard, this type of research is valuable, but should not be used incorrectly.

People like beats because of the marketing - this is the huge issue with such a subjective hobby. Pepple like targets because of the mwrketing (reviewers, influncers etc). So then you have to ask the question - does any of it even matter as long as the consumer thinks they like what they've purchased? Does it matter why they think they like it?

So Harman fired AKG engineering team, and they was reiventing v shaped headphones and created a a fancy standard just to realize oldschool studio headphones was always truly "hifi"

46:01 shows there's a bigger preference window at 4-5k than even in bass, I think this should be its own band for the development of the next version of Harman, in addition to the bass and treble dials. I think if the headphone show adjusted this region alone as opposed to the entire treble they would have gotten much better results. With how many de-essing plugins exist this is obviously an important region.

I keep cracking up. Bass = aggression. Let me into the ladies sample so I can skew us back. Give me all the bass. 😂