New Speaker Comparison Tool

[GXAlan]

https://www.crutchfield.com/S-CQ8N8qwEpDu/speakercompare/default.aspx

Crutchfield record speaker in anechoic chamber
Crutchfield records several headphones
They take the information together and let you hear some of the "tonal" differences between the speakers.

Edit: My enthusiasm has diminished the more time I spend with it. It's possible that there is a better combination of headphones that more accurately captures the differences between speakers. It's also a comparison of the direct response only as it's all done anechoically. Even though they show the use of an ear-shaped microphone, the recordings are not binaural.

I know Harman proved that you could EQ to simulate different headphones using a reference headphone -- but I'm not convinced that you can EQ to simulate different speakers (at least with what they have right now).

 

[thewas]

Thanks for posting, interesting approach, that of course needs to be validated.
Of course the problems of individual HRTF, different room responses and that a multidimensional soundspace is reduced to one transfer function per channel remain, but for a first tonality comparsion (which as we know is the most important for rating a loudspeaker or headphone) it still might be better than most usual dealer or home comparsions where loudspeakers are placed sub-optimally.
Just for a fun first test I took 4 loudspeakers which have good generally good reviews and decent measurements (Wharfedale Diamond 225, Focal Aria 906, Revel PerformaBe M126Be, Dynaudio Special Forty) and interestingly when listening to them through my Sennheiser HD600 (that are listed in provided correction options) the by far cheapest (Wharfedale) sound in the mids more natural to me than the rest.

 

[amirm]

I know Harman proved that you could EQ to simulate different headphones using a reference headphone -- but I'm not convinced that you can EQ to simulate different speakers (at least with what they have right now).

Harman has performed a lot of studies like this. But they first confirm the system is working using controlled testing and statistical analysis. That seems to be lacking here.

 

[GXAlan]

Harman has performed a lot of studies like this. But they first confirm the system is working using controlled testing and statistical analysis. That seems to be lacking here.

Agreed. The concept sounded promising so I posted it before I started really testing the site, I found that it was not particularly useful. It is a good concept, but I am not convinced they took the actual route of recording music in the environment using binaural mics and then applying a correction for specific headphones. I haven’t gone through their patents to see what they are actually claiming to do.

Likewise, they have too many headphones which wouldn’t be transparent even with EQ.

It’s great marketing though, which is good for everyone, and I love that it goes to 11 for the volume.

 

[amirm]

The mic array they are using is aiming to just capture the direct sound at a few angles. I don't know why they would ignore the rest of the off-axis sound that impacts how we hear the sound.

 

[edechamps]

Harman has performed a lot of studies like this. But they first confirm the system is working using controlled testing and statistical analysis. That seems to be lacking here.

My understanding is that these Harman studies not just compensated for differences in headphones - they also measured individual listener's HRTFs and compensated for that too. Crutchfield doesn't seem to do that.

Also it's not clear what acoustical environment (as in, what kind of room) they're simulating on playback.

 

[thewas]

My understanding is that these Harman studies not just compensated for differences in headphones - they also measured individual listener's HRTFs and compensated for that too.

Did Harman really do so? I don't remember reading such from Olive at his headphone simulator articles but my memory is poor, so does anyone maybe have a link or quote?

 

[Count Arthur]

I couldn't get it to work.

 

[spacevector]

This seems like a great tool assuming it captures the tonality accurately - I think we can mostly agree that this is paramount for speaker preference setting.

One cool enhancement could be blind testing. User chooses a price range and form-factor and the tool spits out options. User browses through and selects models that will work in their space - size, aesthetics, what have you. The tool then plays samples without disclosing which model is playing and makes the user note their preference in competing samples. At the end of the audition, the tool shows most preferred speaker(s) by the user and then does not allow them to buy any other speaker.

Yo Crutchfield, if you are reading this - please implement and send me a check. Thank you!