Innovative personalized 3D sound solution for headphones: introduction and beta release

[myHRTF]

Dear ASR community,

Since this is the first post on this forum, kindly allow me to introduce myself:

I’m from Munich in Germany and 15 years ago I started to measure my own head-related transfer functions (HRTF) with microphone capsules inside my ears, hoping to replace speakers and room acoustics with headphones and digital signal processing one day.

After many years of research I have developed a new method to adapt the sound of specific headphone models to the user’s individual hearing online, taking into account important aspects of their HRTF. The objective is to create a lifelike simulation of listening in a premium recording studio through headphones without acoustic measurements being required.
Based on encouraging feedback, I have founded myHRTF to make this solution available.

We have just released the beta version and hope to gather feedback about the sound and website usability from interested ASR members. Currently, we support just under 50 over-ear headphones and in-ear monitors (IEM) with more to come soon.

In addition to standard stereo, you can choose between wide stereo (+/-45 deg.) and crosstalk-free sound reproduction. Switch between near-field and far-field listening, with room acoustics in line with international standards, plus simulation of concert hall acoustics by mouse click.

You may try it online for free, you just need your headphones and a ruler or similar:

myHRTF Personalized Filter Designer

Looking forward to your feedback.

All the best,
Philipp

 

[90sNintendoKid]

I'm extremely interested in this and I don't own any of the headphones in the list.........

How would the surround version work? For example what sources could I use? (Apple TV 4k for example).

I'm hoping this is going to save me money compared to an A16 Realiser.......

 

[myHRTF]

I'm extremely interested in this and I don't own any of the headphones in the list.........

How would the surround version work? For example what sources could I use? (Apple TV 4k for example).

I'm hoping this is going to save me money compared to an A16 Realiser.......

Thank you for your interest. If you let me know the headphone(s) you own, I will check what we can do for you.

What you get in the end are personalized filters (impulse responses) that you can import into virtually all software applications that support so-called convolution (signal processing method). Hence, music or movie sound needs to go through some sort of PC or laptop that provides this processing in realtime.

When we talk surround or even immersive audio in movies or on TV we need to get to the decoded multichannel audio first in order to apply above mentioned processing. The most cost effective way is to use an Apple PC or laptop in combination with the Apple TV app (and/or Apple Music app) since Apple PC and laptops provide for this decoding. Inside this PC or laptop you would route the audio via virtual audio cables (e.g. BlackHole) to a VST host software that runs the convolution processing (e.g. Hang Loose Convolver and Host). You may connect his PC or laptop to a TV or beamer via HDMI for a bigger screen if you like.

We will describe this solution in our support section together with step-by-step guidance as soon as we launch the surround option (Autumn this year).

Philipp

 

[GA16SE]

In your first post in this thread, you’ve written: After many years of research I have developed a new method to adapt the sound of specific headphone models to the user’s individual hearing online…..

Would you like to provide a few links for your findings as published in the specific literature? One more question please, as @90sNintendoKid has mentioned the Realiser A16 in his post, how would your innovative personalized 3D sound solution for headphones compare and contrast to SVS (Smyth Virtual Surround) solution, which is found in Smyth Research’s Realiser A16 processor.

 

[myHRTF]

In your first post in this thread, you’ve written: After many years of research I have developed a new method to adapt the sound of specific headphone models to the user’s individual hearing online…..

Would you like to provide a few links for your findings as published in the specific literature? One more question please, as @90sNintendoKid has mentioned the Realiser A16 in his post, how would your innovative personalized 3D sound solution for headphones compare and contrast to SVS (Smyth Virtual Surround) solution, which is found in Smyth Research’s Realiser A16 processor.

Few years ago I made the decision to give it a try and commercialize my work. With this decision, publishing results would have required to patent them beforehand. Unfortunately, patenting and defending patents is out of reach for a small start-up serving a niche market. Hence, I decided to publish my work as a product.

In 2010 I implemented the methodology of the product that you mentioned in Matlab using HD650, but without head-tracking. Direct comparison is difficult since this method is limited by the room acoustics that was available during calibration and more often than not this room acoustics is flawed.

In contrast, our solution is based on simulated room acoustics that is

• free from resonances and flutter echoes,
• free from strong reflections within the first 20 ms,
• free from reflections from unfavorable directions,
• in line with Griesinger concerning both the reverberation decay profile and LOC, and with a
• reverberation time in line with Recommendation ITU-R BS.1116-3.

You can test it online free of charge and I want to strongly encourage everyone with an interest in this topic to invest 20 minutes of their time.

Philipp

 

[myHRTF]

Having followed ASR for quite some time already, I know that there are several members with exceptional dedication and experience in speaker virtualization for headphones, for example @jaakkopasanen , @dasdoing , @Lion♡ , @kemmler3D , @OK1 , @Robbo99999 , @Mr. Haelscheir , @olieb , @Muflatschki Gurkovski , @fcserei , @LionIT . I would love to hear your opinion! You just need one of the supported headphones and a ruler or similar:

myHRTF Personalized Filter Designer

Please let me know in case your headphones are not supported. I will check if I can integrate your specific model quickly (over-ear and IEM).

Happy listening,

Philipp

 

[Hugo9000]

Interesting, but it's a pity you don't have support for Sennheiser HD 600 or HD 650, since so many own one or both of those models.

 

[myHRTF]

Interesting, but it's a pity you don't have support for Sennheiser HD 600 or HD 650, since so many own one or both of those models.

You are absolutely right. Next week we will upload more headphone models, including HD6XX, HD600 and HD650.

Philipp

 

[aquathesecond]

Is it possible to add support for the HIFIMAN HE-X4? Thanks!

 

[Robbo99999]

Having followed ASR for quite some time already, I know that there are several members with exceptional dedication and experience in speaker virtualization for headphones, for example @jaakkopasanen , @dasdoing , @Lion♡ , @kemmler3D , @OK1 , @Robbo99999 , @Mr. Haelscheir , @olieb , @Muflatschki Gurkovski , @fcserei , @LionIT . I would love to hear your opinion! You just need one of the supported headphones and a ruler or similar:

myHRTF Personalized Filter Designer

Please let me know in case your headphones are not supported. I will check if I can integrate your specific model quickly (over-ear and IEM).

Happy listening,

Philipp

Thanks for tagging me on this, I'll definitely give this a try, it's interesting. I don't have time to do it tonight, but I'm replying here so that I get subscribed to this thread which should help me remember to try your system out. I think I'll try it with my HD800 headphones as they're supposed to be low unit to unit variation and I've got new pads on mine. So you're obviously trying to take the headphone used out of the equation by EQ'ing each model of headphone to a certain target that then has the HRTF filters put on top? As an end result how far different in tonality is it off from Harman Curve tonality? So talking bass levels and just overall tonality?

EDIT: I'll also try this with my HE400SE. So I'll be trying it with HD800 & HE400SE.

EDIT#2: and before I forget to ask, what measurements did you use to characterise each of your headphone models? Did you measure them yourself & if so using what device or rig? Or did you use an existing database of headphone measurements, like Oratory's for instance?

 

[myHRTF]

Is it possible to add support for the HIFIMAN HE-X4? Thanks!

Good news: HE-X4 will be part of next week's upload.

Philipp

 

[myHRTF]

Thanks for tagging me on this, I'll definitely give this a try, it's interesting. I don't have time to do it tonight, but I'm replying here so that I get subscribed to this thread which should help me remember to try your system out. I think I'll try it with my HD800 headphones as they're supposed to be low unit to unit variation and I've got new pads on mine. So you're obviously trying to take the headphone used out of the equation by EQ'ing each model of headphone to a certain target that then has the HRTF filters put on top? As an end result how far different in tonality is it off from Harman Curve tonality? So talking bass levels and just overall tonality?

EDIT: I'll also try this with my HE400SE. So I'll be trying it with HD800 & HE400SE.

EDIT#2: and before I forget to ask, what measurements did you use to characterise each of your headphone models? Did you measure them yourself & if so using what device or rig? Or did you use an existing database of headphone measurements, like Oratory's for instance?

Thank you very much for your interest, your feedback is most welcome!

Your questions are very relevant. Let me answer them tomorrow, it's getting late over here...

Philipp

 

[kemmler3D]

Having followed ASR for quite some time already, I know that there are several members with exceptional dedication and experience in speaker virtualization for headphones, for example @jaakkopasanen , @dasdoing , @Lion♡ , @kemmler3D , @OK1 , @Robbo99999 , @Mr. Haelscheir , @olieb , @Muflatschki Gurkovski , @fcserei , @LionIT . I would love to hear your opinion! You just need one of the supported headphones and a ruler or similar:

myHRTF Personalized Filter Designer

Please let me know in case your headphones are not supported. I will check if I can integrate your specific model quickly (over-ear and IEM).

Happy listening,

Philipp

Definitely interested but can't give an ETA on testing right now, just had a baby so audio stuff is off the priority list for the moment...!

 

[Robbo99999]

Thank you very much for your interest, your feedback is most welcome!

Your questions are very relevant. Let me answer them tomorrow, it's getting late over here...

Philipp

No worries, I'll be back on tomorrow to take a look.

 

[myHRTF]

So you're obviously trying to take the headphone used out of the equation by EQ'ing each model of headphone

Correct, you need to adapt to every headphone model individually since most over-ear headphones show wildly different frequency responses. You can hear that easily on our website when you intentionally select a wrong headphone model.

As an end result how far different in tonality is it off from Harman Curve tonality? So talking bass levels and just overall tonality?

The target curve is different from Harman and depends on the intended direction of arrival, for stereo this is front left / right. Hence, tonality is a little bit different, too. It sounds like you sit in front of speakers: slightly less bright and with more body compared to Harman headphone sound.

EDIT#2: and before I forget to ask, what measurements did you use to characterise each of your headphone models? Did you measure them yourself & if so using what device or rig? Or did you use an existing database of headphone measurements, like Oratory's for instance?

We need to measure the headphones on our own. In fact, most of the time and money goes there.

Looking forward to your feedback.

Philipp

 

[myHRTF]

just had a baby so audio stuff is off the priority list for the moment

Congratulations! Being a dad myself I fully understand this and will wait for your feedback patiently.

 

[Robbo99999]

Correct, you need to adapt to every headphone model individually since most over-ear headphones show wildly different frequency responses. You can hear that easily on our website when you intentionally select a wrong headphone model.


The target curve is different from Harman and depends on the intended direction of arrival, for stereo this is front left / right. Hence, tonality is a little bit different, too. It sounds like you sit in front of speakers: slightly less bright and with more body compared to Harman headphone sound.


We need to measure the headphones on our own. In fact, most of the time and money goes there.

Looking forward to your feedback.

Philipp

Hi, thanks for the answers. What rig did you measure the headphones on? How many units of each headphone did you measure? Did you measure the headphones with new pads, worn pads or something inbetween?

 

[olieb]

Welcome!
It is great to see that people are interested in binaural virtualisation and want to make it readily available. Thank you for that and godspeed!
These are my quick and dirty first impressions.
First of all, you need a stop button for the sound on your website. And when you are at it, you might improve the UI in other ways too.
About the sound, it is interesting. I am interested in classical music foremost and checked those examples first (which for me did not work the first time but only after choosing from the pop-down a second time).
It produces a "realistic sound" in some way, lifting the sound out of the head and putting me in a generous virtual space filled with sound, but ...
There is quite a lot of reflections even with studio and 1m distance selected. Way too much for my taste. It sounds a bit like being in a row quite far into the hall where "proximity" is all but lost. Spacious, absolutely, but everything rather cloudy. "Crosstalk ON" gave a little bit more focus (as expected) but nowhere near what I would prefer. And selecting "Concert Hall" made everything even more cloudy.
And the orchestra, in particular the middle part directly in front, seemed floating for me in mid air several meters above ground.
Some of the other examples (i.e. the first "Break ...") sounded a bit like coming through a barrel for me. Changing the parameters did not do much about that. (Switching the playback options is tedious as there are dependencies between the menu items in different menus.)
Using the bypass function it became obvious that there is EQ going on for the HD800s, but not the kind I would apply. Violins (and everything else with treble) sounded very piercing.
For any kind of meaningful comparison the bypassed signal should have (optional) EQ too. A room virtualisation should not turn the sound balance upside down as this will make tonality fully dominate the experience.
The Beethoven did not sound good to me at all, neither bypassed nor with the filter. I preferred the Händel piece in bypassed mode (I assume normal stereo?) even with in head localisation, although I really hate that.

Very sorry for the bad feedback, but I want to be honest.
I do not know what exactly your software does and what is going on behind the scenes. This test does not reveal that. Perhaps there is a lot of improvement in the pipeline. And of course others may have a much more positive experience!
But I don't see how a rough estimate of concha dimension together with similar for the head can give a good HRTF model. And generic HRTF solutions exist for quite a long time. And there are good ones like APL Virtuoso.
As I understand, there are algorithms to create an HRTF from multiple photographs of the ears/head. To me it would be much more promising to offer that as a service (in .sofa format).

 

[Robbo99999]

I'll give my feedback on this after trying just now with HD800 headphones, here are my settings, I managed to measure everything and somehow even manage to measure my ears on my own including finding out I can see a lot of my ear canal from the side (courtesy of photo):

I think with these settings the virtual speakers were level with me, so they weren't above or below. My biggest first impressions are that the music doesn't sound wrong tonally, so I think that's roughly right, and it does to me sound like you're in a room listening rather than listening on typical headphones - to me it sounded like listening to a live performance in a room more than recorded music. Now this makes me think you might be loosing some fidelity but it's really hard to know when the sample tracks are not your own tried & tested tracks that you use for things like EQ experiments, because it's pretty darn hard to judge the effect of the filters from music that you're not familiar with. So I think my biggest request would be to allow for the option of uploading a track and then having the ability to play it through your filters. It's really hard to judge without using it on your own music. As I said though biggest impressions were that it doesn't sound tonally wrong and it sounds like listening to live music in a room I'd say, but we need to be able to try it on our own music to know for sure.

EDIT: when I say "sounds like live music in a room", I don't think all recorded music should sound like that, afterall that's not how my anechoic flat speakers sound in my room, so that's why it's important to be able to test these filters on your own music that you know well so we can be sure on the differences.

 

[ob1]

Hello Philipp,

I have given your software a try and thought I could share my feedback. I wish I could have tried it with my over hear headphones, notably Sennheiser HD560S, but the only ones listed I had were IEM, Truthear Crinacle Red.

I had noticed that you had warned users of the -10 dB preamp gain required, but I was wrongly under the impression that the bypass version would be 0 dB and personalized HRTF at -10dB, while it is actually the other way around. My bad. I wished I had the time to match to level match both versions, will try it again later.

I am unsure how valid is my opinion, since I know for certain my ear canal is unusual. I have a formal training in hearing aid dispensing and my fellow students always commented how "tortuous" and "bobsleigh-like" were my ear canals. In sum, take my feedback with a grain of salt, I might be an outsider...

In short, I was positively impressed how externalized was the rendition and it indeed felt like listening through speakers.

Whatever the angle or crosstalk setting, I felt that the soundstage was a bit too diffuse/imprecise/blurred to my tastes. However, I have had a similar impression with every software I have tried.

As for the tonality, I thought it was right, perhaps a bit rich with prominent lows and highs, but it would make sense that the +10dB boost is responsible for this change. I have a preference for Amir EQ, but thought that I was mostly close to that, save for the bass was more prominent.

My training was not very deep (undergraduate level), yet I am unsure why is it that HRTF almost always involve a simulation, while in my field, we would actually measure a few millimeters from the ear drum (with and without the ear canal probe) to get actual HRTF measurements. (Of note, I graduated, but never worked in the field, please excuse me in advance.)

As others have mentioned, I would love to be able to test it with my music.

Thank you and congratulations on your effort, your software look promising.