Aural ID from Genelec: Making headphones like stereo?

[oivavoi]

Haven't seen any discussions of this here yet, so I thought I'd start a thread on it.

Headphones don't sound like stereo, we all know that. Stereo soundstage happens in front of you, headphone soundstage happens inside the head. The reasons for this are complicated (google "head related transfer function"), and has to do with how the ear/brain uses direction cues to localize sounds and create a stereo illusion in front of us. As far as I know there has until now only been one really successful way of making headphones sound more like stereo - the Smyth realiser hardware. Problem: It's expensive, and almost impossible to buy (has it become vaporware?).

But I recently discovered that Genelec has now created an algorithm which promises to do the same. https://www.genelec.com/aural-id

One only needs to take a video of ones head and torso with a smartphone, upload the movie to Genelec, and they will then generate a personalized file/plugin which can be uploaded to different computer programs. This will then equalize and change the signal so that our brain thinks that the sound is coming from the front, not from inside the head. Cost is around 600 euro.

It seems to be geared towards the pro market for now, but I see no reason why this can't be used by audiophiles and music lovers as well... Any thoughts?

The only drawback for me personally is that I've gone back to CDs for my listening, so in that case I would need to find a smart way of using a raspberry PI for example as a digital interface between the CD player and the headphone amp, which seems somewhat complicated.

 

[Rockfella]

Nice info! Sounds radical. It will cost a bomb though

 

[oivavoi]

Maybe @Thomas Lund wants to chime in?
Anything to think about if one wants to use this as an audio consumer?

 

[fredoamigo]

you go back to the cds for your listening and you sell the D&D...what's your therapist?

 

[oivavoi]

you go back to the cds for your listening and you sell the D&D...what's your therapist?

Hehe The Dutchmen had to go because they didn't fit in the living room after moving, unfortunately! As for CDs... well, I've come to the opinion that streaming is bad for my relation for music, and recently decided to quit Tidal and Roon. Wrote about this in another thread:

This may be slightly off-topic in this thread, but it comes from a person who has been streaming for years, and is now heavily debating with himself whether to go all in on Roon or not, or just skip the whole streaming thing and go back to CDs for ever (just leaving airplay functionality in the hifi system for the occasional youtube concert):

Aren't there some dangers with this kind of algorithms, this limitless freedom of discovery? That we become shallow in our musical listening, just ever going on to the next musical thing instead of really delving into music we can get intimately acquainted with? Is streaming to music what tinder has been to romance? The "swipe next" culture overtakes music?

When I was in my teens and twenties - before streaming - I remember that I knew many CDs by heart. Whether it was pop or rock, or classical pieces and interpretations, or classical jazz albums from Mingus or Coltrane - I knew them intimately. I did not have a limitless supply of music at home, so I really listened to what I had. This, I believe, made this music become a part of me. I knew every single word on Abbey Road or After the Gold Rush, I knew every single note Keith Jarret played on the Köln Concert. And so on.

The last decade - in my 30s - I don't think there is any new music I've only listened to which has become a part of me in a similar way. I suspect this is because of streaming. I almost never to listen to a recording more than twice, at the very max, sometimes three if it's something really special.

I feel that I have lost something. As a singer in a semi-professional classical choir, I also know that intimate exposure to music over time does something. When we practice a difficult piece every week for a couple of month, and then perform at two or three concerts, I really get to know it. It's like the different parts of the piece start speaking to me in a different way after a while - it really gets to me.

Listening to a CD repeatedly had some of the effect on me. But are we losing this now, in our new musical culture? I think so. I've certainly lost it myself, at least, and wonder whether I will need to drop streaming completely to get it back. So even though Valence looks promising, and might turn out to be better at predicting my preferences than other algorithms (Spotify has done the least bad job so far), I'm torn as to whether this would actually be a boon to me, everything taken in account. Maybe I would become more musically enriched by listening to my old CDs one extra time instead.

 

[Stefano_Milano]

Isn't there a software that transforms stereo files into binaurals simulating a virtual head in whose ears signals arrive from right and left with different intensity?

 

[oivavoi]

Isn't there a software that transforms stereo files into binaurals simulating a virtual head in whose ears signals arrive from right and left with different intensity?

There are many software solutions for inducing crosstalk etc into stereo recordings, thus possibly making them sound more "binaural". The problem is that the actual HRTF is unique to each individual, so for it to really work one needs to measure the actual head and torso of the listeners. As far as I know this is only done by the Smyth hardware and the Genelec software.

 

[Stefano_Milano]

There are many software solutions for inducing crosstalk etc into stereo recordings, thus possibly making them sound more "binaural". The problem is that the actual HRTF is unique to each individual, so for it to really work one needs to measure the actual head and torso of the listeners. As far as I know this is only done by the Smyth hardware and the Genelec software.

I did understand. This is why I continue to prefer listening with speakers rather than headphones. Even at low volume the sound experience is more realistic than having the singer in the middle of the head. The best solution would always be to have the binaural version of the recording but unfortunately it is not possible.

 

[Blumlein 88]

One big benefit (I suppose never actually used one) of the Smyth Realizer is they kept track of your head movements and let you move your head. So you'd hear it like two speakers fixed in your room as you move your head. I wonder if the Genelec software, even if it works well, will still be less nice to listen to because the soundstage will move around with your head.

 

[Jumbotron]

Waves NX app for Windows (and MAC, I think) is another of the software solutions. It also offers some customization based on two measurements. One is the user's head circumference. The other one is the distance between both ears. It is not expensive.

However, I am using the Headphone Surround Virtualization (HeSuVi) free software, which emulates Waves NX and similar solutions. I am not using it for listening to music though, just for movies and games.

 

[oivavoi]

One big benefit (I suppose never actually used one) of the Smyth Realizer is they kept track of your head movements and let you move your head. So you'd hear it like two speakers fixed in your room as you move your head. I wonder if the Genelec software, even if it works well, will still be less nice to listen to because the soundstage will move around with your head.

Interesting comment. Didn't think of that. Haven't tried the Smyth realiser either, but I've read enough about it to assume that it really works as advertised... Another question where it would be good to get comments from @Thomas Lund !

 

[oivavoi]

I did understand. This is why I continue to prefer listening with speakers rather than headphones. Even at low volume the sound experience is more realistic than having the singer in the middle of the head. The best solution would always be to have the binaural version of the recording but unfortunately it is not possible.

I agree. I always prefer stereo on good speakers. And unfortunately having a binaural recording isn't enough... I've listened to many binaural recordings, and while they can sound somewhat better than stereo recordings over headphones, it doesn't sound like stereo. The reason is that people's heads are different, so such recordings go for an "average head": "Due to the diversity in HRTFs it is impossible to create a binaural effect compatible for everyone’s ears. Therefore the simulated dummy head recording algorithm uses average HRTFs to create a moderate binaural effect for everyone" (italics by me).

I would really like to try the Smyth realiser if I can get hold of one, though.

 

[Soniclife]

One big benefit (I suppose never actually used one) of the Smyth Realizer is they kept track of your head movements and let you move your head. So you'd hear it like two speakers fixed in your room as you move your head. I wonder if the Genelec software, even if it works well, will still be less nice to listen to because the soundstage will move around with your head.

This strikes me as important when you are watching video, in surround sound, I don't feel it's important for listening to music in the same way, but maybe the illusion is stronger with it.
Where does the bacch system fit into this?

 

[Thomas Lund]

Thanks for prod, oivavoi. I have primarily considered binaural personalisation from basic human and pro perspectives, so bear with me for maybe going off on a tangent in relation to this forum.

If two people listen in the same room, they are able to precisely discuss what they hear. That is not the case if they share a pair of headphones. Experienced audio engineers - like audiophiles - are highly active listeners, making full use of e.g. outer ears, efferent pathways to the inner ears, movements and any cross-modal amplification they can get. Here are two brief summaries about such issues,

https://www.genelec.com/-/how-does-it-sound-on-headphones-personalised-headphone-listening-•-part-ii
https://www.genelec.com/-/professional-listening

Aural ID replaces a headphones' one-size-fits-all frequency response with a custom compensation, based on an approximation of the actual listener’s external ear, head and upper torso features. That is, however, only a first step towards a natural listening experience.

A processor then needs to calculate direct sound and room reflections, based on the >800 angles of arrival in your personal map (a SOFA file). Furthermore, movements need to be taken into account for low latency audio processing updates. Finally, cross-modal amplification (seeing the same environment you hear, balance, acceleration, body posture, smell, temperature etc.) affects our listening experience.

While the Smyth Realiser actually addresses many of those points in an integrated solution, Aural ID is currently more of as an academic tool for investigating various factors of binaural reproduction systematically. For easy home listening, a hedonistic approach like Harman’s target curves, is one way of qualifying a generally agreeable headphone listening experience.

 

[Kal Rubinson]

This strikes me as important when you are watching video, in surround sound, I don't feel it's important for listening to music in the same way, but maybe the illusion is stronger with it.

I think it is because we have been trained by experience to use small (or, even, large) head movements to aid in our perception of sound sources and ambiance. We cannot completely suppress such movements without using a head-clamp and/or paying more attention to head movement than the music. As a result, we are likely to find the lack of cues in response to head movements unnatural albeit at a subliminal level.

 

[Soniclife]

I think it is because we have been trained by experience to use small (or, even, large) head movements to aid in our perception of sound sources and ambiance. We cannot completely suppress such movements without using a head-clamp and/or paying more attention to head movement than the music. As a result, we are likely to find the lack of cues in response to head movements unnatural albeit at a subliminal level.

That seems the same as what Thomas explained above, something I'd not considered, but might well do myself now it's been pointed out.

 

[oivavoi]

Thanks for prod, oivavoi. I have primarily considered binaural personalisation from basic human and pro perspectives, so bear with me for maybe going off on a tangent in relation to this forum.

If two people listen in the same room, they are able to precisely discuss what they hear. That is not the case if they share a pair of headphones. Experienced audio engineers - like audiophiles - are highly active listeners, making full use of e.g. outer ears, efferent pathways to the inner ears, movements and any cross-modal amplification they can get. Here are two brief summaries about such issues,

https://www.genelec.com/-/how-does-it-sound-on-headphones-personalised-headphone-listening-•-part-ii
https://www.genelec.com/-/professional-listening

Aural ID replaces a headphones' one-size-fits-all frequency response with a custom compensation, based on an approximation of the actual listener’s external ear, head and upper torso features. That is, however, only a first step towards a natural listening experience.

A processor then needs to calculate direct sound and room reflections, based on the >800 angles of arrival in your personal map (a SOFA file). Furthermore, movements need to be taken into account for low latency audio processing updates. Finally, cross-modal amplification (seeing the same environment you hear, balance, acceleration, body posture, smell, temperature etc.) affects our listening experience.

While the Smyth Realiser actually addresses many of those points in an integrated solution, Aural ID is currently more of as an academic tool for investigating various factors of binaural reproduction systematically. For easy home listening, a hedonistic approach like Harman’s target curves, is one way of qualifying a generally agreeable headphone listening experience.

Thanks for the explanation, @Thomas Lund ! Very interesting. But in order to simplify this a bit a binaural noob like me, can you say a bit more about it actually feels to use aural id? Does it make the stereo image happen in the front of the head like with speakers, for example?

 

[Thomas Lund]

Aural ID rendering of a composite stereo source at +/- 30 degree angles is different but, less natural than stereo over speakers. It helps if the source format is immersive but I would still prefer an in-room system.

Possibly, we might learn to appreciate personal HRTF rendering more, even without room reflections, cross-talk, movements and cross-modal amplification; I'm just not there yet. Maybe it also stands in the way how I have learned a certain stereo headphone sound, where it ought to have been the other way around. 27 years of using the same pair of AKG K240s influences a listener, at least this one K240s from when AKG was independent are well engineered, by the way. Their frequency response has stayed the same, L/R and seating variation next to nothing.

 

[maxxevv]

Creative Technology already has a product(s) that does that. But as with all things, YMMV ... some take a liking to it, but some don't. Its a mixed bag.
Personal experience from trying a friend's SFXI dongle is that it works somewhat but the DSP processing seems to have some effect on the quality of sound reproduction. A mixed bag in my opinion

https://sg.creative.com/sxfi/

 

[JJB70]

In terms of simply listening to music in stereo will this do anything more than applying some cross feed?