[oyebto]

Noticed it hasn’t been answered, just wondering is the Mobius made in USA like the other products?

 

[Alcophone]

This might work with some HMDs that have a USB output like the Vive. But some of the new ones and rift only have analog output. We can certainly get any USB or analog output and feed it into the headphones and use either 3D off or auto-align mode. We had some students at the USC GamePipe lab use the mobius this way for a demo. It was interesting.

Yeah, for the Rift you would need a USB extension cable parallel to the Rift's cable. For wireless models HMDs it will get tricky

But whats the point? Dont VR headsets already do their own surround virtualization (or at least offer it) and they obviously already have head tracking. So wouldnt the same features in the mobius be redundant in that particular instance? I mean the audio is already being processed that way, so you could still use them normally just without processing, but that case would be no different than any other headphones. How is what the ossic doing any more sophisticated than what the VR headsets can already do with their own tracking and virtualization?

The Mobius might support sensing the distance between your ears (read that somewhere as a consideration), which allows the spatialization to more accurately calculate the delay between both channels necessary to make it seem like a sound is coming from a certain location.
The Ossic X will definitely support that. In addition, the Ossic X has four small drivers per ear (above, below, in front and behind your ears) that get fed independently. They are used to make the high frequencies (most important for localization) come from where they should (based on the relative location of whatever is making the noise), and the shape of your ear does the rest. The traditional approach has to simulate your ear, which usually means simulating an average ear, since all sound will come from the side. As a result, the Ossic X should be more precise than the Mobius, with both better than what you can do in software alone. Though the Mobius could be matched if you know the distance between the ears.

 

[TidalWave]

Battery life is rated at 10hrs+. I'm curious if the 10-hour mark is for all the features on and running and moderately high volume? What about moderate-safe listening volume, stereo 3.5mm input (BT off) without 3D emulation--is it possible to get 15-20 hours out of these headphones?

 

[mashuto]

Yeah, for the Rift you would need a USB extension cable parallel to the Rift's cable. For wireless models HMDs it will get tricky


The Mobius might support sensing the distance between your ears (read that somewhere as a consideration), which allows the spatialization to more accurately calculate the delay necessary to make it seem like a sound is coming from a certain location.
The Ossic X will definitely support that. In addition, the Ossic X has four small drivers per ear (above, below, in front and behind your ears) that get fed independently. They are used to make the high frequencies (most important for localization) come from where they should (based on the relative location of whatever is making the noise), and the shape of your ear does the rest. The traditional approach has to simulate your ear, which usually means simulating an average ear, since all sound will come from the side. As a result, the Ossic X should be more precise than the Mobius, with both better than what you can do in software alone. Though the Mobius could be matched if you know the distance between the ears.

From what I remember reading about the mobius, I dont think it senses it, but you can maybe input your head size into it (similar to waves nx). But it sounds like the mobius wont have the option to completely take over the virtualization from the HMD, just use the audio it already provides. What I was questioning though is whether thats really a drawback, the virtualization and head tracking already exist with HMD's so wouldnt having separate virtualization and head tracking in your headphones just be redundant?

I also have not heard either the ossic or the mobius since neither is released yet, so I cant say, but I know that those 7.1 headphones with the multiple speakers were all supposed to be fairly gimmicky. And I also know from experience that surround virtualization works extremely well with regular old 1 driver per ear headphones. So, again, people are asking if this can work with VR, and I am just wondering if its even necessary (at least the 3d stuff) since VR headsets already do that for you.

 

[acia]

We have been researching spatial audio for several years now, we have our own ambisonics microphone and we have both recorded and tried to faithfully reproduce it in our headphones. Reproducing multi-channel audio through a pair of headphones, be it for gaming, movies or music works on the same fundamental principles. You are trying to reproduce the experience that you get from mult-speaker setup from a pair of headphones.

Sound localization in a virtual audio scene needs five components:

1. Inter aural time delay -- Depending on the where the sound source is located, there is a small time difference between when our left and right ear hears the sound. Our auditory system is tuned to use this to localize sound. However there are places where left and right ears are equidistant from the sound source (for example directly in frot or directly behind). There can also be locations in the virtual audio scene where the inter aural delay can be exactly the same for two different locations (for example slightly to the right in front and slightly to the right but behind). These regions are called the 'cone of confusion'
2. Inter aural intensity difference -- Our head geometry approximates a sphere and is often 'in the way' causing the sound intensity level for one ear to be lower than the other ear depending on the direction of sound. This is more true for frequencies higher than 1kHz. This is another clue our brain uses to localize sound. However here too, when the sound is directly in front or behind or in the cone of confusion, we have difficulty localizing sound.
3. HRTFs - Our ear-shape, head and shoulders together cause reflections that are asymmetric and directional in nature and modify the sound we hear (both frequency and phase) depending on the location of sound. Our brain is used to this and helps us localize sound. However HRTFs are individual specific. Generic HRTFs can be effective to some degree to provide directional cues but they are not great to disambiguate when the sound source falls in the cone of confusion.
4. Reflections: 1, 2 and 3 above provide information on the direction of sound but not so much on depth. Our brain uses reflections to judge depth but our brain is also smart enough to 'subtract' or ignore these reflections so we do not hear 'echos' when we go about doing our day to day activities.
5. Head movements: We rarely hold our head steady in place in the real world. Head movement is a integral part of both our visual and auditory perception. Without head movements we find it hard to judge visual depth in a 3D space. Without head movements both human and animal subjects find to disambiguate location of sound if the sound source is directly in front, above , below or behind us or in the cone of confusion ( especially in the absence of any visual cues). In the presence of visual cues (such as some activity in the monitor), head movements accentuate the experience giving a larger sense of space. More importantly, being able to track head movements provides an auditory experience that is closer to reality so our brain is not constantly on over-drive to piece conflicting visual and auditory information. As a result the experience becomes immersive and natural, and one forgets you are wearing a headphone, which is exactly the experience we want to provide.

Many 3D audio applications may provide 1,2, 3, and 4 to a varying degree, but without head-tracking to provide a real-time feedback of the relative position of our head, they fall short of providing a truly immersive and believable experience.

One more factor to consider is all the requirements I have listed above rely on precise timing and faithful reproduction in both frequency and time domain. We believe for a immersive 3D experience, a very linear and transparent transducer with negligible distortion and high dynamic range is needed and our planar transducers fit that requirement very well.

i am sold by this detailed explanation alone.

 

[Alcophone]

From what I remember reading about the mobius, I dont think it senses it, but you can maybe input your head size into it (similar to waves nx). But it sounds like the mobius wont have the option to completely take over the virtualization from the HMD, just use the audio it already provides. What I was questioning though is whether thats really a drawback, the virtualization and head tracking already exist with HMD's so wouldnt having separate virtualization and head tracking in your headphones just be redundant?

I also have not heard either the ossic or the mobius since neither is released yet, so I cant say, but I know that those 7.1 headphones with the multiple speakers were all supposed to be fairly gimmicky. And I also know from experience that surround virtualization works extremely well with regular old 1 driver per ear headphones. So, again, people are asking if this can work with VR, and I am just wondering if its even necessary (at least the 3d stuff) since VR headsets already do that for you.

For the Mobius, it might sound the same, so functionally it would be redundant. An HMD's headtracker will also be superior because it usually has an absolute reference (Vive's light house, Rift's cameras), allowing to correct for the drift that gyros invariably exhibit over time. On the other hand, doing it in the headphone might result in better latency, even if it's otherwise doing the same calculations (since you have immediate access to the latest orientation, rather than having to a round trip through the PC). There's also a chance that the DSP can do a better job than what software could do (since the CPU is busy doing all kinds of things). Just like graphics cards can do certain tasks much more efficiently than a CPU could, so that you couldn't reach the same visual fidelity with just a CPU, if the DSP in the headphone did the heavy lifting of spatialization, it's possible that better results could be achieved than just with a CPU. But that's just speculation.

In the Ossic case, I hope what I said before makes it clear why it might have an edge over existing approaches. You can't compare it to 5.1 or 7.1 headsets. The game doesn't know where exactly the corresponding drivers are located in the headset, for example - unless there's an equivalent to the calibration mics used by surround receivers, I don't know. With the drivers so close to your ear, I imagine calibration is even more crucial with those headphones. And if you want to keep the price low, you might not use the best (linear, fast) drivers. And they have to be tiny, yet further away than in an IEM, yet cover a wide frequency range - challenging.
So obviously, you can't just slap extra drivers in there and automatically get better results. In the Ossic case, you have to do a good job separating the frequency bands, calculating how much to play through what driver at exactly the right time, etc. A straightforward 5.1 set would simply direct the existing channels to one or two drivers. Your left ear wouldn't hear the rear right channel at all (again, I don't know how simple they really are). The Ossic will simulate virtual surround speakers, calculate what signal from which virtual speaker would reach the location of each small driver at what point and play that back. Very different approach (with lots of challenges, so might still sound like crap).
And you have to prevent reflections within the earcup, which Ossic hopes to achieve through essentially an anechoic chamber in the earcups. How well they are doing that, I don't know, and whether existing multi driver surround headsets do that, I don't know either.

If existing spatialization is good enough for you, that's great. Maybe you're not very demanding, or maybe the generic models used match your physiology very closely.
I have an HTC Vive, and yes, there is spatialization, and it's good enough to tell roughly where things are on the horizontal plane, but compared to the impressive visual fidelity achievable today (through great effort) the available audio fidelity seems rather primitive. I don't feel like something's actually present in the room and made that sound. I don't feel like I could pinpoint exactly where a sound came from or that I could aim blindly or use a shield blindly. Virtual audio has never felt eerily real to the point of giving me goosebumps. This binaural recording, however, has:



Getting close to that with virtual sound is the dream.

What Ossic is trying to do is in some ways more ambitious. But that's also why Audeze will very likely ship way before Ossic does, with Audeze apparently still setting a new benchmark for 3D Audio, which would be a great achievement. But it's possible that Ossic will surpass them soon after.

I preordered the Ossic X because I want super precise audio in VR, and because I want to get good surround sound when watching movies on the TV, without my wife complaining about the rear speakers being to close (which they are). The Mobius doesn't seem to be well suited for either (unless my computer is the source). So I didn't pull the trigger. Still, it's high on my list of things to try at RMAF, or maybe even CAS? Hope so

 

[Zyker]

You know, I forget that the X7 even has that function. I will try that first thing tomorrow morning. Stay tuned. I have never used that function on the X7 so I don't have any expectations on how it works.

Thanks! I look forward to hearing your thoughts on it... I'm hoping that I'll be able to go completely wireless with it, including the mic.

 

[Panimation]

Have heard various rumours that these are not going to be made in the USA , totally understandable considering the price point. Can anyone from Audeze confirm this?

 

[Mani ATH 87]

I hopped on the bandwagon for these, I expect that they are going to sound great from the early impressions. Can't really go wrong at the price point if they deliver like they say they are going to.

 

[singingbee]

I want one but the technology and features will not work on my switch

 

[acia]

For the Mobius, it might sound the same, so functionally it would be redundant. An HMD's headtracker will also be superior because it usually has an absolute reference (Vive's light house, Rift's cameras), allowing to correct for the drift that gyros invariably exhibit over time. On the other hand, doing it in the headphone might result in better latency, even if it's otherwise doing the same calculations (since you have immediate access to the latest orientation, rather than having to a round trip through the PC). There's also a chance that the DSP can do a better job than what software could do (since the CPU is busy doing all kinds of things). Just like graphics cards can do certain tasks much more efficiently than a CPU could, so that you couldn't reach the same visual fidelity with just a CPU, if the DSP in the headphone did the heavy lifting of spatialization, it's possible that better results could be achieved than just with a CPU. But that's just speculation.

In the Ossic case, I hope what I said before makes it clear why it might have an edge over existing approaches. You can't compare it to 5.1 or 7.1 headsets. The game doesn't know where exactly the corresponding drivers are located in the headset, for example - unless there's an equivalent to the calibration mics used by surround receivers, I don't know. With the drivers so close to your ear, I imagine calibration is even more crucial with those headphones. And if you want to keep the price low, you might not use the best (linear, fast) drivers. And they have to be tiny, yet further away than in an IEM, yet cover a wide frequency range - challenging.
So obviously, you can't just slap extra drivers in there and automatically get better results. In the Ossic case, you have to do a good job separating the frequency bands, calculating how much to play through what driver at exactly the right time, etc. A straightforward 5.1 set would simply direct the existing channels to one or two drivers. Your left ear wouldn't hear the rear right channel at all (again, I don't know how simple they really are). The Ossic will simulate virtual surround speakers, calculate what signal from which virtual speaker would reach the location of each small driver at what point and play that back. Very different approach (with lots of challenges, so might still sound like crap).
And you have to prevent reflections within the earcup, which Ossic hopes to achieve through essentially an anechoic chamber in the earcups. How well they are doing that, I don't know, and whether existing multi driver surround headsets do that, I don't know either.

If existing spatialization is good enough for you, that's great. Maybe you're not very demanding, or maybe the generic models used match your physiology very closely.
I have an HTC Vive, and yes, there is spatialization, and it's good enough to tell roughly where things are on the horizontal plane, but compared to the impressive visual fidelity achievable today (through great effort) the available audio fidelity seems rather primitive. I don't feel like something's actually present in the room and made that sound. I don't feel like I could pinpoint exactly where a sound came from or that I could aim blindly or use a shield blindly. Virtual audio has never felt eerily real to the point of giving me goosebumps. This binaural recording, however, has:



Getting close to that with virtual sound is the dream.

What Ossic is trying to do is in some ways more ambitious. But that's also why Audeze will very likely ship way before Ossic does, with Audeze apparently still setting a new benchmark for 3D Audio, which would be a great achievement. But it's possible that Ossic will surpass them soon after.

I preordered the Ossic X because I want super precise audio in VR, and because I want to get good surround sound when watching movies on the TV, without my wife complaining about the rear speakers being to close (which they are). The Mobius doesn't seem to be well suited for either (unless my computer is the source). So I didn't pull the trigger. Still, it's high on my list of things to try at RMAF, or maybe even CAS? Hope so

Life cycle of these toys counted in months not years it seems to me.


Timing of the release of products may be more important than superiority of technology.

 

[Lobaba]

Will this be demo-ed at can jam 2018? can we go and try?

 

[KMann]

The Mobius might support sensing the distance between your ears (read that somewhere as a consideration), which allows the spatialization to more accurately calculate the delay between both channels necessary to make it seem like a sound is coming from a certain location.

Since most of us carry just a single head over our shoulders, we prefer the simplicity of measure once, enter it in the app and store it on Mobius till one feels the need to change it. It helps keep the design simple and not add to the cost. This would cover a majority of use cases.

Regarding multi-driver approach, vs single driver:

• Yes high frequencies have directional cues both because of Inter aural intensity difference (head shadowing) and shape of our outer ear and ear canal. Due to both of these factors, the higher frequencies (greater than 1kHz) are affected to varying degree based on the direction of sound. Whether one uses a single driver or multiple drivers, what reaches the ear drum should feel like sound waves that went through a HRTF. I.e, our ear or brain does not care if it came from four drivers or single driver or N drivers as long as all the signals add up to represent something similar to what our head, shoulder and ear geometry do to the sound.
• When using multiple drivers, it is very hard to predict how the sound form the drivers interfere with each other, they can cancel or accentuate. Since everyone's ear geometry is different, it is very hard to predict and model how this interaction would be and compensate for the interaction to make them all sound coherent together. With a single large, full range planar driver, this is a non issue. Also when a single large driver is replaced by several small ones, base reproduction and perception is affected.
• When trying to emulate 3D sound , some form of HRTF modeling is used, then this HRTF model is shot through our outer ear which again adds your own HRTF on top of the simulated HRTF. It is very important to not have this 'double HRTF' else it will result in an unnaturally bright sound signature. Being able to reasonably estimate what the the transducers transfer transfer function would sound to an average ear and remove the effect is very important for providing a 3D audio sound that does not sound colored. Having multiple drivers that may be tuned differently and be provided different signals will make it very hard to estimate and remove the second HRTF created by the drivers.

Please note that my comments are regarding multi-driver approach in general and not with respect to a specific implementation.

 

[Alcophone]

Since most of us carry just a single head over our shoulders, we prefer the simplicity of measure once, enter it in the app and store it on Mobius till one feels the need to change it. It helps keep the design simple and not add to the cost. This would cover a majority of use cases.

Agreed, mostly. Though I'm not sure how I would measure it without some special contraption for that purpose. And the live sensing approach has advantages when sharing the headphone, but not many people will do that often, so it's fair. Would help you demo it, though

Regarding multi-driver approach, vs single driver:

• Yes high frequencies have directional cues both because of Inter aural intensity difference (head shadowing) and shape of our outer ear and ear canal. Due to both of these factors, the higher frequencies (greater than 1kHz) are affected to varying degree based on the direction of sound. Whether one uses a single driver or multiple drivers, what reaches the ear drum should feel like sound waves that went through a HRTF. I.e, our ear or brain does not care if it came from four drivers or single driver or N drivers as long as all the signals add up to represent something similar to what our head, shoulder and ear geometry do to the sound.

Right. The tricky part being getting an accurate HRTF for the user. It's pretty clever of Ossic to mostly side step the issue of the complex ear geometry by using the ear itself as part of the HRTF. Except that the actual sounds can only come from four fixed locations per ear, so that's still a compromise that requires compensation (using the top and front drivers simultaneously for sounds coming from, say, 45° above, which could very well sound off since it's hitting different parts of the ears first).

• When using multiple drivers, it is very hard to predict how the sound form the drivers interfere with each other, they can cancel or accentuate. Since everyone's ear geometry is different, it is very hard to predict and model how this interaction would be and compensate for the interaction to make them all sound coherent together. With a single large, full range planar driver, this is a non issue. Also when a single large driver is replaced by several small ones, base reproduction and perception is affected.

Makes sense. I hope eventually there will be a video about the challenges they had to overcome.
For bass, the Ossic X has a fifth, larger driver in each earcup for the bass... somewhere around 50mm if I remember correctly, can't find it right now.

• When trying to emulate 3D sound , some form of HRTF modeling is used, then this HRTF model is shot through our outer ear which again adds your own HRTF on top of the simulated HRTF. It is very important to not have this 'double HRTF' else it will result in an unnaturally bright sound signature. Being able to reasonably estimate what the the transducers transfer transfer function would sound to an average ear and remove the effect is very important for providing a 3D audio sound that does not sound colored. Having multiple drivers that may be tuned differently and be provided different signals will make it very hard to estimate and remove the second HRTF created by the drivers.

Yeah, I was wondering how you dealt with that. Since you don't have a personalized HRTF (unless I missed something), the deviation hits you twice. Have you considered an approach based on the iSINE, just also with cups to house battery, electronics, etc., or a separate box? No ear geometry to worry about, and more precise control over the distance to the eardrums than large, soft pads.
Not as comfortable, for sure, that's why I wouldn't want the nuraphone, for instance.

Please note that my comments are regarding multi-driver approach in general and not with respect to a specific implementation.

Noted. I'm still focusing on the Ossic X since it's the one I know most about.

 

[KMann]

Agreed, mostly. Though I'm not sure how I would measure it without some special contraption for that purpose. And the live sensing approach has advantages when sharing the headphone, but not many people will do that often, so it's fair. Would help you demo it, though

Inter aural delay and inter aural intensity difference are a function of the distance between your ears and the approximate head size, it is not rocket science and all we need is a simple measurement with a measuring tape to compute the right values.

Right. The tricky part being getting an accurate HRTF for the user. Except that the actual sounds can only come from four fixed locations per ear, so that's still a compromise that requires compensation (using the top and front drivers simultaneously for sounds coming from, say, 45° above, which could very well sound off since it's hitting different parts of the ears first).

Multiple drivers within a very confined space capture only a very little portion of what we experience in real life even if the multiple drivers are placed in some extreme angles. For example, how would one capture the HRTF of a center channel directly in front or a rear channel directly behind. Even then it will not include the effect of head and torso.

Makes sense. I hope eventually there will be a video about the challenges they had to overcome.
arger driver in each earcup for the bass... somewhere around 50mm if I remember correctly, can't find it right now.

when the frequency response is split over multiple drivers, a crossover that is transparent is needed, which will be challenge with multiple drivers.

Yeah, I was wondering how you dealt with that. Since you don't have a personalized HRTF (unless I missed something), the deviation hits you twice. Have you considered an approach based on the iSINE, just also with cups to house battery, electronics, etc., or a separate box? No ear geometry to worry about, and more precise control over the distance to the eardrums than large, soft pads.
Not as comfortable, for sure, that's why I wouldn't want the nuraphone, for instance.

The compensation we apply is based on experience and uses both measurements and critical listening and works well with the 3D emulation algorithms. The DSP presets and 3D emulation works very well for localization in a gaming or movie environment. For a purist perspective for music listening (which includes me) who values tonality more than emulation for critical listening, switching off 3D would give you a neutral sound signature tailored to your ear as the only HRTF involved is your own.

(Edited for clarity)