AVS Forum banner

1 - 20 of 178 Posts

·
Registered
Joined
·
1,824 Posts
Discussion Starter #1 (Edited)
I'd like to share my experience in building a virtual surround sound system. There's constantly questions about what headphones to use for movies, and everything save for one expensive device that isn't out yet seems rather lackluster. I've used lots of software and equipment and the end result of being able to put on a pair of headphones and have such natural sounding high fidelity external virtualization of a 7.1 loudspeaker system is truly marvelous. A front soundstage anchored to the display, clearly rendered and discrete side and rear surround channels, long resolved pans, every bit of the most complex surround movie mixes(looking at you Ready Player One race scene) rendered in full detail over 2 tiny headphone drivers. With the added benefit of providing a virtual 2 channel loudspeaker system for music listening, as well as being able to change the room you're in with the push of the button, no phoney reverb effect, this is the real deal, treated studios, untreated listening rooms, home theaters, huge movie theaters.

Research and development of room impulse responses and binaural synthesis has been done for decades, and is finally cultivating in consumer products. So if you see anyone claiming headphone surround sound is a gimmick based soley on a budget gaming headset, point them to the plethora of scholarly articles over the past 20 years on the subject.

The following are all modeled and not using measurements:
Dolby Headphone / Atmos for Headphones
Windows Sonic
Creative CMSS 3D / SBX
Razer Surround
DTS Headphone X
Toneboosters Isone
Flux Hear
Sennheiser GSX
Dspeaker headspeaker
Waves NX and the Audeze Mobius

1) Darin Fong's Out of Your Head, and now Impulcifer as well, are the only products aside from the Smyth Realisers which use measured room impulse response applied to the binaural processing. The measurements of movie theaters, home theaters, recording studios, or hifi listening rooms were made using the Smyth Realiser A8 or your own head and speakers with Impulcifer.

Since the program is a stand alone piece of software, parts of the Realiser aren't used, like headphone correction. So this must be done manually using something like Equalizer APO, to equalize the headphones to have a flat response. This is very important and something I don't think alot of people do when they try OOYH or other HRTF based virtualization, and ends up sounding very bright/grating/noisy in the 2-6khz region due to double HRTF, but after the EQ, sounds very natural. Of course something with a relatively flat uncompensated response could be used without EQ, but anything that follows the Harman curve you'll want to bring 2-6khz down ~5-10db.

Impulcifer has auto eq measuring which can be applied to OOYH as well.

2) When using these programs, I've found headphones with more responsive drivers render the virtual rooms better. Planar magnetic and electrostatic drivers being the best. Anything with a fast response, little to no treble ringing(good imaging), and low distortion. Soundstage ability is important as well for better externalization of the virtual speakers. Open headphones are great, deep angled pads, etc. e.g. the Sennheiser HD800. I use a pair of Audeze EL-8s, MrSpeakers Alpha Primes, Stax SR-207, Hifiman HE-400i, Sennheiser HD650, and I also have a pair of Audeze Mobius. The EL8s work best for me with a pad mod(I use the Dekoni LCD Velour pads, they reduce a bit of treble ringing and push the drivers out further), the Alpha Primes muster a nice room size despite being closed, the Stax have scary realism but sound very close(less soundstage / small room size).

3) Lossless movie audio bitrate is 10-15mb/s, doubled for the binaural rendering and room reflections, crammed into a 48khz container. Due to the large amount of data and complex waveforms from lossless 7.1 audio, binaural rendering, and measured room reflectioms, the circuitry makes a huge difference in rendering the audio. I have several devices I've tried, ranging from the standard motherboard Realtek chip, the dac and headphone out in a Denon X4200W, a JDS Labs ODAC/O2, an Audio-GD NFB-11, and an Audio-GD NFB-7. Using the computer motherboards headphone output results in a neat surround sound effect with little fidelity, scaling up to the NFB-7, which renders the virtual loudspeakers with incredible realism and accuracy, especially with the more reverberant rooms, due to the undistorted transients of the non feedback amplifiers and the power delivery of the double stage discrete regulation.

4) Digital and power conditioning improves realism further, improves fidelity substantially, and reduces listener fatigue. I use a device from Uptone Audio called the ISO regen, which isolates from the PC, regenerates and clocks the data, with ultra low noise voltage regulation, an Uptone Ultracap LPS powers the iso regen and the usb to i2s conversion board in the dac, as well as isolating both from the ac mains, and a Topaz Ultra-Isolator Line Noise Supressor, used to isolate the dac and amp from the ac mains, these were primarily used for scientific measuring equipment in labs. There are posts from the early 2000s on this forum of people using the isolation transformers as affordable power conditioning.

5) Add a bass shaker system, a small 20w shaker carefully attached to a chair frame really helps with immersion for special effects and improving extwrnalization due to not only feeling bass on your head. I duplicate the audio with Voicemeter so the shaker system is separate from the headphone system. When properly dialed in sometimes between the shaker and headphones that can play low(20hz) I'll get shocking sub bass hits that feel suprisingly natural like actual subwoofers, chest hits and all.


So to some it up, I feel OOYH and Impulcifer are the best headphone surround programs available at the moment, except of course for the Realiser. The measurements of rooms(PRIR, personal room impulse response) makes the biggest difference when compared to other options. The programs scale very well with better equipment, but isn't necessary, just make sure the headphones you're using play flat(HPEQ, headphone equalization to avoid a double HRTF and harshness in the 2-6khz range). Using open headphones with fast responsive, low distortion drivers and good imaging as well as using hifi DACs, amps with ample power supplies to render all the data, and conditioning devices to greatly improve the realism of the virtual room presets, if that's what you're after.

The program is free to try, a "gaming" preset without room measurements has an unlimited demo for a week and costs $25 by itself, the remaining presets can be demod in 2 minute intervals(after 2 minutes you only need to select a different preset and go back to the same one). The license includes 1 preset for $150, and presets are $25, and often goes on sale twice a year for half off.

For impulcifer you need to buy in ear microphones and have speakers to measure.

Unfortunately using a PC is necessary to get the best surround sound except for the Smyth Realiser A16.
The Dspeaker Headspeaker is no longer available, but you still may be able to find demo units or used units. This was the next best option to the Realiser for non pc use. It has an optical input so only lossy(dolby digital ac3 or dts) decoding though. It used modeled algorithms instead of measurements but included a head tracking device and was able to have much better externalization.
-Sony MDR-HW700DS, is a wireless device with hdmi input and can decode lossless audio but is limited to a single pair of headphones and has hdmi 1.4 ports so no full 4k passthrough, however, you can use something like the AVR Key to pass the audio separately to the 700ds' base. The surround sound is good but the externalization not as much. Theres also the optical based DS6500 which performs similar but with less fidelity due to the lossy audio.
-Some gaming sets and dac/amp units have an optical input, can accept dd/dts lossy 5.1, and built in surround processing, like SBX, dts headphone x, etc. With some of these you can at least use a better pair of headphones, however, they won't have the most external of virtual speakers but it will have more surround than a stereo headset.
-Creative is developing the SXFI TV and wireless headphones which should be a better option than what's available for non pc users. https://www.hardwarezone.com.sg/feature-sneak-peek-creative-sxfi-theater-headphones-and-creative-sxfi-tv

If youre more interested in headtracking and anchoring the virtual speakers than measured room imupulse, options include.
Redscape Audio
Waves NX
or the Audeze Mobius

Full system for reference:
Transport: Supra USB, Uptone ISO Regen, Uptone Ultracap LPS 1.2
DAC: Audio GD NFB 7, amanero usb to i2s board with isolator on i2s output.
Amp: Audio GD NFB-1AMP
Power: Topaz 91092-31 Ultra Line noise Isolator
Shaker system: duplicated signal with voicemeter, usb to dac, equalizer, crossover, amplifier, 2 shakers
Headphones: Mrspeakers Alpha Prime, Audeze EL-8, Sennheiser HD650, Hifiman HE-400i, Stax SR-207(with SRM252S energizer)

Additional commentary on equipment at post #100 :
https://www.avsforum.com/forum/378-headphones/3034902-headphone-surround-sound-4.html#post57949110
 

·
Registered
Joined
·
26 Posts
This is a very interesting topic for surround headphone enthusiast.

although im not very familiar with some terms and devices , i subbed to the thread to track further developments.

thanks for sharing and keep up the good work.
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #3
This is a very interesting topic for surround headphone enthusiast.

although im not very familiar with some terms and devices , i subbed to the thread to track further developments.

thanks for sharing and keep up the good work.
If you have any questions feel free to ask.
 

·
Registered
Joined
·
1,461 Posts
Hi! Trying now the free samples of OOYH I'm interested. The beginning of "Money" by Pink Floyd is impressive.


I'm using Beyerdynamics DT-880 and I use the Equalizer APO with satisfaction for my PC desktop monitors (nothing special, some Audioengine small monitors). I own a Umik calibrated microphone but... I'm not sure how I could equalize the DT-880 to flat. How can I measure them?


Being a fan of surround music and a headphone enthusiast, I've always been interested in replicating the experience through headphones so... you have my attention. :)
Sooo... questions:


1) How to make headphones flat? Where do I get the measurements, since I can't use a regular microphone to do that through REW.


2) Does OOYH allow to pre-process sound? I ask this because of portable use. There are parametric equalizers for Android (the platform I use) but the program itself runs under either Windows or MacOS. If I could pre-process stuff, I could then transfer it on my smartphone and have proper multichannel playback on the go (this has been one of my wildest tech dreams since forever). In the FAQ I see they mention Android version coming, but there's no release date.


3) I don't understand if the program allows one to measure one's own room or if you are bound to use the presets provided with it. Which one is it? Reading the website it looks to be the latter, I guess.
Yeah, you are bound to use their presets. Some of them are quite good, though. I'm particularly fond of the Magico Q7 speakers.


4) Apart from the ability to measure one's room and head tracking, is this basically Realiser for $150? Do I get this right? Because having had a look at the presets available, that's the impression I get. Have you ever had the chance to test Realiser, if so how does this compare? Edit: further reading, there don't appear to be Atmos presets, so there's that.


5) Looking at presets, I'd be interested, for reasons expressed above, in buying a 7.1 preset. The question is: what happens when source is 2.0, 4.0 or 5.1? Unless I add extra DSP only the speakers with signal play back, correct? I hope it's like that.
It's like that, I tested with the trial.
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #5 (Edited)
1) How to make headphones flat? Where do I get the measurements, since I can't use a regular microphone to do that through REW.
Two ways.
1.) Listen to a frequency sweep, most likely the headphones will play fairly flat until 1-2khz, and then rise about 10db, this is the area you want to attenuate.
2.) Several sites have made measurements of headphones, I'm particularly fond of Innerfidelity's. All 3 880s have been measured.
https://www.innerfidelity.com/headphone-measurements


2) Does OOYH allow to pre-process sound? I ask this because of portable use. There are parametric equalizers for Android (the platform I use) but the program itself runs under either Windows or MacOS. If I could pre-process stuff, I could then transfer it on my smartphone and have proper multichannel playback on the go (this has been one of my wildest tech dreams since forever). In the FAQ I see they mention Android version coming, but there's no release date.
Yes you can use programs to capture the output of OOYH. I personally haven't tried this yet, but I was talking with someone recently who really enjoyed converting the track to DSD after running it through OOYH and playing it back on an Ibasso DX200 and Audeze LCDi4.
I believe the programs he used were IShowU for capture and Audacity, and Xivero XiSRC converter


3) I don't understand if the program allows one to measure one's own room or if you are bound to use the presets provided with it. Which one is it? Reading the website it looks to be the latter, I guess.
Yeah, you are bound to use their presets. Some of them are quite good, though. I'm particularly fond of the Magico Q7 speakers.
The measurements for the presets were made using a Smyth Realiser A8.


4) Apart from the ability to measure one's room and head tracking, is this basically Realiser for $150? Do I get this right? Because having had a look at the presets available, that's the impression I get. Have you ever had the chance to test Realiser, if so how does this compare? Edit: further reading, there don't appear to be Atmos presets, so there's that.
Correct, it's basically just the convolution engine of the Realiser A8, the Realiser also employs the headphone correction(HPEQ) which is why I eq separately. I haven't had a chance to test the Realiser yet, expensive Dx. I'm hopeful that once the A16 comes out Darin will eventually update the program for Atmos use, the problem currently is that it would need to decode the Atmos track as well which would increase cost with licensing etc. If you can find it, test the Atmos Encounter Demo, height effects render suprisingly well.


5) Looking at presets, I'd be interested, for reasons expressed above, in buying a 7.1 preset. The question is: what happens when source is 2.0, 4.0 or 5.1? Unless I add extra DSP only the speakers with signal play back, correct? I hope it's like that.
It's like that, I tested with the trial.
Yup, I use this for virtual stereo loudspeaker music listening all the time when I'm at my desktop setups now. There are ways to upmix 2.0 and 5.1 to 7.1, some cheap and average, some complicated and expensive, I tried the matrixing in FFDShows audio filter but didnt find it to convincing.
 

·
Registered
Joined
·
1,461 Posts
Thanks a lot for your replies.


As far as equalizing my DT-880, these are the measurements from the Inner Fidelity website. Problem is I really would not know what to input in APO Equalizer. It was "easy" through REW handling the measurements, as that allowed for outputting a text file that you could feed APO Equalizer. I don't know about going in blindly inputting frequencies and quality values :( Are there any tips for doing this? Sorry if I sound whining, it's not my intention but after listening to OOYH for a while... I'd really like to take advantage of it at its maximum level.



I'll investigate about iShowU alternatives on Windows, no MacOS here. :)


Is Darin Fong connected to Smyth Research? From what you write about Atmos support I'm led to believe so but I might be mistaken in interpreting what you are writing.


I'm perfectly fine with listening 2.0 stuff in stereo, automatic upmixing leaves a lot to be desired. And I have a collection of over 200 albums in real surround to listen to, so I'm ok with this. :)


Just one downside in finding out about this now. I've missed the Black Friday sale. :D
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #7 (Edited)
Thanks a lot for your replies.
You're welcome! I really enjoy the program and want to share, I always see people asking about headphone surround for movies and this is hands down the best option next to the Realiser.


As far as equalizing my DT-880, these are the measurements from the Inner Fidelity website. Problem is I really would not know what to input in APO Equalizer. It was "easy" through REW handling the measurements, as that allowed for outputting a text file that you could feed APO Equalizer. I don't know about going in blindly inputting frequencies and quality values :( Are there any tips for doing this? Sorry if I sound whining, it's not my intention but after listening to OOYH for a while... I'd really like to take advantage of it at its maximum level.
Not at all. You dont know something until you know it ;] I'd be more than happy to write a couple filters later tonight that you can input and test. The simple jist is use the parametric peaking filters. So when you see the peak on the dt880s at 3.5 khz the filter would be:
Frequency: 3500hz; level: -5db; q factor 2

I tend to use a couple filters at the same level so I can easily turn them on and off to dial the levels in to where I want. So I'll have 2-3 filters every 1000hz starting at 2500hz of about 4db with a qfactor of about 2.5

The lower the q factor the wider the frequency range filtered, the higher the narrower.

Is Darin Fong connected to Smyth Research? From what you write about Atmos support I'm led to believe so but I might be mistaken in interpreting what you are writing.
I don't believe so, he just wrote(or had someone write) a convolution engine and used the PRIR taken from the Realiser.

In regards to Atmos support, what I mean is currently the only way of decoding Atmos is in an AVR with the built in decoders. On Windows for instance, there's no way of decoding the track and sending a 16 channel pcm track with all of the object data to ooyh, or anywhere else for that matter. The Realiser A16 will take atmos or dts x bitstreams and do all the decoding itself.

Just one downside in finding out about this now. I've mixed the Black Friday sale. :D
You know what, I don't think it happened this year, or I missed it too and there were a few presets I wanted. Dx
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #8
As far as equalizing
Here's the filters I use on a pair of HE-400is as well as the innerfidelity measurement, and what the EQ looks like after all the filters.

With EQAPO it's nice and easy to hit those power buttons for the filters on and off while listening to music and get it just right. These kind of filters are fairly similar for what I use for different headphones, just different combinations and q factors.


:::peaking Filters:::
Frequency: 2500hz
Gain: -3db
Q Factor: 3

Frequency: 3500hz
Gain: -3db
Q Factor: 2

Frequency: 3500hz
Gain: -4db
Q Factor: 2.5

Frequency: 4500hz
Gain: -3db
Q Factor: 2

Frequency: 4500hz
Gain: -3db
Q Factor: 2

Frequency: 4500hz
Gain: -3db
Q Factor: 2

Frequency: 5000hz
Gain: -3db
Q Factor: 2

Frequency: 9000hz
Gain: -4db
Q Factor: 2

:::High Shelf:::
Frequency: 8000hz
Gain: -3db
Fixed
 

Attachments

·
Registered
Joined
·
1,461 Posts
Not at all. You dont know something until you know it ;] I'd be more than happy to write a couple filters later tonight that you can input and test. The simple jist is use the parametric peaking filters. So when you see the peak on the dt880s at 3.5 khz the filter would be:
Frequency: 3500hz; level: -5db; q factor 2
I've already lost you here. Because if I look at the Innerfidelity chart for the DT-880 250ohm I see it doesn't peak at 3500Hz, it's actually down to -5db. It's flat from 400Hz till 1000Hz, then it starts going down, with some up and downs but always below 0db, until 7500 where it has a +3.5 (more or less) at 7500. Am I reading the graph wrong? Or is my reasoning wrong and I should not aim for 0db?
I don't believe so, he just wrote(or had someone write) a convolution engine and used the PRIR taken from the Realiser.
Oh. So he's got an A8 from Smyth and he's using that to create the different presets.
In regards to Atmos support, what I mean is currently the only way of decoding Atmos is in an AVR with the built in decoders. On Windows for instance, there's no way of decoding the track and sending a 16 channel pcm track with all of the object data to ooyh, or anywhere else for that matter. The Realiser A16 will take atmos or dts x bitstreams and do all the decoding itself.
Yeah... as far as I understand it, Atmos does not have discrete channels. The sound is "distributed" among the channels according to speaker placement and calibration. So you are correct, the software should do the decoding and then distribute the sound throughout the "virtual" channels.
You know what, I don't think it happened this year, or I missed it too and there were a few presets I wanted. Dx
You're right. It happened in 2017. In 2018 he did it for CanJam SoCal.
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #10
Am I reading the graph wrong? Or is my reasoning wrong and I should not aim for 0db?
The red graph is compensated . Grey is uncompensated and I find that easier to work with. Mind me asking what they're plugged into? Pc, dac/amp, avr, etc.
 

·
Registered
Joined
·
1,461 Posts
The red graph is compensated . Grey is uncompensated and I find that easier to work with. Mind me asking what they're plugged into? Pc, dac/amp, avr, etc.
Usually they're connected to the AVR in my signature. RX-A3010, Yamaha. I used to listen to them on a dedicated amp in my office but now that I've moved to home office I've opted for desktop monitors for PC use.
 

·
Registered
Joined
·
1,461 Posts
I hope not to be straying too much offtopic in turning this into a "Equalizer APO How To" but... on desktop I'm more likely to use my run of the mill AKG K450. Experimenting with the Inner Fidelity measurements and Peter's Equalizer APO Configuration Extension (Peace) I "inverted" the frequency response chart from Inner Fidelity and came up with this equalization. I screengrabbed them one on top of the other for your evaluation.




Do point out all the tremendous mistakes I surely must have done. After uploading the picture I realized that I had two 1.000 Hz bands and I eliminated one, while working on the other to come up with a similar result.
What do you think?


 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #13
What do you think?
It's an interesting thought, but you're boosting the frequencies I'd recommend you reduce. Try just a single peaking filter, 3500hz, -10db, q factor 3.5 to 4.

Or if you're using a graphic eq, you would reduce 3500 hz by 10db, 3000hz and 4000hz by 7db,and 2500hz and 4500hz by 4db.

You typically want to avoid boosting when eqing because it can cause clipping and/or increased distortion. Looking at the grey graph of the K450, the main thing you want to adjust when using OOYH is the peak at 3.5khz.
 

·
Registered
Joined
·
1,461 Posts
It's an interesting thought, but you're boosting the frequencies I'd recommend you reduce. Try just a single peaking filter, 3500hz, -10db, q factor 3.5 to 4.

Or if you're using a graphic eq, you would reduce 3500 hz by 10db, 3000hz and 4000hz by 7db,and 2500hz and 4500hz by 4db.

You typically want to avoid boosting when eqing because it can cause clipping and/or increased distortion. Looking at the grey graph of the K450, the main thing you want to adjust when using OOYH is the peak at 3.5khz.
Yeah, I have a -16db preamplification applied, to avoid clipping. I'll try your suggestion, I use APO so it's no problem implementing the single filter at 3.500 you suggest.


Tried, as it's fast enough to do. Obviously it sounds a lot different, muddier I'd say... but it's better if I take a break and come back to it later. With the above filters I tried approximating the ideal curve explained on Inner Fidelity as much as possible but maybe I went it the wrong way. Actually, what you kindly define as an "interesting thought" might be an idiotic thing to do. Don't restrain yourself, tell it as it is! I'm here to learn. :)


Actually, if you feel this is too specific or offtopic, don't hesitate to tell me so. Thanks for all the help so far. :)
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #15
Yeah, I have a -16db preamplification applied, to avoid clipping. I'll try your suggestion, I use APO so it's no problem implementing the single filter at 3.500 you suggest. Tried, as it's fast enough to do. Obviously it sounds a lot different, muddier I'd say.
The single filter sounded muddier? Muddy usually doesn't refer to that frequency range, darker would be a better description. Just to clarify, the preamp and the other eq were all off and you were using ooyh?
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #16
And this might sound silly, but its happened before, did you turn the volume back up to a comfortable level? After eqing 3-4khz down so much its much quieter and you can increase the volume.
 

·
Registered
Joined
·
1,461 Posts
Yeah, sorry. I'm Italian, English is not my first language. So darker might be more appropriate. Yes, I did raise the volume. But no, I wasn't evaluating it with OOYH, just with regular stereo music. I'll experiment some more and come back to you. Thanks again.
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #18
Yeah, sorry. I'm Italian, English is not my first language. So darker might be more appropriate. Yes, I did raise the volume. But no, I wasn't evaluating it with OOYH, just with regular stereo music. I'll experiment some more and come back to you. Thanks again.
Ah k, using that eq without ooyh will sound awful. ;]
 

·
Registered
Joined
·
1,461 Posts
Ah k, using that eq without ooyh will sound awful. ;]
What I'd like to understand is toward what one is supposed to equalize to. I see you state that the Harmon curve shouldn't be followed. Correct? To the Harmon curve one should apply approximately -10db of correction in the area between 2kHz and 4kHz, right?
So for regular "musical" use one should equalize toward the Harmon curve and then have a different APO profile with -10db approximately applied between 2kHz and 4kHz to the regular "musical" profile. Is that what you meant?
 

·
Registered
Joined
·
1,824 Posts
Discussion Starter #20
What I'd like to understand is toward what one is supposed to equalize to. I see you state that the Harmon curve shouldn't be followed. Correct? To the Harmon curve one should apply approximately -10db of correction in the area between 2kHz and 4kHz, right?
So for regular "musical" use one should equalize toward the Harmon curve and then have a different APO profile with -10db approximately applied between 2kHz and 4kHz to the regular "musical" profile. Is that what you meant?
Only when using Out of Your Head do I find it necessary to flatten out that 2-5khz peak.

For eqing for normal listening its entirely up to you how to eq the headphones, theres no right or wrong. If you find them to be to bright, try pulling down that same area around 2-4khz by 2-5db, if you find them bass light use a low shelf under 100hz, if you find them piercing/sharp use a high shelf over 8khz.
 
1 - 20 of 178 Posts
Top