STAryo, thank-you for checking with Ralph. Wow, 72 channels is ambitious. I'm working towards 32 and thought that was a lot!
FWIW, I had an ambiophonics trial a few years ago and seem to remember there being a delay. My hope is regardless of ambiophonic software or VST plugin, the total processing delay is less than 2 seconds.
FWIW, I had an ambiophonics trial a few years ago and seem to remember there being a delay. My hope is regardless of ambiophonic software or VST plugin, the total processing delay is less than 2 seconds.
It will be a dual system with 7.2.4 Atmos/DTS:x for video and stereo/ambiophonics 2.0/4.0 for music. So in music mode, 8 speakers will be used for ambiance. I need 32 channels for all the active crossovers. I would do more channels, but I'm currently limited at 32 with DACs.
It isn't quite set up yet as I'm working on a solution to delay video for ambiophonic, FIR crossover, and DRC latency and I need another amplifier.
Any word on ambiophonic latency? I don't need an exact time, just an estimate. The video delay hardware/software requirements for audio sync vary depending on how much delay is needed (e.g., buffer frames in VRAM, RAM, or hard disk).
If the latency is more than a couple seconds, that can get very expensive. If the delay is a second or less, it shouldn't be a problem.
It isn't quite set up yet as I'm working on a solution to delay video for ambiophonic, FIR crossover, and DRC latency and I need another amplifier.
Any word on ambiophonic latency? I don't need an exact time, just an estimate. The video delay hardware/software requirements for audio sync vary depending on how much delay is needed (e.g., buffer frames in VRAM, RAM, or hard disk).
If the latency is more than a couple seconds, that can get very expensive. If the delay is a second or less, it shouldn't be a problem.
Ralph has not responded yet. Meanwhile, Hotto did provide some guide where he said latency depends on buffer size. Ambiophonic Transcoder .
Meanwhile, are you using convolution engine? I use SIR2 and the 360 degrees IR is st Cecilia concert hall.
Meanwhile, are you using convolution engine? I use SIR2 and the 360 degrees IR is st Cecilia concert hall.
Ralph’s reply:-
The latency is minimum. I have never had a problem with video lip sync. The latency is on the order of milliseconds, since the direct sound samples are not processed in RACE only copied and reused like concert hall reflections. However there is latency in the buffers that feed whatever program is running RACE. So you need to use as small a buffer size as the computer will permit. Since there is also video delay, you can have the sound earlier than the video.
The latency is minimum. I have never had a problem with video lip sync. The latency is on the order of milliseconds, since the direct sound samples are not processed in RACE only copied and reused like concert hall reflections. However there is latency in the buffers that feed whatever program is running RACE. So you need to use as small a buffer size as the computer will permit. Since there is also video delay, you can have the sound earlier than the video.
Ok, dragging up an old thread, but thought experiment time.
I ran across a cool HRIR measurement package a while ago (GitHub - jaakkopasanen/Impulcifer: Measurement and processing of binaural impulse responses for personalized surround virtualization on headphones.). The primary purpose of the software is basically the same as the Smyth realizer - use binaural in-ear mics to measure the response of your speakers in-room, and then reproduce that over headphones via multi-channel convolution filters.
I've had the idea for a while that headphones 'should' be an ideal approach to ambiophonics since the crosstalk is already removed. If you could just recreate the hrtf/hrir and some type of reverberant field, you should get close to what ambiophonics offers.
So, it's pretty easy to tweak the code, and I made binaural measurements of a single center channel, and processed them to give the in-room response of the speaker to each ear appropriately processed via HRIR. Loaded them into JRiver so that the Left/Right channels were filtered by the appropriate in-room hrir measurement.
The result was......disappointing for the most part. The effect on center-anchored vocals was fantastic - nicely externalized, and very realistically rendered. Unfortunately anything off-center either L or R definitely did NOT exhibit the mind-blowing ambio spatial effect - it was basically a horseshoe which collapsed back to be close to the typical headphone experience at the R/L limits.
I have to admit I'm left a bit puzzled by this - it seems that this really should capture the theoretical behavior of a front dipole, but obviously does not. I am planning to try a 2-channel setup with the more typical 20-degree separation, but this approach has always struck me as a pragmatic concession to the XTC math rather than a goal and so I don't really expect an improvement. The other possibility is that the captured IR is too short to fully recreate the experience (under 1 second), but I think extending the length of the impulse will take a bit more work.
My question is basically "what am I missing?". How can/should this process be altered/augmented to get closer to the free-field ambio effect?
I ran across a cool HRIR measurement package a while ago (GitHub - jaakkopasanen/Impulcifer: Measurement and processing of binaural impulse responses for personalized surround virtualization on headphones.). The primary purpose of the software is basically the same as the Smyth realizer - use binaural in-ear mics to measure the response of your speakers in-room, and then reproduce that over headphones via multi-channel convolution filters.
I've had the idea for a while that headphones 'should' be an ideal approach to ambiophonics since the crosstalk is already removed. If you could just recreate the hrtf/hrir and some type of reverberant field, you should get close to what ambiophonics offers.
So, it's pretty easy to tweak the code, and I made binaural measurements of a single center channel, and processed them to give the in-room response of the speaker to each ear appropriately processed via HRIR. Loaded them into JRiver so that the Left/Right channels were filtered by the appropriate in-room hrir measurement.
The result was......disappointing for the most part. The effect on center-anchored vocals was fantastic - nicely externalized, and very realistically rendered. Unfortunately anything off-center either L or R definitely did NOT exhibit the mind-blowing ambio spatial effect - it was basically a horseshoe which collapsed back to be close to the typical headphone experience at the R/L limits.
I have to admit I'm left a bit puzzled by this - it seems that this really should capture the theoretical behavior of a front dipole, but obviously does not. I am planning to try a 2-channel setup with the more typical 20-degree separation, but this approach has always struck me as a pragmatic concession to the XTC math rather than a goal and so I don't really expect an improvement. The other possibility is that the captured IR is too short to fully recreate the experience (under 1 second), but I think extending the length of the impulse will take a bit more work.
My question is basically "what am I missing?". How can/should this process be altered/augmented to get closer to the free-field ambio effect?
I did forward the link to Ralph ( the founder of Ambiophonics Institute) and he told the approach was valid and similar to a project he is currently working with a University to make one similar headphones. You may want to write to him and ask for his feedback.
Meanwhile, if you are interested in loudspeaker binaural I can pre process the files for you for evaluation. The problem with other plugins is that you really need to understand the settings for perfection and there is no one setting for all. I use A modified RACE version and seemed to work fine across all recordings including QSound. Unfortunately, audiophiles are not helpful to provide feedback so the experiment is limited to my system at around 20 degrees. Ralph tried the recordings with his smaller system and it worked very well there. Just PM or email me. You can find my email address in my page.
Cheers!
I was looking at some of the free papers over at the AES, and it looks like Microsoft and Facebook have been pouring money into psychoacoustics:
AES E-Library
I didn't dig too far into the papers, but it looks like they're attempting to figure out things like "how can we make a sound appear to be far away when it's radiating from a point in space that's very close?"
There's a few dozen papers on the topic, for free, along with 100+ free papers on other topics. Kudos to Scott Hinson for publicizing the stash of free papers.
AES E-Library
I didn't dig too far into the papers, but it looks like they're attempting to figure out things like "how can we make a sound appear to be far away when it's radiating from a point in space that's very close?"
There's a few dozen papers on the topic, for free, along with 100+ free papers on other topics. Kudos to Scott Hinson for publicizing the stash of free papers.
If anyone is interested in trying ambiophonics and doesn't want to deal with using a personal computer or laptop to do it, I have one of the miniambio processors for sale on eBay right now. These were discontinued years ago and are very rare. I'm selling it for less than half of what I paid.
Another neat thing about the processor is that it can be used in a car. Heck you could even make it portable. (The processor runs off of USB power, so even a plain ol' battery pack for your cel phone will power it. Ask me why I know this lol
Another neat thing about the processor is that it can be used in a car. Heck you could even make it portable. (The processor runs off of USB power, so even a plain ol' battery pack for your cel phone will power it. Ask me why I know this lol
I have the parts to make two good pairs of speakers that I would like to use Ambiophonics, but the pair are different
I read somewhere that Ambiophonics can be implemented with -
Is there any combined Ambiophonics plus crossover DSP system (hardware & software/ plug-in) which as as well as creating the extra Ambiophonics channels - can also handle the speakers’ active crossovers - including adjusting for phase?
.. so phase is the same, or close to it - across all four speakers
I read somewhere that Ambiophonics can be implemented with -
- a laptop/ PC/ soundcard
- Four channels of amplification, and
- modest speakers - with consistent phase response - especially in the crossover regions
Is there any combined Ambiophonics plus crossover DSP system (hardware & software/ plug-in) which as as well as creating the extra Ambiophonics channels - can also handle the speakers’ active crossovers - including adjusting for phase?
.. so phase is the same, or close to it - across all four speakers
Why would you need four channels for Ambio?
I've done it various ways, and a plain ol' barrier works pretty darn well. I've done two different setups using a barrier, over the past year or so, and my current set up uses a barrier. I'm just using a sheet of plywood that's covered in Bubos Acoustic Panels.
About seven years ago I (accidentally) stumbled upon a way to do something that's similar to Ambio, but sounded superior IMHO. I can describe how I did it you'd like. I ended up trashing the project because it made everything sound TOO big. The soundstage was distractingly huge.
OK, so I came up with this accidentally:
A few years back, I was reading about a Bose system that was created for Cadillac. The Bose press release hinted that the system would have noise canceling.
In headphones, the way that noise canceling works is that it records the ambient noise and then simultaneously plays back an inverted version, canceling the noise.
I speculated that someone could achieve something similar without microphones by AUGMENTING what the speakers are playing.
In other words: here's the stereo triangle. What if we ADDED some speakers near the listener, to AUGMENT the sound that's radiated from the loudspeaker itself?
So I tried it, and it was basically mind blowing. The soundstage becomes about 180 degrees, it's like an arc. It sounds a million times better than any Dolby Atmos setup I've ever heard.
I also tried some variations on it; for instance, I did a version of it where the speakers were there for crosstalk cancellation.
The reason that it works so well, I think, is because doing crosstalk cancellation on the other side of the room is kinda stupid. The entire idea of crosstalk cancellation is that sound from the left speaker gets to your right ear, and vice versa. But if you do the cancellation on the other side of the room, the sound would basically have to travel like a laser beam in order for the cancellation to work properly. If you look at Ralph's ambio setup, he's using super-directional planar speakers, for the reason I describe.
But if you do the cancellation right where you're sitting, it works a million times better.
Another factor is that the cancellation speakers will reduce the bass. It's inevitable. So if you're going to use cancellation speakers, by putting the cancellation speakers much closer to the listener, it means you can run them at a lower power level, which means it will do a lot less to cancel the bass.
I tried this experiment using speakers that were IN PHASE and also speakers that were OUT OF PHASE. The difference between the two wasn't dramatic. Keep in mind that if the speakers are out of phase, you have to swap the channels. Basically there are two variations:
1) a crosstalk cancellation speaker. This would be placed to cancel out the sound from the opposite speaker. So you would place a speaker to the LEFT of you playing the RIGHT channel and out-of phase.
2) or an "augmentation" speaker. This would be placed to augment the sound from the same-side speaker. So you would place a speaker to the LEFT of you playing the LEFT channel in-phase.
For both options, delay is required to compensate for distance. Sound travels 34cm in one millisecond, so add one millisecond of delay to the speakers near you, for every 34cm of distance that the main speakers are AWAY from you.
Attachments
here's my ambio setup
The barrier separates a pair of JBL center channel speakers. The JBL speakers are particularly directional on the horizontal axis
I covered the barrier in Bubos sound absorbers (I still need to buy more.)
On my previous setup I had the barrier go nearly all the way down to my desk. You'd think it would be distracting, but surprisingly it's not, especially if you sit exactly equidistant.
It's interesting how the sound stage gets better and better as it gets closer to your face; a gap of even two or three inches is audibly worse than having it come all the way to my face.
Another option that might be worth a look is to apply a little bit of ambio processing (but not a lot.) I find that the electronic processing can sound kinda weird so use it sparingly.
I listen to a lot of podcasts, and for that the ambio setup simply sounds like mono.
The barrier separates a pair of JBL center channel speakers. The JBL speakers are particularly directional on the horizontal axis
I covered the barrier in Bubos sound absorbers (I still need to buy more.)
On my previous setup I had the barrier go nearly all the way down to my desk. You'd think it would be distracting, but surprisingly it's not, especially if you sit exactly equidistant.
It's interesting how the sound stage gets better and better as it gets closer to your face; a gap of even two or three inches is audibly worse than having it come all the way to my face.
Another option that might be worth a look is to apply a little bit of ambio processing (but not a lot.) I find that the electronic processing can sound kinda weird so use it sparingly.
I listen to a lot of podcasts, and for that the ambio setup simply sounds like mono.
