In Gerzon patent US5594800 the "energy preserving" requires parameter w to be 1. However with this value of w there is no solution of Gerzon coefficients to be found giving the same Trinaural coefficients.
Then we can say Trinaural is not "energy preserving" (in Gerzonian sense). And thus Trinaural cannot be a sub version of them.
I'm not sure do we need "energy preservation" ?
- Elias
Last edited:
Yes and we also need to make the decoder behave differently for high (>5kHz) and low frequencies (<5kHz). Explanation is given in here:
http://decoy.iki.fi/dsound/ambisoni...for Multispeaker Stereo (TRIFIELD)_Gerzon.pdf
You did try the freq dependent matrix ? Do you think it makes a perceptual difference worth pursuing compared to frequency independent decoder ?
Attachments
It is interesting to note that Gerzon energy preserving decoder for below 5kHz gives speaker signals for mono signal (L=1, R=1):
Ls = 0.577
Cs = 1.15
Rs = 0.577
so
Cs = 2*Ls = 2*Rs
which is the same ratio as Trinaural decoder for mono signal
Ls = 0.5
Cs = 1
Rs = 0.5
Cs = 2*Ls = 2*Rs
This means that for freqs below 5kHz they present the same phantom imaging for center panned sounds.
- Elias
Ls = 0.577
Cs = 1.15
Rs = 0.577
so
Cs = 2*Ls = 2*Rs
which is the same ratio as Trinaural decoder for mono signal
Ls = 0.5
Cs = 1
Rs = 0.5
Cs = 2*Ls = 2*Rs
This means that for freqs below 5kHz they present the same phantom imaging for center panned sounds.
- Elias
On the other hand 5kHz is quite high freq. The difference of matrixes may depend on music type and freq content therein.
Also, if the tweeter is non-conventional e.g. ceiling firing configuration where high freqs are presented in more diffusive nature to the listener, the difference between matrixes may be small above 5kHz.
- Elias
Elias,
I have been reading your article on your website, and some of your posts here on DIYaudio for the past couple of days. The trinaural approach now pikes my interest, especially that I have already experimented with ambiophonics (binaural) and found the results interesting.
I currently have enough speakers and amps to give trinaural a try, but I am yet to find a simple (AKA cheap) line-level analog circuit schematic to mimic the setup you describe in your articles (L-0.5R, 0.5R+0.5L, R-0.5L).
Could a simple resistor divider network be enough? I know it won't be frequency dependent, but could be enough for an early experiment.
Sorry if i'm posting in the wrong thread, or if the question has already been answered.
Regards
Nick
I have been reading your article on your website, and some of your posts here on DIYaudio for the past couple of days. The trinaural approach now pikes my interest, especially that I have already experimented with ambiophonics (binaural) and found the results interesting.
I currently have enough speakers and amps to give trinaural a try, but I am yet to find a simple (AKA cheap) line-level analog circuit schematic to mimic the setup you describe in your articles (L-0.5R, 0.5R+0.5L, R-0.5L).
Could a simple resistor divider network be enough? I know it won't be frequency dependent, but could be enough for an early experiment.
Sorry if i'm posting in the wrong thread, or if the question has already been answered.
Regards
Nick
Would a speaker level circuit suit you ? You can manage with only 2 channel amplifier, if their outputs have a common ground.
Also the passive RC filter can be used if needed.
An externally hosted image should be here but it was not working when we last tested it.
- Elias
Thanks for the quick reply Elias.
I now have more questions, what would the values of Rps & Cps have to be for 8ohm drivers?
Or to be more general, what's the equation that dictate Rps & Cps?
Furthermore, only one of my amps in my stable has a common ground at the speaker output, and it's the amp that sounds the least good.
So, although I know that for most people, being able to derive trinaural audio for a single two channel amp is the better/easier/cheaper solution, in my specific case, a line-level circuit would be better suited.
Just wondering if there is such a passive circuit (or an easy active one) available at the moment.
I now have more questions, what would the values of Rps & Cps have to be for 8ohm drivers?
Or to be more general, what's the equation that dictate Rps & Cps?
Furthermore, only one of my amps in my stable has a common ground at the speaker output, and it's the amp that sounds the least good.
So, although I know that for most people, being able to derive trinaural audio for a single two channel amp is the better/easier/cheaper solution, in my specific case, a line-level circuit would be better suited.
Just wondering if there is such a passive circuit (or an easy active one) available at the moment.
For 8 ohm drivers you may start experimenting with
Rps = 4.7 ohm
Cps = 4.7 uF
Roughly the function can be describes as:
With smaller values of resistor the center signal is attenuated more and side signals are boosted more.
With smaller values of capacitor the filtering action starts higher in frequency.
There is plenty of room for experimentations I believe, since every listening room is different including wall absorption, lateral dimensions and the listening distance.
Happy experimenting !
- Elias
Lynn Olson describes how proper 3-ch more accurately reproduces tonal qualities in the C Ch vs. stereo's phantom C. I absolutely agree. I am Trinaural user since 2007 and never prefer 2ch.
I often switch back to 2-ch and quickly adapt, thinking Trinaural is unnecessary. Immediately upon switching back to Trinaural I notice how inferior and how mucked up was the tonal color of stereo's phantom C. You have to really know the true currency to recognize the counterfeit.
For Trinaural I increased L/R spacing 62% (Golden Ratio) beyond Cardas' 4/5th specification. Trinaural has wider stage and images extend farther vs. stereo. Audience members appear 75' beyond side and front walls.
With all due respect, James never disclosed Trinaural "algebraic processing" so there's nothing to "gather".
I know only one living person who likely has access to the formula and he has every reason James did to keep it secret.
Trinaural released early 2000s for $1500 and is now $2500. With average discount I consider it decent value or better.
Your C image may indeed be perfect, no sarcasm. My best references are Crump's last CES (two large Soundlab stats, Blowtorch, hot as a mutha JC1s) and Kimber's best ever IsoMic never again to be displayed, eight floor to ceiling stats (2 per corner), proprietary IsoMic 4.0 DSD, oodles of Pass class A amps.
My center image in stereo equals about the best I've heard. After switching to Trinaural it is immediately apparent that the stereo image that seemed "perfect" a minute prior, has extreme tonal aberrations. In stereo the saxophone sounded more like a plastic toy well played. In Trinaural the glory of the instrument is crystal clear and obvious.
It's as subtle as a 2x4 to the thigh.
Last sentence: Bingo!
Lynn, are you ever wrong?
I'm 75 minute drive NE of Salt Lake City. Anyone who wants their "sacral" 2-ch paradigm destroyed is welcome!
Last edited:
I am also giving respect, but I cannot help noticing in Austria they are building DIY trinaural decoder
DIY Trinaural Decoder - HiFi Forum
I don't know how they are able to do it without knowing the matrix coefficients ?
However, it is very easy to measure the coefficients by test signal if one has access to the original Trinaural equipment. Afterall, it is said to be linear static matrix, right.
Last edited:
Excellent point.
The audible spectrum is 20-20kHz, while "middle" C is a mere 262 Hz! (1st fret of the 2nd highest string, 2nd string from the bottom, standard guitar tuning...listen to it next time you pass a guitar)
Notice how "piercing" high is that 1kHz test tone.
Excellent! Thanks for posting that! Now that I read this, I remember it indeed crossed my mind years ago that all one must do is input signals and check what happens at the output. Moderator please delete my stupid post above!
If the math is so basic and easy, why do Japanese mass market receivers not employ it? All their multi channel synthetics sound like crap. I suppose it just reflects how dumb they are.
What is estimated parts cost to make DIY at least equal to the original in pure sound quality? Also, is it easy to adapt the math to the digital domain? Is it possible that Dolby Pro Logic II properly implemented with state of the art digital technology can create a new digital benchmark? If so, strange indeed! DAC makers take note! Time to add value!
Jim Smith said he thought Trinaural C Ch quality was inferior to the L/R. I had to admit, after he said this, I though I noticed a disparity in quality. All three of my speakers are identical highly refined 2-way stand mounts crossed in Trinaural @ 80 Hz. Elevating the C Ch speaker 1.75" (no other change) seemed to completely cure the disparity.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.