My system is under complete renewal, New speakers, new electronics.
The matrix stereo will remain !
I'm brand new to this. I tinkered with my friend's amazing 3 speaker system (Titan Reference Loud Speakers) and his Krell showcase preamp did a fine job of making 2ch music a 3ch affair. My own 2ch system blows me away with every excellent new track I feed it, but the sweet spot [quoting James Romeyn] is truelly nothing more than a sweet dot. If I want a little more headphone-like stereo seperation I'll lean forward 6 to 12 inches -- a lateral shift beyond a couple inches destroys the phantom center.
So I'm hearing that each track may have a different "ideal" center gain level. An entire album is usually recorded with the same microphone/s and studio, though I just played Music Nuda (2004) where every track sounded like they used an omni + figure 8 and her last track sounded like a coincident mic. Regardless, we don't listen to albums anymore, we listen to playlists.
So there is this parameter that exists at the track level. If each file had it's own "operation mode" coefficient embedded [as metadata tags] then our 2ch library would always playback in optimal 3ch conditions (Foobar let me add an arbitrary tag to a FLAC file). If not in the file, then in the CUE sheet. Or, add the Elias tag to LRC files which would leverage the current network of lyrics servers already in use. And if you wanted to get crazy, the LRC file could specify different coefficients at any point within the song. I swear I can hear when in a mid-side recorded song the distribution moves -- I presume engineers are doing this in the mix/eq process to accentuate a passage.
So I'm hearing that each track may have a different "ideal" center gain level. An entire album is usually recorded with the same microphone/s and studio, though I just played Music Nuda (2004) where every track sounded like they used an omni + figure 8 and her last track sounded like a coincident mic. Regardless, we don't listen to albums anymore, we listen to playlists.
So there is this parameter that exists at the track level. If each file had it's own "operation mode" coefficient embedded [as metadata tags] then our 2ch library would always playback in optimal 3ch conditions (Foobar let me add an arbitrary tag to a FLAC file). If not in the file, then in the CUE sheet. Or, add the Elias tag to LRC files which would leverage the current network of lyrics servers already in use. And if you wanted to get crazy, the LRC file could specify different coefficients at any point within the song. I swear I can hear when in a mid-side recorded song the distribution moves -- I presume engineers are doing this in the mix/eq process to accentuate a passage.
Good, then you may not yet be biased
I don't know about Krell, but I highly doubt it has a static linear matrix. What it most likely have is one of those 'steering' algorithms, like all the other devices on the market.
Certainly varying the level changes the sound. But what is 'correct' is much harder to determine. Usually one has to choose which parameters to optimise in a linear matrix. Those may include large sweet area, spaciousness, center focus, etc.
I see how Elias made a set of speaker wires that could drive 3 speakers from a 2 channel amp. This never occurred to me as an option (well outside my brain's solution domain). Would I be correct in saying it would never fly for an audiophile?
I had assumed that a signal level appliance (like the Trinaural processor) was the only real answer. But unless said processor can receive ques on how to handle each song differently, then a signal level appliance is also a half measure. When I say "que" I'm really thinking USB input that can instruct the processor which parameters to use. (I'm assuming this is not something that a processor can figure out on it's own using a microphone the way dynamic EQ processors do)
But it feels stupid to build a processor with both digital and analog inputs. May as well be all digital, and fork into 3 channels before converting to analog. So that leads me to where Markus is with his Reaper application. So if I'm using a desktop PC with a 24/192 capable RME sound card I will now have the option to exercise formulas for 3 or MORE speakers in the front.
If a hardware appliance is built for audiophiles, it will need to be a 3ch DAC that will hopefully receive it's track ques from the internet. Until then 3ch will be best realized by software geeks using programs like Reaper.
I had assumed that a signal level appliance (like the Trinaural processor) was the only real answer. But unless said processor can receive ques on how to handle each song differently, then a signal level appliance is also a half measure. When I say "que" I'm really thinking USB input that can instruct the processor which parameters to use. (I'm assuming this is not something that a processor can figure out on it's own using a microphone the way dynamic EQ processors do)
But it feels stupid to build a processor with both digital and analog inputs. May as well be all digital, and fork into 3 channels before converting to analog. So that leads me to where Markus is with his Reaper application. So if I'm using a desktop PC with a 24/192 capable RME sound card I will now have the option to exercise formulas for 3 or MORE speakers in the front.
If a hardware appliance is built for audiophiles, it will need to be a 3ch DAC that will hopefully receive it's track ques from the internet. Until then 3ch will be best realized by software geeks using programs like Reaper.
^
Please see http://www.diyaudio.com/forums/mult...s-3-speaker-stereo-matrix-15.html#post3405726
The software allows to do an ideal energy preserving, frequency-dependent matrix decoder.
Please see http://www.diyaudio.com/forums/mult...s-3-speaker-stereo-matrix-15.html#post3405726
The software allows to do an ideal energy preserving, frequency-dependent matrix decoder.
I'd like to think I'm like most consumer minded audiophiles in that I'm wired to crawl the forums for ages for a winning product and then use my modest mid-fi budget to procure a path to audio heaven. I totally agree with you Elias about over thinking, but I'm not actually a DIY'er so I can't do what you or Markus did without some serious hand holding. So far my armchair impression of exceptional 3ch is that the best experience is exclusive to the hobbyists using tools that allow them to experiment with freq-dependent algorithms, gains, etc that best suit the music of the hour. Armed with that impression I will cease to search for a line level appliance.
FYI Markus I installed the AcourateConvolver trial -- can you give me copies of your config and CPV files? Even if I don't figure out Convolver I will probably go for their windows based room EQ software. I know room EQ appliances are generally expensive and I don't think a black-box appliance would reveal the full detail of what it found wrong with my room. My Thiels are already coherent/time-aligned so it will be interesting to see where it finds problems. Both me and my speakers are over 3 feet off our respective back walls and I still hear some of the low-freq standing waves.
FYI Markus I installed the AcourateConvolver trial -- can you give me copies of your config and CPV files? Even if I don't figure out Convolver I will probably go for their windows based room EQ software. I know room EQ appliances are generally expensive and I don't think a black-box appliance would reveal the full detail of what it found wrong with my room. My Thiels are already coherent/time-aligned so it will be interesting to see where it finds problems. Both me and my speakers are over 3 feet off our respective back walls and I still hear some of the low-freq standing waves.
Hi,
Though there is another very similar topic here, I followed this one seeing as it's newer, old thread nevertheless.
I wanted to highlight something about this matrix below:
This page describes a product which does a similar thing, though the output is only stereo (from stereo input).
My guess would be it increases portion of lateral energy relative to the direct sound with transition at 700hz. If you add EQ the side, do same but inverse to the mono portion, the output should remain linear to a degree. I could be wrong. See attachment.
Slightly off topic but one of the best ways of adding de correlated (stereo) signal to a track is with reverb, especially at low freqs. The legendary Lexicon PCM 70 had a R-L,L-R preset... I assume it junks the recorded ambience and adds artificial ambience. Artificial ambience rules IMO, the following study concludes it is a preferable property which enhances listening. See this study: Artifical vs Recorded Ambience. I personally agree...
If intensity panned mono is preserved and spatial properties discarded (done by inversing the whole matrix back onto the input), the fullness of stereo can be added with a decent verb'. The first PCMs accepted mono input and generated stereo output.
I find alot of newer recordings lack the spatial element, which come through either disastrous or are non existent. If one can derive panned mono portion (if it's panned at all) and add reverb, decode to tri-field this should be better.
With the shelving EQs prior too but it largely depends on the recording. Perhaps some ambience detector voltage followers? I guess this is nearing the steered logic territory, though I disliked that approach when I heard it > pumping artifacts,
I don't know how else the Francenstein works though and/or whether the 'Spatial EQ' in attachment is related.
Any comments?
Though there is another very similar topic here, I followed this one seeing as it's newer, old thread nevertheless.
I wanted to highlight something about this matrix below:
This page describes a product which does a similar thing, though the output is only stereo (from stereo input).
My guess would be it increases portion of lateral energy relative to the direct sound with transition at 700hz. If you add EQ the side, do same but inverse to the mono portion, the output should remain linear to a degree. I could be wrong. See attachment.
Slightly off topic but one of the best ways of adding de correlated (stereo) signal to a track is with reverb, especially at low freqs. The legendary Lexicon PCM 70 had a R-L,L-R preset... I assume it junks the recorded ambience and adds artificial ambience. Artificial ambience rules IMO, the following study concludes it is a preferable property which enhances listening. See this study: Artifical vs Recorded Ambience. I personally agree...
If intensity panned mono is preserved and spatial properties discarded (done by inversing the whole matrix back onto the input), the fullness of stereo can be added with a decent verb'. The first PCMs accepted mono input and generated stereo output.
I find alot of newer recordings lack the spatial element, which come through either disastrous or are non existent. If one can derive panned mono portion (if it's panned at all) and add reverb, decode to tri-field this should be better.
With the shelving EQs prior too but it largely depends on the recording. Perhaps some ambience detector voltage followers? I guess this is nearing the steered logic territory, though I disliked that approach when I heard it > pumping artifacts,
I don't know how else the Francenstein works though and/or whether the 'Spatial EQ' in attachment is related.
Any comments?
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.