The Grateful Dead's Wall of Sound (2019)
audioacademy.in
audioacademy.in
The effect of that setup is that only the difference between the two microphones is amplified; common signal in both (i.e. the sound coming out of the speakers) is nulled out, but the difference signal (the voice) makes it through. It apparently wasn't quite perfect but was absolutely a lot better than wailing feedback.
The thing that made it sound so good was that any given speaker only reproduces a single source, but the article touches on that. The mic arrangement I described is simply what makes it possible.
While the WoS laid much of the groundwork for how modern PAs are designed and operated, it was more of a white elephant than anything, and many of it's actual ideas were discarded. It was totally impractical to tour with and they lost money doing so. The only real technical legacy it has is of using coherent phased line arrays.
Really it's whole reason for existence (getting a coherent, in phase, non-canceled signal at an extended distance from the stage) isn't even relevant, as these days secondary speaker arrays with delay lines (to sync them perfectly with the mains) is almost childs play. Literally plug and play. Modern PAs can self-tune the whole system just from playing a short burst of white noise through the system, and listening for the response.
A technology which was developed by/with the Grateful Dead, by the way.
From what I understand they essentially financed the modern PA industry by spending a ridiculous amount of money on sound equipment. People don't realize how much money they made - they were pretty much the top grossing tour band for about 15-20 years, playing about 90-100 shows a year. And they could (and did) use those shows to experiment with sound in a way that probably no other band has done since.
I haven't watched any D&C shows, but I expect their sound quality was just as good, if not better, than the Dead's.
"""What tools do modern sound engineers have at their disposal that you didn’t? Computers and all the things that became possible at the advent of computers. It’s removed the limitations to creativity. Nowadays, in terms of concert sound, you can not only correct the sound for any room, you can ongoingly correct it in real time as the room changes and as the temperature changes and as the humidity changes. We used to do that in the ’80s and ’90s – I had a complete weather station at my mix board and we tracked temperature, humidity, barometric pressure, because they change sound dispersion, sound quality. We corrected the systems accordingly. We did long, long studies and mapped it out, and we had curves so we could predict where the sound was going. We had to do that by hand. Nowadays, it happens all by itself."""
Much has changed from the days when we had to implement balanced binary trees of tapes (analog tape copies were lossy, so you wanted to minimize the total depth of copies).
(I took part in lots of tape trees back in the nineties -mostly Neil Young, with a good helping of Grateful Dead and occasionally some Phish thrown in for good measure. Then CD-R became a thing and we did lots of those for blanks+postage. Good times! (I even did DAT trees, as I had three (!) DA-P1s gifted from a local radio station going out of business)
I once got called down to the customs authority to explain what I was up to - they noted I got loads and loads and LOADS of seemingly innocent recording media in the mail, only to ship them out again at a later date.
Nothing showed up when they inspected the packages for drugs, so if I didn't mind - would I PLEASE explain what was going on?
The whole system travelled in two articulated lorries; but it took so long to set up that eventually they used four lorries, and two systems. One set would be for tonight's gig, the other would be on the road to tomorrow's gig, or starting to set up.
I think the mikes for the vocals were three mikes, not two (you can see triple mikes in The Grateful Dead Movie). I can't remember the rationale for using three mikes. The movie also shows the WoS being set up, under the direction of Ramrod, with Garcia gently urging him to get the speaker stacks hauled up higher, because that would be "really cool".
It's still very common to use largely-separate signal chains for the PA, and for recording (and/or broadcast). These days we often do it with analog or digital splits, but analog splits can [still] be problematic and digital splits didn't exist yet back then.
Analog splits might have been particularly problematic when the source is two microphones fighting eachother. And it's often [still] the case that a microphone that works well for PA is not a microphone that is ideal for non-PA use.
(Sure, we like to think of live recordings or broadcasts as being "being there," but that's almost never actually the case. When recording and/or broadcast is involved, those things are often handled as separately as is practical -- often with gear and engineers that are outside of the walls of the venue.
This allows the FOH folks to get it sounding good in the room, the recording folks to get it sounding good on tape/DAW, and the broadcast folks to get it sounding good on radio/TV. They're different goals, and they've each got different approaches.)
It was Donna.
It's like saying relay computers were dumb... Boolean logic was new and no one had ever attempted stuff like that before.
each speaker only carried a single instrument.
I also believe I heard some of the precussion mics were targetting only one or two speakers.
At least in the case of the speakers for Jerry, they had a a seperate McIntosh hi-fi amp for each speaker, being fed out of a Fender-derived preamp and a many-way splitter. Owsley basically bought the every one that model amp that was in stock at dealers on the west coast. Hundreds of thousands of dollars just on those amps - they were something like 2 or 3k a pop even then.
The only reason they were even able to afford in the first place was that Owsley (Yeah that Owsley, who was also their primary sound engineer) had so much illegal cash from a decade of making most of the LSD consumed in the United States. Band never even paid for most it. It was more this crazy idea Owsley had and mostly paid for that they kind of rolled with.
That sort of thing was more than a bit of a pattern in that camp, and was a large part of the band's downfall. It got to a point where it seemed like half of Marin county was on the payroll, and there was so much money going out that they had to tour constantly, wether they wanted to or not. The heavy touring clearly had a major toll on Jerry both physically and mentally. A two or three year hiatus around '91 or '92 would have done him (and probably some of the other guys) a world of good.
Being a non-profit facility, there are no fancy DSPs to notch out the hum or anything like that, so more creative solutions were investigated. It was determined through dumb experimentation that orienting an identical microphone 180 degrees to the one with the hum and setting the gain similarly would nearly eliminate the hum.
Eventually, the working theory became that a relatively new large pad transformer installed across the street was being picked up by the microphones. Orienting one microphone 180 degrees from the other caused the hum to be picked up out of phase from the main, and thus could be mixed in to cancel out the main mic hum.
Ultimately the real solution was simply buying better microphones, but there was a period of some months while a microphone sat off stage, pointed backwards.
I would have expected that this decades-old and well established component of power infrastructure would have been commoditized by now and integrated into any dedicated AV performance/production space such as an auditorium.
A simple notch filter won't fix 60 Hz hum. Or, at least, when I've tried to eliminate annoying 60 Hz hum in my own amateur recordings, it's never been very effective. The problem is that the 60 Hz hum isn't a sine wave. It's more like a square, so you've got a bunch of harmonics up the frequency spectrum to worry about too. You can try to also notch out 120, 180, 240, etc. but it starts to get weird sounding fast.
On the same pole facing down at the entire band, you have the omnidirectional mic trying to capture the whole thing as good as possible, suitable for live broadcast from this other channel alone. They didn't have stereo radio yet anyway.
Afterward back in the studio, starting with only the two channels on reel-to-reel tape, the "difference" channel can be phase-inverted to an auxiliary tape, then you have three channels suitable for mixing.
And with analog techniques like this they can be mixed "down" to 5.1 surround sound.
From a single mic stand and only two live signals.
But as for the acoustic implementation, even that has a long history. The Dead borrowed the idea from fighter pilot headsets, the only difference being that pilots were contending with a noisy cockpit rather than feedback. Same general idea that the unwanted sound hits both mics somewhat equally while the voice hits both mics somewhat unequally.
Therein lies the difference between Cat5 and Cat5e: Cat5 pairs were all twisted at the same frequency, while Cat5e pairs must not be twisted at the same frequency.
(This is also how we were able to transition so quickly and cheaply from Cat5 to Cat5e: All that was needed to produce Cat5e on a line that had been producing Cat5 was just a relative speed change on the machines that twisted up the individual pairs.)
What drives this? Singers and speakers are both localized sources, so I'd expect the mics to pick up similar phases for each.
I bet it's distance! Falloff depends on distance to source, so there should be a larger difference in volume for closer sources.
Edit: Apparently this is not the case!
https://www.soundonsound.com/techniques/line-arrays-explaine...
For the most part, each performer plugged into the "Wall of Sound" had their own vertical 1xN stack of drivers.
3 drivers in a vertical line away will have less distortion than 3 drivers in a horizontal array; the horizontal drivers will suffer from comb filtering for listeners who are not located dead center at the middle of the array. (This of course assumes your audience is dispersed horizontally as opposed to floating randomly in space)
Modern home loudspeakers hew to this philosophy as well to an extent. As opposed to big "monkey coffin" 70s speakers with a random array of drivers sprayed across the front of the speaker[1], modern tower speakers have a vertical array of 2 or more drivers whose centers are aligned in a vertical line[2].
____
[1] https://www.reddit.com/r/BudgetAudiophile/comments/yburht/at...
[2] https://www.audiosciencereview.com/forum/index.php?threads/r...
But you cannot say that on HN. HN is the place where people believe soundbars are as good as high-end audiophile speakers!
Why those bands, and not others? What do they have in common?
https://en.wikipedia.org/wiki/Les_Rallizes_D%C3%A9nud%C3%A9s
https://en.wikipedia.org/wiki/Taper_(concert)
(a disadvantage to too much ethology reading: I can't remember "Bill Graham Presents" without thinking of baboon behaviour)
Not only did GD (and particularly Jerry Garcia and John Perry Barlow) eschew the 'intellectual property' model of music, their thought-leadership has lived on to become much of what we today consider fundamental internet technology and methodology.
Early decentralized crypto-economics, peer-to-peer file sharing, and the founding of the Electronic Frontier Foundation were all developed by some combination of deadheads and musicians and tapers, particularly on an internet service called The WELL.
The Green Pill Podcast had an entire episode exploring the bluegrass roots of blockchain technology; I was humbled / psyched to be a guest and play several of my songs, as well as some traditionals that GD also played. It's here if you're interested: https://www.youtube.com/watch?v=Y3s9Fu4yu7o&t=2898s
Owsley (wall of sound engineer) was one of the original major LSD manufactures... and a dead show was always where you went to score ACID if you lived on the east coast. This remained true well into the 90's.
I'm going to guess that all of those early internet pioneers that you mentioned also have fond stories of LSD.
The Dead, Bill graham, hells angles, peoples temple, Patty Hearst.... There is a continuum of culture that spills out of San Francisco to this very day.
What the Dormouse Said by John Markoff is exactly that. A dive into how psychedelic counterculture made its mark on folks at Stanford, folks at XEROX PARC, Doug Engelbart, Jim Fadiman, Steve Jobs, etc. Not the most cohesive narrative but fascinating stories
https://www.amazon.com/What-Dormouse-Said-Counterculture-Per...
Who's got my miracle?
Lagniappe: (was this real blotter?) https://m.media-amazon.com/images/M/MV5BYTI5M2YyYTctNDBjZS00...
Barlow was also big on the WELL, an early influential time-sharing messaging system (https://en.wikipedia.org/wiki/The_WELL a Bay Area legends from a time when the Bay Area dominated in both technology and music). In many ways Hacker News is an intellectual inheritor of the WELL (along with Usenet and Slashdot).
The FLAC codec was developed by deadhead tapers.
The core of it is that speakers are bad at polyphony, so if you can avoid it, you can produce something that sounds more natural to human ears, and do so in a larger area. The way to avoid it is more speakers, more stacks/arrays/etc. You don't necessarily need an array per instrument, because modern loudspeaker arrays are indeed much better than they used to be, and modern loudspeaker processing fixes a multitude of problems.
This is also what Dave Rat is on about, in addition to IM and combing concerns, when explaining why a speaker per instrument sounds better than a mix: not just that it's a single point source, not just that it's doing one single job, but that all the sounds come from discrete points nowhere near each other. Just like an acoustic band using no reinforcement.
deal with the phasing over so many speakers?
https://www.soundonsound.com/techniques/line-arrays-explaine...
The bass guitar gets its own pile of speakers. The lead guitar gets its own pile of speakers. The vocals get their own pile of speakers. (And so on, and so forth, until everything on the stage was covered -- maybe with some mixing on a stack to fit something else in, but probably not for things like Jerry's guitar and Phil's bass).
So, yes: There were aspects of the Wall of Sound that absolutely behaved like modern line arrays do (even though "line arrays" as we know and use them today did not begin to gain wide popularity until somewhere around the middle of the 1990s).
There's no magic necessary to implement a line array: It can just be an array of speakers that are arranged in a line.
(Of course, it can also be much, much more complex than that -- but that complexity isn't an inherent part of what a "line array" be. It can involve things like phase-steering and per-element EQ and 3D predictions [and 3D measurements!], but it does not have to be that way in order to be a line array.)
The core of it is that speakers are bad at polyphony
https://www.erinsaudiocorner.com/loudspeakers/kef_r5_meta/
https://www.erinsaudiocorner.com/loudspeakers/sony_sscs3_tow...
more: https://www.erinsaudiocorner.com/loudspeakers/
The way to avoid it is more speakers, more stacks/arrays/etc.
I'm sure you know that, just clarifying for others.
If he has written about this elsewhere I'd love to read more!
Dave Rat has a good youtube channel, he also has some sort of insider subscription thing that I've never bothered with. I wouldn't say his ideas necessarily translate to every situation, but, he really presents these ideas as tools to use in a toolbox, not as gospel.
One interesting thing is that, we generally view comb filtering as universally 'bad', when, in reality, our ears do an excellent job of sorting out comb filtering when it comes to natural sounds. In fact, comb filtering is how we can locate a sound in 3d space with only two reference points (and, if you try, only one reference point moved around a little). That's remarkable, and points back to how speakers comb filter instead of mere comb filtering itself.
In practice, say you have a rock band, and your sound system has two arrays with subs spaced 40ft apart. Now, you're going to get a less than ideal pattern from that in the ranges where the bass guitar and kick drum live. How do you fix it? The answer is fairly simple, you simply run bass/kick in stereo, then, you delay the bass on one side by a little, kick on the opposite side just a little, then add some kind of EQ difference to the delayed side of each, then play with the delays until it sounds right.
Why does that work? Or does it really work? It's odd, because the math says "no no, it'll sound bad", the reality is, it can tighten up that comb to the point you don't really hear it and you can get good bass coverage out of a fairly poor system design.
So we do the obvious thing: We put speakers at each side of the stage, instead. This at least solves the obvious Newtonian physics issue.
But there's a problem with that. Now we have two sources of audio in one room, and the sounds that many in the audience hear are a sum these two sources, mixed together acoustically -- even though those two sources arrive at the listeners' ears with different amounts of delay (due to relative distance).
And when those two sources are playing exactly the same thing (an instrument or vocalist panned to the middle, say, as is quite common), that always results in predictable comb filtering for many people in the room. And it's the same comb filter across the whole spectrum, for all of the sounds (every instrument, every vocalist, every everything) that come from both sets speakers.
One approach that Dave uses is to try to make sure that each side of the PA is always playing something different from the other one. This might mean using two kick drum mics (placed differently), and two microphones in front of a guitar cabinet (also placed differently), and so on. One microphone is for the left speakers, and one microphone for the right speakers.
They still mix acoustically, but because they're different signals to begin with, they don't comb filter as efficiently as they would if they were identical signals.
And because each of these things are all mic'd differently (with one mic per speaker-stack), one instrument's inevitable comb filtering will be different from another instrument's inevitable comb filtering. This randomizes (ish -- it can be predicted and measured) the comb filtering that is involved in the sum of the entire PA, and that can make for a better listening experience for most people.
(And where two microphones can't be used, like for a vocalist, then the idea is to do something, anything different between the left and the right output channels for that singular source. It could be different EQ (what a wonderful phase-scrambler conventional EQ is), or delay, or different reverb settings -- but it should be different somehow coming from left and right speakers.)
The math works fine. If comb filtering is inevitable, and it is, then it is better to have a random-ish mix of comb filters of diminished depth and that each affect individual musical elements differently than to have one singular and very brazen comb filter that affects everything uniformly for any given listener.
The problems with Dave's ideas here are this: It doubles the channel count at the board (which can be worked around just by throwing more money at it), it more-or-less doubles the expenditure on microphones and snakes, and it more than doubles the complexity of getting an initial mix working well.
But most importantly: It defies the conventional workflow of trying to mix sounds on a stereo PA system in a huge room like it is the same as listening to a high-end stereo system at home and sitting in the sweet spot. Dave Rat's ideas generally don't/can't combine to make a central sweet spot, or to produce a perfect stereo image: There's a left mix and a right mix, and they're both very deliberately different mixes that are each made from sources that are split up to be as different as is reasonably possible.
So the biggest problem with Dave's ideas aren't that they don't make sense somehow (they do make sense), but that it goes against the conventional target.
(But the conventional workflow that aims for the idea of precise stereo image as a target is kind of inherently bullshit anyway, because only a small percentage of attendees can stand in that sweet spot to hear that perfect stereo image at one time, while the other 90+% of concert-goers simply can't physically be there (because Newton, again). Most listeners aren't anywhere near centered on the stereo PA, and will hear what psychoacoustically seems to come chiefly from either the right or the left stack (because Haas effect), and will have no ability to be in a place where a good stereo image is possible to begin with. The conventional target sacrifices the auditory experience of many for the benefit of few.)
Despite being an insane murderer, Phil Spector was a musical genius.
Why does every blog try to write a novel. Please just mainline me information.
But wait, then there wouldnt be room for 700 ads
If it doesn't make your pants flap in the breeze, turn it up!
I mean if you look at this effects it's not surprising he can get all those tones. He's got multiple signal paths that get split and recombined with various effects applied to one or the other (being played thru speakers that are miked), a Leslie (rotating speaker in a cabinet that adds some interesting warbles), a bunch of different gains with distortion, along with a bunch of "digitech" that does real-time complicated distortion. All of those run in chains so effects compound.
Some day I really want to have the time to build my own guitar similar to his as well as my own pedals. I've spent various bits of time in the background working on related things, but haven't allowed myself to really focus on it, because I tend to get a bit obsessive until I've fully explored my interests.
Here's a nice example of looping to make one instrument sound like many: https://www.youtube.com/watch?v=UsBINddmVcY
and here's a person who runs her harp through various effects pedals (she has a library of them): https://www.youtube.com/watch?v=C1Uv7JcnhhI
Emily Hopkins! I've followed her for a while. She just did a collab with Anthony Fantano and I think she blew up with it. Neat stuff. Makes me want a unique instrument to play. I saw Bela Fleck perform with a tribal African band and they had an African ukelele run through a wah pedal. Was super funky and not like anything I'd heard
I found this story [1] fascinating, and really enjoyed this interview [2] where he talks about working with Joe Walsh, a fellow amateur radio operator, and many others. He had a very midwestern modesty.
[0] https://news.ycombinator.com/item?id=39573756
[1] https://web.archive.org/web/20111005221916/http://www.perfor...
What is weird is that they don't talk about this. There is a graphic on products that can do this that show the speaker behind the mic. That's it, no text at all. I think the moral of this story is that marketing is weirder than engineering...
This image, addressing concert PA arrangements, has been floating around for decades, highly influenced by the GD setups.
One should not, of course, trust fifty-year-old memories, especially about something so subjective as sound quality. But, for what it’s worth, my recollection of the Hollywood Bowl in particular is very positive—clear, solid sound with distinct separation of the instruments, while not so loud that my ears rang afterwards (as they did after other rock concerts I attended in those days, until I started wearing earplugs to protect my hearing).
sound engineer that worked on coachella and they talk about the switch to vertical stacks of speakers
https://www.reddit.com/r/WeAreTheMusicMakers/comments/pjeq4/...
Google Maps indicates it moved from Santa Rosa to Rohnert Park.
Rolling Stone article from a decade ago: https://www.rollingstone.com/music/music-news/dead-wood-is-t...
Gosh now I’m imagining if people did this with books. At least chapter books are mostly the same.
That’s enough to cover a typical Playing>Uncle John’s>Drums>Space, and probably not get all the way through the reprises. You’ve easily got another two hours of jam.
"I went to a Grateful Dead Concert and they played for SEVEN hours. Great song."