I decided to video and upload a short tour of my new workspace for music and electronics and general maker stuff.
Here’s the latest picture of my basement studio space/lab:
You can see some of my new purchases, too, like the Akai Miniak and the TC Electronic Impact Twin interface.
I’m broke, but I’m happy.
Well, I’m not sure I’d really refer to limitations caused by my lack of funds as “principles.”
As you might recall, I mentioned some time ago (it was long enough ago that I don’t feel like going back and finding the post) that I have decided to eschew software synthesizers, largely because the money it would cost to buy all the good ones would be better spent elsewhere in my studio. Combine that with my love of classic musique concrète and ideas of found sound, and I became convinced that I could make music with a signature more uniquely mine without any software-based synthesis.
Now, I’ve known about Csound for a long time. Even with my background in science and coding, though, it still intimidated me. The problem is, I continue to see people do really cool things with it, like Dave Seidel’s Sruti Drone Synth. Add to that the latest version of QuteCsound and I can’t help but thinking that maybe now would be a good time to re-enter the world of computer music.
One of the stumbling blocks, for me anyway, has always been the lack of way to integrate things like Csound and Puredata into a regular musical workflow. There is a VST version of Csound, but I could never get it to work with Reaper. QuteCsound has simplified the world of real-time Csound a lot, but internally routing the audio within the computer is a bit of a pain, and I find QuteCsound to be unstable on Windows anyway.
So now I find myself seriously considering buying a cheap PC barebones kit from some place like TigerDirect and building a dedicated synthesis box with Linux. Since I want to get an Impact Twin for my main audio interface, that’ll leave me with this Echo MiaMIDI as a spare, which would work perfectly for a balanced out for said synthesis machine. Thus I would be going back on my own principles, which I never set out in any sense of righteousness in the first place.
In the end, though, Csound is a form of computer synthesis that doesn’t have to sound anything like NI or any of the other popular software synthesis systems. So maybe it’s time to dust off the old Csound Book I bought some time ago and try out a few things.
Well, I went and done did it!
You can now reach this blog via my new domain name, eternalprototyper.com. This is long overdue, especially since I was able to get such a cool name for the site.
That’s about it for now. I have some plans for the future, including a complete overhaul of the look of the site. For now, though, I’m going to order off for a bunch of parts because there’s making that needs done. Bye for now!
After complaining about the effin’ state of science on television, I thought it might be nice to share my thoughts on a show that I actually like.
Food Jammers is a Canadian show about food, oddly enough. It’s also about three guys and what they do with food, and that’s where things really get interesting. Like when they made their own coffee roaster with a couple of metal colanders, some threaded rod, and copper pipe. Or when they made their own sodas by hacking their fridge (literally “hacking”: they drilled holes in the poor thing!) so the soda liquid, now cold, could flow out and interface with a CO2 system they bought.
In short, they’re Makers. They’re always on the lookout for materials that will work in their projects, they keep their design fluid enough so they can use what’s available to them, and they’re not above a bit of dumpster-diving to get a usable bit.
The show does have one interesting quirk, though, and I’m not just talking about the presenters. You see, there are a lot of how-to shows on television that deal with more complicated subjects. Norm Abram’s been building high-end furniture for years on the New Yankee Workshop. Alton Brown’s had more than a few construction projects on Good Eats. The Food Jammers, though, do one thing none of these other shows do.
In every episode, at least one of their projects will meet with a disaster in mid-episode. Now, I can’t say for certain that they plan this, or whether the projects are complicated enough that such failures just happen anyway, but the one thing they never do is give up and go back to noodling on the synth. [EDIT: I think I can say for certain now that it's all staged, but the point about not giving up remains.] Always, they re-evaluate the design, come up with a solution, and go back and try it until they get what they want, whether that’s the perfect coffee roast or a dehydrated turkey to take camping.
There’s a lesson in this for all of us. You have to be ready for failure, and be willing to reconsider your ideas and make decisions on what is and is not working.
My biggest complaint is not about the show, but about the channel that carries it here in the States. There are 39 episodes of Food Jammers made, split up into three seasons of 13 episodes each. The Cooking Channel, however, insists on playing the same six episodes over and over again, and I fear people will get burned out and the show will get pulled from rotation before we get to see all that these three fun-loving hipsters have to offer.
Oh, and by the way, their light bulb-based food dehydrator is actually a pretty good design.
There’s a new episode of Catching the Sound up. The third one, actually. But that’s not what this post is about.
The first two episodes of our erstwhile show on sound design were produced by recording the audio directly from our Skype conversation, and then laying in the intro and outro, the bumpers, the example sounds, and so on in editing. This produced a final result that was listenable, but not pristine in any sense.
Now, we had thought to set up our sessions to record each of our mics locally, then all the tracks could be sent to me for editing. That way, the audio would sound good, at the expense of the complexity of syncing three (or more, possibly) tracks during editing. We felt this would be too complex, and lead to more points of failure, as one recording not working for some reason would scuttle an entire episode.
However, the noise levels bothered me, and the quality of Skype, although great when compared to the telephone, is not so good when compared to, say, FM radio. So we finally decided to try recording our three mics separately, using a countdown and a handclap for sync. It turned out to be a lot easier to do than I thought it would be, and the latest episode is now sounding a whole lot better than the previous two.
As for the points-of-failure issue, I still continue to record the Skype conversation directly as a back-up. That way, if a technical glitch somewhere messes up the regular recording, we can still release a lower-quality-though-still-listenable version that week. Then I can just record a slug at the beginning of that episode explaining what happened.
So there you go, my “in a nutshell” guide to podcasting with people scattered throughout the land. Remember to have fun, and if you become a famous podcaster known all over the world, try to remember us little people.
I didn’t think it was possible to make a science-oriented show more sophomoric than Junkyard Wars.
I finally decided it was time to effin’ watch a few effin’ episodes of G4′s attempt at an effin’ science show. After all, I can’t really be proper effin’ critical of a show that I haven’t effin’ seen. At the start of the effin’ show, we get an effin’ slate giving the usual effin’ disclaimer: These experiments and stunts are effin’ dangerous, and you shouldn’t effin’ do them yourself, and so on. The rest of the effin’ show is then all about three effin’ people behaving in exactly the same way they effin’ should NOT be while doing dangerous effin’ experiments.
And if you thought I used too many effin’s in that last paragraph, try watching an episode of the show sometime.
So what about the science? Well, it’s in there, sorta. You see, they bill this as being a show aimed squarely at adults. Apparently, that means puerile humor, occasional fraternity house hijinks, and making jokes about George Michael in a port-a-potty. And after all that, the science is at a level no higher than your typical first-year high school class.
I’m not going to be totally negative, though. Each episode usually features a segment where they show how to make interesting or useful gadgets, like a battery-powered mini-bicycle or a sun-powered cell phone charger made from solar yard lights. Both of those were from the same episode, I might add. But did the show itself actually give us all of the information needed to do these things? Not really, but it wouldn’t be hard to find. And of course, don’t forget that we aren’t supposed to try any of this at home.
I’ve talked a lot about wanting to do my own science show as a self-produced podcast, also aimed squarely at adults. So how would I go about it?
To be honest, my first inclination was also toward adult-oriented humor, but I quickly grew out of that idea. The fact is, the people who want to watch a show like this want to be informed, and the entertaining part should be there, but secondary. Most of all, though, I don’t want to simplify any explanations. People who watch science shows really need to be shown the real work required to find a solution to a technical problem. In the real world, we need algebra and calculus and statistics and laws of physics and chemistry to navigate us from not knowing what to do to something that works, whatever it may be. And sure, you can prototype a device 150 times to finally arrive at a solution, but if you really and honestly understand the maths and the science behind the principles you’re working with, that shouldn’t be necessary. See, math saves you money.
My idea is to present a problem to be solved, and then solve it. First, by carefully defining the problem. Then considering potential solutions. Then, after picking a solution, go through all the steps needed to implement it. No simplification. No holding anything back. Now THAT would be effin’ cool.
I’d still need to tell people not to try it at home, though.
As if the basement studio space and the modular synth were not enough, I have two other projects I would really like to pursue. Well, three if you count the PICSynth, but I don’t have to design that one.
Of course, part of my wanting to pursue these projects has to do with wanting to play with PIC microcontrollers. Why PIC and not Arduino/AVR? Well, the PIC line also includes a DSP model, called the dsPIC. That’s actually a pretty big deal, especially for one of my synth ideas. But I’ll start with the other one first.
I’ve always wanted a string synthesizer. Your basic seventies string synth consisted of 12 oscillators for the highest octave of the keyboard, and then divide-down circuitry would be used to make the remaining lower octaves. This signal would then be fed through waveshapers to get a sawtooth wave from the square waves made by the division circuits, before going through a chorus and onto an output amp controlled by the keyboard. This is just one way to do it, really, and in fact, the Paia Stringz’N’Thingz string synth output pulse waves instead, at an 80% or so duty cycle before going to the chorus circuit. Pulse waves are the path I’ll be taking, largely because they’re dead easy to make with even the cheapest PICs.
The PIC18Fs come with eight 8-bit input and output channels, which are switchable between the two directions, if I’m reading the data sheet right. (In fact, if someone can chime in about this, I would greatly appreciate it.) So, the idea is to use one PIC chip per note in the octave, for a total of 12 PICs. Each PIC would then be responsible for outputting all of the octaves of it’s assigned note.
The key here is to use the digital outputs to create the analog waveform. It sounds weird to not use the DACs, but there’s beauty in simplicity here. All we have to do is set the output pin high and then low rapidly, at the right frequency for the tone needed. Bang! Instant square wave. And if I’m correct there won’t be any aliasing, either, as this is one of those rare instances where a particular digital signal can also be an analog signal. The duty cycle is then just a matter of lengthening the time the pin sits low versus high during each cycle. This does mean 37 separate signals will then need to be summed in the analog domain, which will mean a lot of soldering, but it can be done.
Let’s talk about taking input from the keyboard. The first thing I thought of is to hook up all of the octaves of a note from the keyboard to an input pin for that note’s PIC. Each octave of the note would go through a unique resistor divider, so that each octave would have its own DC voltage level. The trick here is that every possible summation of these voltages would also need to turn out to be a unique level, so that any combination of keys could be discerned by the firmware on the chip. I don’t know if this would be possible for four voltages. The PIC18F series has an A-bus (again, from what I can find in the data sheets and online) which has four 5-bit I/O pins. I could use one per octave, seeing as then I’d only need to check for on and off and the resolution of the inputs won’t really matter.
The hard part will then be applying an ASR envelope to the output amplitude of the signal, before we get to any other analog components. One ASR and amp at the output means that the amplitude envelope will reset only after all keys are up. This can create artifacts if you have a sound with a slow decay, and then hit a new note during that decay (the envelope resets itself and kills the decay instantly, which is really noticeable if you also set a slow attack). Also, if you are holding some notes, and then add new notes in, they will come in instantly at full volume. Pretty much all string machines of the seventies had these same issues, though, and if they lived with them then, I can live with it now. Who knows, I might even find a solution.
On the analog side, I’ll of course want a chorus, and maybe a phaser as well. Maybe even also a filter. If I do put in a filter, it’ll be the first in the chain (filter -> chorus -> phaser) and will probably be a low-pass resonant filter with an LFO. I don’t want to mess with envelopes at that point. The chorus and phaser circuits I can probably just gut from guitar pedals found on Craigslist.
And if that wasn’t enough, here’s project number two. It’s an additive synthesizer.
This is where the added DSP power of the dsPIC would come in handy: generating a metric buttload of sine waves, all at different frequencies and amplitudes, and adding them together. In a high-end additive synth, each sine wave (there might be as many as 128 of them) would have its own frequency envelope and amplitude envelope. That’s a terribly large amount of calculating to be doing 44,100 times a second.
My idea to streamline this is to give each collection of sine waves (hereby known as a voice, as in “voices of polyphony”) only one each amplitude and frequency envelope. The envelopes would then be applied to all sine waves, but not necessarily equally. Each sine wave would have a coefficient that would be multiplied by the incoming envelope signal before being applied to the sine wave itself. This is still a lot of calculation, but it’s a lot less than custom envelopes for each partial (that’s what we computer music people call a sine wave in an additive synth). This idea would also make programming easier. Can you imagine editing pitch and amplitude envelopes for each and every partial 64 or 128 times? I don’t want to think about it either. One of the other cool things is that the coefficient for any particular partial can be negative, so it’s amplitude or frequency would go up when the others were going down and vice-versa.
From there, I would put in one dsPIC per voice, probably up to eight of them. A ninth processor, which could be a regular PIC, would then process the keyboard input and manage polyphony.
Again, I don’t know how much power I can squeeze out of a dsPIC. I did, though, find this cool video of a monophonic virtual analog synth made with one.
I think that’s a good thing to end on, actually. Have a good one!