Recently Synthodeon has been working on something called the '3i'. This is, among other things, a MIDI controller that will allow new types of playing that currently impossible due to the fundamental limitations of MIDI. While I cannot yet talk about exactly what the 3i is and does I can say a few things about what it needs to do MIDI-wise.
First of all MIDI has been tied to the notion of a MIDI cable and to some extent a "keyboard" - after all this is what it was originally designed for. There are sixteen channels on the cable and each channel supports things like pitch wheels, notes and all the standard MIDI stuff. Most of today's keyboard controllers support other things like CC messages for knobs and sliders.
But on the whole the "concept" is still very simple: Channels are tied to sound controllers (maybe with splits using extra MIDI channels to support two different sounds or to overlay sounds). Keys pressed operate within a given channel using Note On/Off messages, and so on.
So what don't I like about MIDI?
First off I don't like the basic model because its so very limited - particularly when you look at what can be done with things like Kontakt or Logic or any other MIDI-able DAW-type software. These apps have the ability to tie different samplers or functions to specific MIDI channels, i.e., I can have the same sampler on three consecutive MIDI channels. But, for example, most keyboards can only access a single sampler.
Secondly certain types of MIDI software in Kontakt, e.g., guitars in Orange Tree or Vir2, have funky multi-key controller options for their samples, e.g., you have keys to select a chord or note and separate "strum" or other note-sounding keys, reminiscent an AutoHarp. This is somewhat awkward and pushes the notion of MIDI away from keyboards - but there is no alternative place to go currently.
Third the MIDI bend (and in general the CC) model is very limited - I can only perform functions at the channel level. What happens if I want to bend a single note? Controllers I am aware of don't handle this at all.
Fourth, with the advent of network software (wireless, wired, USB, etc.) MIDI can run over a virtual channel from point A to point B. However there is no standard model for this. There is stuff like OSC and others but they are not standardized.
Fifth, things like iOS devices have limited asymmetric MIDI functionality. What I mean by this is that on an iPad, for example, you can plug in a MIDI device but only through a "Camera Connector" - and then your device must meet a variety of vague requirements (USB and power) in order to work.
Sixth MIDI is just not reliable enough for serious live music over WiFi regardless of what you do - even on local, private WiFi's with Mac's.
Seventh, I have a lot of devices - computers, iOS devices, keyboards, wind instruments. I want a MIDI model that represents each of these "places" on some sort of virtual "MIDI backbone" so that I can plug something into one device on the backbone and have that connection available at other points on the backbone.
Now I am sure there are a lot of arguments about why I am wrong, or why such-and-such a specific tool can overcome one of the limitations I listed above, but the bottom line is MIDI is twenty-some years old and really hasn't kept up with modern times. I want something that works in a modern way.
(Can you think of anything else you use that old on a daily computer or music basis save something like a physical guitar?)
I can't.
The oldest stuff I have that I use all the time are things like studio monitors and vintage synths - and most of that is not twenty years old.
So I've been looking around and found this site (OpenMuse) which talks somewhat about this in general. There are links to some standards efforts which mostly appear to no longer exist. There are also some patent links to US Patents claiming some form of jurisdiction over MIDI over networks.
Irrespective of any "standards" efforts there are a couple of patents to note. First there is U.S. Patent 5,983,280 "System Using Standard Ethernet Frame Format for Communicating MIDI Information Over an Ethernet Network" and U.S. Patent 6,353,169 "Universal Audio Communications and Control System and Method."
The first patent ('280) has a UK assignee. The patent discusses MIDI using Ethernet broadcast of various forms, multicast, and so on and it appears to be protecting some proprietary hardware and protocols. (See this for a reasonable discussion with which I would concur.)
The second patent, owned by Gibson, uses MIDI to control audio. Again I do not see this as relevant here.
So neither of these patents from my reading address what I will be describing here (though I am not an attorney and my opinions are worth exactly what you have paid to read article).
Are there more patents in this area? Its hard to say (and this is a reason the US Patent system is flawed - there is no way to discover what else is out there for a given topic).
The fact that there are numerous software apps (various OSC things, etc.) out there doing this sort of thing in one fashion or another leads me to believe that if there are they are probably not valid or everyone out there would be getting stomped on.
All that said what I want to see a MIDI software topology, at least at an abstract level, is something like this:
Here each "Node" (Node 1, Node 2, etc.) is a physical device (an iOS device, a computer, etc.). The MIDI ports listed on each device are virtual, i.e., not physical. If I need to link things to the outside world on a given device I can use some sort of MIDI port connector or bridge software to do that.
I want fast, clean delivery of MIDI messages - plain and simple. No other fancy nonsense to get in the way, no wizards, no autoconnect please press okay. (Who wants their shit to go dead during their epic solo? Worse, who want's to thrash through menus and crap trying to fix it afterword while the audience and your band mates stare at you with hate-filled grimaces.)
Basically I want to create MIDI endpoints between devices easily, reliably, deterministically and cleanly.
I want a physical wire (green arcs) between the Nodes (USB speed or better) because I want sub-millisecond response (WiFi is okay for testing and noodling around but not for live). To play live you need this along with the headroom it offers (for really fast pickers).
(I measured my response via MIDI on moderately fast piano piece - I can hear down to about 2.5ms - 3ms between chord strikes. I am not a pro and I cannot play really fast but my ears are pretty good. Sub-millisecond leaves a factor of about 2x which should be enough.)
I don't want automatic or automagic connectivity via something like Bonjour because for live gigs I don't want software "helping me" set things up. I don't really want to (accidentally) be on the venue WiFi (where some clown will start downloading porn videos to his iPhone just as I hit the first note in my solo).
Instead I want hard coded/wired links via fixed device-based TCP/IP settings over deterministic, private connections that stay the way I set them up and come back up spot-on. (Who hasn't had the circuit breaker or generator go out in the middle of a gig at a festival?)
I don't know if a local BlueTooth network is fast enough for this - especially outside of a "computer deveice", e.g., iOS. I have done a lot of testing and so far all I can say is "I am not sure." I am also concerned about BlueTooth interference (audience members with lots of active BlueTooth gear on).
From a device perspective Mac OS X works reliably on stage - at least in terms of no OS, audio or midi glitches and no battery fails - networking and iOS wise this all remains to be seen. I do not yet know about Android for this type of application though from what I have been reading its still immature in the developer area.
No comments:
Post a Comment