In praise of MIDI, tech's hidden gift to humanity If you're not a musician, you may never think of MIDI, the Musical Instrument Digital Interface standard that links up keyboards and other electronic noise boxes. Firefox has, adding the super-niche Web MIDI API in its latest version. That's one of those "uh, OK" decisions which gets weirder the longer you look at it – but then …
There were other things that weren't broken in history, but they were pooh-poohed by the younger generations who were convinced they could "do better", changed the labelling, changed the names, changed the implementation language, and delivered pretty much the same thing that was done 10-20 years earlier.
Anyone who has been in the bowels of IT for 30+ years as I have has seen that happen time and time again...
Re "That's not the worst thing, as if they made it again, the whole thing would probably be javascript and HTML under the hood, which would improve nothing."
It would probably require Javascript and HTML, plus a recent Chromium based browser, and upload everything you scan to some cloud service for *ahem* processing.
The scary thing is, a couple of years back, I had to get a scanner working for a user. This was some God forsaken Lexmark thing that required the users to access it via a web interface. For one thing, the interface was pointless. The user had to be near the scanner anyway to load it with things to scan.
The worst part was the web interface. Not only did it use Javascript and HTML, but it relied an a Java applet to do the actual scanning. This meant we had to set up a special machine off the network to access this scanner, as our systems team banned Java thanks to Oracle's licencing requirements.
IIRC, it did offer a USB port, but the driver was not compatible with Windows 10, Lexmark had not updated it, and the department concerned did not have the budget required to replace what was quite an expensive scanner that, apart from the out of date driver, still worked.
I ended up setting up a private network in their office, just connecting that printer and a PC, with no outside network access.
Most notably there's OSC (open sound control) and while it has never caught on in a mass market way for musical instruments, it does have a lot of uses. it's more flexible, is easily human readable, works over networks by default and as such is more likely to be used for the control of things like synced visuals etc. Tends to get used in media servers etc, the kind that run the big screens you see at concerts - but it is fully specced to control music as well. It's supported in a lot of DAWs these days.
It's a better choice than midi for the modern non musical applications that this article mentions
I used MIDI for my GCSE too, except I used Quartet on an Atari ST. The music teacher loved the idea and was keen to share it around the class especially once she realised the school already had the necessary equipment to do it with.
I still play with it a little, and highly recommend the Roland A49 MIDI keyboard if you prefer key/action quality over features you might not need like tiny drum pads, multiple rows of of sliders and knobs etc.
RAM Music Machine review and demo tune with cowbells. Possibly even more fun than a Currah Microspeech.
Saving to, and loading from, the integrated tape deck, presumably?!
(For the uninitiated, the +2 was essentially a Spectrum 128 with a built-in cassette recorder (and the first new Spectrum produced after Amstrad bought Sinclair's computer line). The +3 came out the following year and had a disk drive, but with the same Z80 CPU and 32-column text/graphics as the original Speccy, I doubt anyone who was serious about MIDI would bother paying £199 for it(!) when they could have the 520STFM which had just fallen to £299 around the same time).
Possibly; though the Spectrum 128, +2 and +3 apparently had MIDI built in (though it required an adaptor cable, which I assume would have been cheaper than £50). And if you had a 128 or +2 anyway, there was no harm in trying that.
I just couldn't imagine anyone using one too seriously for that purpose without a disc drive... but then, I couldn't imagine it being worth buying the (drive-equipped) Spectrum +3 for that either when you could get an ST for £100 more.
(I found out recently that the +3 was meant to be £249(!) when it launched. I can only assume they dropped it to £199- which I remember it being- almost straight away after they got caught by the ST falling to £299 which would have made it look even more ludicrous).
One of the features of the Atari ST was the inclusion of MIDI ports.
I have no musical ability at all so I never hooked mine up to any instrument but from what I heard a lot of musicians did and for a while the ST became the computer of choice for folks like that.
One up for the ST, I reckon.
Since this appears to be developing into one, it might be of interest for those who missed it to note we already had an extensive discussion on the Atari ST's importance in the MIDI field just a couple of weeks back. (This was part of a response to this article on its less well-known successor, the Atari Falcon).
And the absolute quality and timing of the ST's MIDI is still revered to this day. Take that, Amiga fanboys!*
*This whole ST/Amiga thing is so amusing, its still like a kind of cultural identity badge. Although, after all these years I'm even prepared to concede that the Amiga was probably the better machine, especially for games. However, I'll still maintain that the ST was significantly more likely to be used as a multi-purpose home computer for all kinds of fantastic hobbies, and as such it has much more in common with PC's nowadays. I'm not saying YOU (whoever you are) didn't use YOUR Amiga to do interesting things, but I think fewer Amiga owners generally did. The End.
I'm not saying YOU (whoever you are) didn't use YOUR Amiga to do interesting things, but I think fewer Amiga owners generally did.
I can't see how you could describe Cannon Fodder as not interesting!
The ST was used in many recording studios because of it's clean MIDI interface, I saw a picture of Norman Cook's studio recently and he still has on.
The Amiga on the other hand had great graphics, so was used for video work. I seem to remeber the Chart Show on ITV used an Amiga 3000 for the graphics
It should be noted that most uses like that still used add-on cards to provide true 24-bit colour. As stunning as the Amiga's regular 4096 colour HAM mode was by mid-80s standards when it first appeared, it was still only 12-bit (i.e. 3 x 4-bit) RGB which wasn't enough for smooth gradients without posterisation or dithering (alongside some minor colour fringing due to how HAM worked).
That said, the Amiga was still- as far as I'm aware- more suited to general video use, due to its design, than the ST.
And then again, the original ST, for all that it was- for legitimate reasons- considered a good choice for controlling external MIDI keyboards, ironically had incredibly weak built-in sound (which we already went over at a recent discussion, so apologies for anyone getting déjà-vu here!)
HAM is actually a clever bit of salvage on the Amiga's video circuitry. Amiga started life as a chipset for a new Atari home games console (Atari commissioned the design of the chips). One neat trick the designers came up with was to take advantage of how the composite video signalling used by TVs meant that the colour information had half the horizontal resolution of the brightness signal. (The early "4:2:2" digital video coding standard is so-named for this property of analogue video: the luminance, Y, signal is twice the data-rate of the two colour signals, so each byte would have 4 bits Y, then two bits each for the U/Pb and V/Pr components)
So, to get more detail without wasting RAM, the Amiga was to have had a graphics mode where, for each two-pixel block (i.e., the length of a colour clock time) you would set the colour once, but still be able to adjust the luminance in the middle. The video circuit would Hold the chroma value constant from the first pixel while it Modified the luma value (hence Hold And Modify).
For various well-documented reasons, Atari never got to use the Amiga chipset for a new console. Because the Amiga staff had been getting bad vibes from Atari about this deal even before Atari imploded in 1984, they repurpose their designs for workstations and looked for an alternative buyer (which they found, of course). One big change that this needed was to move from the composite YUV colour model used for TVs to the RGB as used by "real" computers. But, unlike composite video, in RGB every channel has the same bandwidth, so the neat Hold/Modify trick wouldn't really work as well. But as most of the silicon was there and debugged, it was left in place and quickly adapted to work with RGB instead. The "fringing" artifacts are because you cannot produce 12x2 bits worth of RGB with just 12 bits of pixel data (HAM mode is 6 bits per pixel), no matter how hard you try..
Proud A2000 owner. Twice over in fact. MED, Deluxe Paint, Imagine and AMOS all did pretty interesting stuff!! Even if I was crap at all but the latter.
No question the ST with Steinberg annihilated the miggy for pro music of course. I would love a Falcon, but just not sensible to get one.
Games wise the ST had a slighty quicker CPU so marginally advantageous on some tasks (Elite 2) in the out of box config.
Never understood the obsession with cannon fodder. Dreadful game. Plenty of better stuff on both systems.
(Warning; boring ST vs Amiga distraction follows!)
> I'll still maintain that the ST was significantly more likely to be used as a multi-purpose home computer for all kinds of fantastic hobbies, and as such it has much more in common with PC's nowadays.
With respect, is there any evidence that this was actually the case, or is this just wishful thinking trying to rationalise its limitations into (somehow) being a virtue? (i.e. its games were less good so people *must* have been forced to use them for other things?)
> I'm not saying YOU (whoever you are) didn't use YOUR Amiga to do interesting things, but I think fewer Amiga owners generally did.
It's undeniably obvious that a high percentage of the Amiga user base bought and used them almost exclusively for games, but I'm pretty sure many people in the early days also did so with the ST (before the Amiga became more affordable).
The ST *was* definitely more suited to MIDI use (the Amiga less so for technical reasons I linked to here). But then, the Amiga was better suited to (e.g.) hobbyist graphics, and also for non-external sound (e.g. soundtracker/OctaMED music), both of which I was into.
And yes, I'm aware that might be the type of example you had in mind above and- you'd argue- not necessarily representative of the overall user base from back in the day. Possibly- but then again, simply asserting what "[you] think" was the case isn't any more valid either unless *that's* also backed up with evidence.
If the ST user base was latterly more serious, it was more likely because those who *had* used it for games moved on to the Amiga. Just as the games players moved on from the Amiga to the PC and Mega Drive and SNES, leaving being the hobbyists and more hardcore fans and those who still used it for the remaining niches (video and graphics in the Amiga's case, MIDI in the ST's).
There was an ST game ... "Midi Maze" that used daisy chained STs to create a 1st person multiplayer shooter (your avatar was just a big smiley face) in a very simple maze.
You could even have one machine with overview mode that showed where everyone was lurking. My friends an I had great fun with it in the pre doom days.
Back then ('87-ish), I figured out a way of increasing the max players in Midi Maze from 16 to, theoretically, "unlimited". The idea was to bridge MIDI rings together over (IIRC) the modem port. I never did it (there weren't that many ST owners in the local Atari club!) but I was certainly confident it was feasible, as I think there was something to do with the game itself not actually being limited in the number of players, just whether a machine was linked in to the master.
(I did a similar thing with Morrowind to make that 2 player over serial, but we don't care about that cos it's not MIDI hehe!)
[Edit to add: (again IIRC, it's been a long road!) Midi Maze was around at the launch of the ST, or very close, to help promote the machine's Mad MIDI Skillz...]
And the early Demo scene stuff. Yeah, the Amiga had a place in music history to be sure, but the XT seems to have been used by as many pros as the Amigas did for video production(Toaster, as many mentioned already).
The later machines were pretty awesome, but the 500's blew away most of the other machines of that era from a user experience standpoint. PC graphics were broken, most of the Mac's people could actually get their hands on were monochrome, and many of us state side were still on an Apple ][ series or a Commodore.
The cool kids had either an Amiga 500 or a Apple 2 GS. I think I still have an old XT chassis layed up with my old Apple ][ e and my godfather's Osbourne CPM machine. I had Amiga envy till high school when I started building PCs from parts.
The ST only had MIDI as an afterthought because its sound chip was so shite and Jack Tramiel just wanted the machine out ASAP.
And as for the ST being rock solid, that only happens because the software has to take over the entire machine, the Amigas timing was rock solid with OctaMED running over Workbench and anything else you had running.
Yes, proper Amiga fanboy here. Wish I'd never sold my A4000T but I wanted to buy a house :(
As someone who ditched the ST for the Amiga (because it was generally a much better machine), I'll still concede that there were apparently legitimate technical reasons that the Amiga was less suited to MIDI than the ST, whatever the latter's inferiority in other respects.
Mr Strorm,
You've demonstrated more than adequately your hardware knowledge of both families, the STs (the best) and the Amigas (acceptable, at a push).
May we acknowledge that, regardless of differences in MHz, we are part of the same 68K family? I suggest, in the spirit of the season, that all Atari[1] and Commodore devotees come together, handshake, CISC and sync our common tunes together!
To you, sir, and all here around the Vulture's Aerie, I say, Merry Christmas and may 2023 bring you loss-free I/O.
[1] Still the best! <Sorry! Just hadda!>
Interestingly, I was about to post a link to this comment which made clear that- if anything- I was actually more of an "Atari" (*and* "Amiga") fan than a "Commodore" fan. (It just depends *which* Atari era and machines you have in mind).
But I notice that you actually replied to that, so you've probably already seen it(!)
You're right that Atari is the best... but only when it comes to the 8-bit computers! :-)
Merry Christmas to you too!
Hehe! ...but only when it comes to the 8-bit computers! - I ain't biting! lol
Yeah, the 8-bit Ataris were tops, and the machines I really cut my programming teeth on. Assembler in DLIs & VBIs messing with the colour palettes and display modes to achieve, in the early 80s, HDTV resolutions and one meeellion colours! JK! I can't remember the colour palette details and I have to go out now. I'll check later and update....
Thanks for the warm wishes - I hope you and yours had a good day, and best of luck for '23!
Spot on :) A networking interface, with low overhead, all-you-need protocols.
Were you using STs? I mentioned above creating a hybrid ring/star topology[1], did you try anything along those lines?
[1] Standard MIDI rings, connected via modem or printer ports.
(Has anyone got a few STs I can borrow?! :)
4 STs: a 1040 STFM (with a disc controller upgrade and 4 meg of ram), MegaSTe with an internal 200Mb IBM HD, a falcon 030 and what started as a 520STFM until it was Frankenstein'd with a 68010, a more stable clock, TOS 1.62 and the fixed version of the MMU that made the analogue ports work properly without the weird timing issues...
We basically did a ring, with a master detection protocol, based on timing how long it takes to process a specific set of generate/pass/terminate messages. in that setup, the falcon would pretty much always win, unless it was also doing other stuff. there were usually only 3 of us, so the mega ste would usually end up master (because it was mine, and i wrote the master detection code :) ) and we used the frankenstein as a pass though/ network monitor/ heartbeat server/ scoreboard. (our ring relied on a heartbeat providing timing sync messages to keep everything in order... and the heartbeat messages were tagged by each machine in the ring so we knew which machine was bogging down in our home brew protocol :) )
We also worked out how to stream the game code from the master and then run it, so only the master was loading the code, everything else was running a client that just waited for instructions...
I can't think of any other area of digital technology where I can take a device made in 2022 and plug it into a device made in 1985 (with cables that are still widely available) and they will just work seamlessly. The genius of Dave Smith and Ikutaro Kakehashi was that they realised musicians were used to taking a a guitar built in 1980 and plugging it into an amp built in 1965, or two analogue modular synths from different manufacturers, and for them to just work together - and would expect digital instruments to work in exactly the same way. Built-in obsolesence is something that musicians (where the cult of "vintage" thrives) would never tolerate. A Jupiter 6 would set you back
I can't think of any other area of digital technology where I can take a device made in 2022 and plug it into a device made in 1985 (with cables that are still widely available) and they will just work seamlessly.
This comment was typed on a model M keyboard made in the 1980's and plugged into a PS/2 interface on a PC at work bought about a year ago.
The build quality of those old keyboards was pretty awsome though. Inspired a generation and the foundation of the gaming keyboard industry, which I now use at work with the lights turned down.
Before the excellent modern options we have now I was still using a stack of adapters on an AT board all the way to the early USB era.
I still have a model M I use daily, one on each of my other home systems, plus a couple spares, my parents have one that my dad types away on quite heavily too on a daily basis. The surplus store here would literally get keyboards by the pallet load (mostly Gateways and Dells) and sell them for $1 a pop, so one day they got a stack of model Ms and I got the stack for like $10.
"Lot of old keyboards out there companies are too cheap to upgrade."
When the BOFH needs to LART the PFY, he doesn't use a lightweight modern flexi-keyboard with parentage from the realms of laptops. There are good reasons to keep old keyboards, mass being one of them :-)
I'm surprised, gamers are usually going on about stuff like "n-key rollover" and stuff, which PS/2 doesn't do AFAIK (or maybe it's just Model M that doesn't do it, I wasn't really paying attention). Then again, half the gamers out there claim to have superhuman perception, and even the results of inane stuff like "how tall are you?" on community chat forums puts the average height of the average gamer at around 6'8". :|
I'm probably not a "proper gamer" as I'm quite happy using a Model M (only reason I moved to MX Reds is OH complained about the noise of my overnight gaming sessions... and MX Reds feel like a vintage BBC Micro keyboard, a bit) and my now quite ancient NEC ×1200 monitor. And I'm kinda rubbish at combat and strategy, I like Fallout 3 more than New Vegas and my eyes glaze over when someone starts going on about The Narrative and various other ways where they think video games should be more like a book or tabletop.
But it's enough for me to explore and spend more time modding them than playing (though I kinda suck at Blender too) and gain enough minor notoriety that a dev unexpectedly gave me a cameo which was... disconcerting, to say the least. But still not a proper gamer.
Then again, I'm not a proper audiophile either in spite of immediately being able to tell the difference between mp3 and ogg (albeit using their default settings at the time, around 20 years back). It probably helped that I'd randomly chosen something by ELO as my test track, not realising that it was a particularly telling example, so that was a happy accident; of course 20 years back was (mostly) before that deaf old git Rick Rubin really kicked off the brickwalling obsession. But I'm "that girl" who fell asleep during a Motörhead gig; not sure what I missed, just that I woke up deafer than when I passed out.
> MX Reds feel like a vintage BBC Micro keyboard, a bit
Snap! The BBC keyboard was one of the two keyboards whose feel I wanted to recapture when I bought an MX red keyboard.
(I had an MX Black keyboard that I bought before the MX Red keyswitch had been introduced, and it was good, but always a *bit* too stiff for my taste.)
I can't think of any other area of digital technology where I can take a device made in 2022 and plug it into a device made in 1985 (with cables that are still widely available) and they will just work seamlessly
I'd contend that the DMX512 standard comes pretty close to that. I believe it was standardised a couple of years after MIDI and is in many ways less flexible (it's essentially one-direction only, for example), but it's considerably faster (250kbit/s) and the 5-pin (or 3-pin on cheap kit) XLR connectors used are an order of magnitude more robust and easily-fixable than fiddly little DIN plugs. It's a different beast electrically; it can cope with much longer cables than MIDI, but while MIDI devices are optically isolated, DMX devices are not which can lead to problems in some situations.
M.
Good shout! MIDI 1 is also unidirectional - I can send a CC to an instrument to set a control, but can't read back the current state of the control, for example if someone physically moves the slider/knob, unless the instrument also has the capability of acting as a controller at the same time as an instrument. That's been addressed in MIDI 2.
My official USITT DMX512 specs show the original as 1986 and later as 1990. The revision was to extend the 'Mark after Break' pulse from 4us to no less than 8us which made it easier to generate and receive using microcontrollers of the time without needing external hardware logic.
One issue that became apparent (when I was younger, much younger) was that MIDI Lag was a thing - somethimes, especially if you had a few instruments linked up, there was a noticeable delay between sending the note and the far end of the daisy chain receiving it.
The other awkward thing about it, from my point of view, was that the IBM PC UART couldn't clock at the right rate so you had to have a Sound Blaster, or equivalent, to provide the interface rather than just mucking about with a soldering iron and some sockets. Or not use a PC.
It is, however, a fine example of something that actually JustWorks(tm) and, as such, should be an object of study for many modern software architects.
Beer - because, over the years, MIDI has resulted in the consumption of many beers, nights of passion and other such fallout that occurs around musicians; one more for the road(ie) won't hurt!
I believe that latency was higher on x86 architectures than it was on the Motorola CPUs that were used in Atari ST and Macs at the time, so musicians used these machines. Recording studios (along with DTP shops) were where Apple hung onto life through the 90s.
Because those users develooed a need for a fast interconnect for audio capture (or high Res scanners,) Apple worked with Sony to create FireWire. The first iPod used FireWire because USB wasn't fast enough at the time. The iPhone comes with Wireless Midi baked in - it's child's play to use an iPhone's screen and g sensors as Midi controllers.
Android had terrible latency for years, now it's tolerable. iPhones had a low latency of around 12 milliseconds from gen 1 onwards.
I had vague recollection from back in the day of MIDI timing being more problematic on the Amiga than the ST (despite both being 68000-based). I was reminded of that during the disussion of the Atari Falcon a couple of weeks back. But I still wasn't sure whether this was hardware-related or due to the somewhat more complicated Amiga OS (and the fact it supported multitasking)...?
Fortunately, nowadays we can look up this sort of thing on the Internet and it turns out it was due to a design flaw with the Amiga's timer chips and the serial port.
MIDI Lag was a thing
My memory of that was that it only became apparent if you daisy-chained kit via in-out-in-out connections. If you used the "thru" connector (yes, I'm aware not all kit had a thru connector), lag was much less of an issue because the MIDI data didn't have to be read by the device and then spat out again; effectively the thru connector was just electrically connected to the in connector (opto isolators aside).
In many setups only a few in-out-in connections needed to be made for the controller (e.g. keyboard) and sequencer (or ST etc.) whereas all your sound modules (certainly in a performance situation) were merely "listeners" and could be hooked up in-thru-in-thru-in.
M.
These days, the trick is to use something like a 4 port MIDI->USB adapter.
Hmm.. Recommendations for a good adaptor?
I still have my trusty old Korg Triton Studio sitting in the corner, and would be fun to get that back up and running. Downside is I'm not sure what state it's in wrt software for the MOSS board and PCM modules. I think Korg has the software for download, but then it'd be getting it across to the keyboard when I don't think I've got a working PC with a 3.5" drive any more.
I've struggled with my old Korg Trinity Pro-X - nobody makes software to talk to any keyboard that old.
But.. but.. Internet! Someone, somewhere has probably figured out a way. I hope. I also found I have a hub! Emagic MT4 with USB, MIDI In and 4x Out. Also has a 'Patch' LED, and perhaps most importantly, no power connector. Which is handy because PSU's tend to get seperated from the things they're supposed to power in my boxes'o'bits. But at some point in the past, that presumably worked with the Korg. Another thought was taking the HD out of the Korg and seeing if I could patch it via a PC, but I'm pretty sure that's going to involve IDE->SATA FUN!. Progress sucks sometimes.
But it's turning into a project, and could be fun. I bought the 76 key weighted version because I had the idea I'd learn to play properly. Not just figuring out it's easier to programme Chopsticks than make my fingers co-operate. I guess if I can get it working, that could be something to try again.
I bought a crappy old Roland drum kit the other month, with the intention of plugging it in to my Mac Mini via the USB MIDI cable and putting it all in to Garageband.
The £15 cable didn't work with the drum kit, which was very annoying!
What worked though was an old Alesis keyboard I bought in 2006 which has two MIDI inputs and outputs. I plugged the drum kit in to that keyboard, which was connected to a 2015 Mac Mini, and it all worked fine.
Obviously of course it'd have been better if I still had my Atari ST but I sold that months ago.
The MIDI protocol really is a wonder. So perfect that you never even think of what it actually does (until you read this article), and other than a crappy no name USB MIDI cable, it always works. It's incredible.
I still have my Korg M1 Synthesizer after many many years.
As a synth, one could adjust the playback of its sound oscillators (PCM samples), using ADSR envelopes filters etc. These adjustments were stored as small records of parameters, often generally called "patches" on synthesizers. They weren't the PCM samples themselves, but settings that described how they were played, but more on that later!
I wrote a librarian tool to download and store these parameters records on a computer and to be able to send them back. Which meant I could make many more patches than the synth's memory could store.
I used the M1's MIDI SysEx specification for data dumps and receives, in the manual to implement the program. A curious thing of SysEx is that the data itself could only occupy the 7 bits of each byte, not the top bit which was reserved for control bytes. So a routine to convert a fully 8 bit patch file to a string of bytes only using 7 bits was required of me to write.
I wrote the librarian in ARM2 assembly on an Acorn Archmedes A310, using RISC OS SWI system calls drive the MIDI interface podule I fitted connected to the Synth. The SWI calls were provided by a ROM containing a RISC OS module on the card itself. It worked fine, very rudimentary and was of great help to me when my Korg's battery used by the battery backed RAM storing the patches, needed replacing. I could dump back the whole factory set of patches, combi patchs and songs.
I'll put it on my github, therobyouknow at some point.
Later I worked on a GUI running in RISCOS where the Korg M1's memory was shown as a file system, each file being a patch and folders to group them too. I called the idea MIDIFS, following the RISC OS naming convention of using the prefix FS for any kind of file system like ADFS or RAMFS. I used EasyC++ which was a brilliantly written GUI library from Dave Allison who worked at BT in Belfast, which included some form of multithreading and pre-emptive multitasking, which was more advanced modern alternative to RISCOS's co-operative.
Interesting note for the Korg M1, it had a PCM card slot for factory made sample ROM cards. In 2018, WaveRex came out from a company called Synthastix. This is the exact same form factor as a regular PCM card and can be used standalone with the M1 like the usual factory made ones. BUT, you can connect it to a PC or Mac and with their app download your own samples to it, via USB, to then then use it standalone as said. It's actually a microcontroller with flash memory inside. Amazing innovation after more than 30 years since the M1 launched. No affiliation benefit or referral to me.
I, with some help from a friend called Dave (<waves>), wrote a clone of the old Acorn MIDI module (written in C, implemented to the known spec) that talks to modern USB hardware.
Was quite nice to plug my keyboard (Yamaha PSS-something) into the Pi, fire up Rhapsody4, and play music directly to the keyboard. It's been my dream ever since I got the keyboard, so it's great that it now works nicely.
I'd be interested in what you have to handle SysEx. My Yamaha, I think, abuses that in order to upload .mid files to a little bit of flash memory. I don't know if it can handle custom samples (patches) too. Unfortunately the process isn't documented and I'm not sure how I'd hijack a USB port (it's an XP box) to work out what's going on.
@heyrick I've started putting my ARM assembly code for MIDIFS here: https://github.com/therobyouknow/midifs
I'm working through my code archive so I'll be looking to add more code files as and when I come across them and understand what they are.
So far the file I've added appears to contain the algorithm for representing the data in 7bit form, from 8bit bytes as described.
Cool, thanks
I don't do fancy things like source control. I've only just been weaned off of floppy discs. ;)
My MIDI implementation and the source are lurking in the myriad of bollocks known as my blog - https://heyrick.eu/blog/index.php?diary=20220809
As someone who still has an official printed "MIDI 1.0 Detailed Specification", bought direct from the MIDI Manufacturers Association in 1990, on my bookshelf you can put me down as some kind of fanboy.
Whenever I have had to implement any kind of UART API I have always ensured 31250 baud is made available if the hardware supports it.
My only complaint is, as the official spec was only available to paid-up members, is "no reproduction", it means a lot of example receiver implementations out there don't handle Running Status correctly or fail to accept that some MIDI messages can be inserted within others. They mostly work but not always and aren't actually spec compliant.
Another quirk is that according to spec the end of a note playing can be encoded as a "note off" message, or a "note on" one for the same pitch but with a velocity of zero. Roland did it the latter way on at least some of their equipment, but some devices only expect a note off. I have a filter device on the output of my Roland W-30 that translates note on messages with a velocity of zero to note off ones.
is another ones custom built fit for task custom keyboard.
Why bother with a 'standard' 100 or so usb keyboard running through msft (or apples) notoriously inflexible keyboard interface when a couple of custom keys on a stream or elgato or xkeys deck can do the job, fully customized, almost out of the box.
it's a damned shame that more 'creative/creator' software in the audio, video, photo, & drawing domains don't make it easier to customize their interfaces with macro or direct midi interface capability.
The first bit of hardware addon I built for my Beeb was a MIDI interface. About 20 quid including box to put it in. As well as the usual music stuff, I also used it to do data transfer with a friend's Amstrad CPC vis its MIDI interface. :)
I got it out a couple of years ago to test some music code and update the documentation, and it still works perfectly. Ever since, I've forgotten to put it and the Casio away and it is still sitting on the shelf above the computer, plugged in and ready to power up and dispense Jean Michael Jarre.
Interestingly, did you know that "computer keyboards"- by which I assume they mean MIDI control- on Queen's 1986 album "A Kind of Magic" was credited to "UMI & BBC B". (A bit of searching reveals that UMI was a maker of MIDI interfaces for the BBC Micro).
I'd have thought that they might have been using the ST by then, but Wikipedia says that they started recording in September 1985 (i.e. only two or three months after the ST came out, almost certainly before any decent MIDI software was available, and certainly not enough time to learn and risk using it over an established and well-understood setup).
Anyway, shows that the BBC Micro was in use- at least there, and possibly elsewhere- for "serious" MIDI work on a major album before the ST took over.
When mobile phones first gained the ability to use MIDI files as ringtones, overnight, all the user-submitted files on the internet were stolen and put up for sale. However, when smarter phones evolved, and could play MP3, many MIDI files reappeared for free.
Technically, I guess, if someone makes available a MIDI version of someone else's music, it's a copyright violation, because the MIDI is fully equivalent to musical notation. But that doesn't seem to make a difference in practice.
As very much an amateur musician, I have fun by downloading MIDI songs and singing or playing along. As it happens, I do own a 1980s Roland synth and a USB/MIDI interface, but mostly I use a software renderer and play from the PC.
When MIDI came out it seemed like a bit of a kludge. Its data rate was geared to the properties of the Zilog SCC (serial) chip. It was a simple serial protocol with limited capability and no thought given to message latency or the need for two way communication. It was cheap 'n cheerful but it did the job, no more and no less.
It arrived at that awkward time between the need for a decent data bus to connect musical instruments and the like and the arrival of reasonably priced Ethernet. Industrial interfaces like HPIB were too big and expensive and early networking was both expensive and unreliable. Custom interfaces could bridge the gap but they had to be built and maintained (and needed to be 'gig proof').
MIDI found its way into broadcasting too as a useful extra tool.
In the late 1980s when the Yamaha DMP7 MIDI mixer was first released, MIDI was used as a way of getting around the security restrictions at major sporting events like the Olympic and Commonwealth Games. One DMP7 was placed in the 'secure' athletes area with the mic and audio connected to the broadcast comms. A second DMP7 was placed in the 'tech' area connected with just the appropriate pair of MIDI in<>out cables running down the corridor. The athlete's mic fader was then remotely operated from the tech area 'by magic' with out the need for the sound operator to have physical access to the tightly controlled secure area.
Still have 2 working Atari ST1040e systems with the SM124 mono monitor, Steinberg Pro 24 (with blue dongle) and a Cubase Notator/Unitor Timecode interface. Automated mixdown on the cheap with 8 track audio and film or U-Matic video. Still waiting for that 'emergency' job.
The main reason I know of Midi files is thanks to Kid Radd, I downloaded the offline version of the webcomic years ago, burned it to a CD, unzipped the file on an old computer with Windows 98 SE and Internet explorer and then had to figure how to make Midi files work for several hours. Thankfully as that machine had a compatible Soundblaster sound card all I had to do was install the right files, good thing I had one of those CDs full of drivers and extras for Windows 98.
That reminds me, not being musical at all, MIDI didn't really come into my awareness other than that some games supported MPU-401 MIDI output and at some stage I acquired a Commodore branded PC sound card that was Soundblaster 16 compatible with an onboard MIDI synth that played much better sounding game music from games which had the MPU-401 option. That sound card lasted through a number of upgrades until there were no longer ISA slots on the new motherboards and anyway, the news Soundblaster cards like AWE32 etc were just as good to my untrained ear.
Equipment. All recorded out onto DAT tape which was rather annoying since it meant I never did manage to save a copy of that eldritch horror sound I composed (it still brings me pleasure that I got a C for it but was the only composition they didn't play at our awards ceremony...).
The other point well made is that midi was, and still is cheap. The actual interconnect chipage is so cheap if can be stuffed onto pretty much anything...like the youngests new keyboard bought for £25 so she can drag it to school (we also have a hand-me-down piano at home, but as old as midi is, it's not victorian).
I might see if I can strap said cheapy keyboard to a laptop and see just how far we can take it.
Hi - are there any practical uses for the midi web browser extension for FF now? Or does this just open the door to future use cases using the browswer which have yet to be developed.
I have a midi controller that looks like a sax (WX-7) and sound source from Alesis (MIDIverb) and a USB to midi interface for the laptop - which I used with sequencer programs in the past.
Does downloading the midi extension enable some kind of control or visibility from the web browser now??
The advantage of the MIDI interface is that it was done in hardware. Speccy serial was always flakey as it was done in software, bit-banging a single-bit I/O port, unless you built a custom hardware solution. MIDI serialisation is very reliable, the baud rate was specifically chosen to be 1MHz divided down by 32. Set your ACIA to do divide-by-16 and feed it from 1MHz divided once by 2.
MIDI signal is sent through an optocoupler at the receiving end, which derives the signal voltage for the galvanically isolated receiving kit.
Single-pair shielded cable (shield connected only at sender side) meaning no ground loops, and cheap, ubiquitous microphone cable can be used, up to ~30ft IIRC.
In audio, ground loops can be problematic, but smart design choices meant MIDI never made that problem worse. Kudos!
.deb packages The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
