I just learnt something about MIDI

[Wiu]

So I was just reading the S.O.S review of the Vermona Perfourmer MKII (I've wanted one for fucking years) and got to this:

'The review model had been fitted with the optional CV/Gate inputs, meaning that MIDI can be bypassed entirely, should you so wish. Why would anyone need that? Well, super-tight sequencing is one possible answer. If you ever aspired to play four notes simultaneously via MIDI, sorry but you're out of luck, it can't be done — via a single MIDI port, anyway. Shock, horror, I hear you cry, but it's true! MIDI is a serial interface and therefore everything that happens happens sequentially, one event after another. This review isn't the place to muse over the subtleties involved in perceiving a few milliseconds between fast, percussive hits, but if you've ever thought simultaneous voices played by a drum machine sounded different when replayed via MIDI, then perhaps these optional jacks are aimed at you.'

I never knew that about MIDI. It's not something I've ever noticed. So assuming that it's a tiny 1ms delay to the following note, does that mean that, for example, my Roland JV1080's 64th voice would delayed by about 64ms? Surely that would be quite noticable? I'm waffling. I'll stfu now

[Lost to the Void]

MIDI is well shit as a protocol.
There was talk in the mid 90's about a new protocol. MIDI 2.0 but it meant cooperation between all the manufacturers and nothing ever seemed to happen.
With the new second wave of hardware children all getting synths from santa, it's probably about time someone brought out a new synth sync protocol, seeing as we are in a new millennium.

[helloitsmeagain]

innerclock has some interesting reading on timing

http://www.innerclocksystems.com/New%20 ... itmus.html

[innovine]

This review isn't the place to muse over the subtleties involved in perceiving a few milliseconds between fast, percussive hits, ...

So assuming that it's a tiny 1ms delay to the following note, does that mean that, for example, my Roland JV1080's 64th voice would delayed by about 64ms? Surely that would be quite noticable? I'm waffling. I'll stfu now

Not really. Midi is 31kbaud, which is about 4 bytes per 1ms. Depending on the synth, thats a midi note or two, per 1ms. I don't know abopt your jv1080, but you would need to send around 60-100 notes, at once, to see a delay of 64ms. Many synths will do extra voices per note, unison mode for example. The article is talking about number of simultaneously transmitted notes, not voices.
say you play a chord with four notes, and manage to hit all at exactly the same time. Each note will be transmitted one at a time, and the process will take 2ms. So the article you mention is true, in a way, but if you can play notes to a 2ms accuracy you're a goddamn talented mofo.

On the start of a bar, you might have a kick, a crash, a hat, a bass note, a 3 note chord and a sample, for example. Total time to transmit all this, 4ms, about the same as a low latency soundcard. It does smear out the initial transients, technically speaking, but its rare you need to give a shit. Note that many sequencers will transmit stuff in track order, so putting your important notes like kick and bass on lower channels, and pads and stuff on higher ones means the important data gets played first. In some cases, if you have a few notes in between your kick and bass, sometimes, but not others, the few ms delay can introduce audible phasing issues on some hits, but not others.

One point the author of above article failed to mention is that these things don't necessarily have to be seen as a bad thing. Lots of analog goodness comes from oscs that drift. Who's to say that hardware 'groove' and such voodoo isn't helped by slight looseness in timing, which your sample-accurate daw gets perfectly, lifelessly correct?


Saying that midi is shit slow is a bit like saying television is a serial sequence of photos, one after the other. Technically true, noticible on occasien, but not something you lose sleep over. 100hz tvs exist. So does OSC. The reason midi is still around is that it does the job ok, and has a very decent legacy.

[Lost to the Void]

But it s is still kinda crappy. Really long overdue for midi2.0 now

[quest]

Anyone follow NAMM? There is something called MIDI HD in the works, could solve all our problems! I also became frustrated with MIDI base specs, I turned to OSC for a lot of my work or synths which went way out of their way to do a good MIDI custom implementation! (Alesis Andromeda uses NRPN's to transmit hi-res data for every single knob on the surface!)

http://www.synthtopia.com/content/2015/ ... milestone/

[Barfunkel]

I'm happy with MIDI. It's as plug and play as it gets and I've never had any jitter, sync problems (except when trying to sync a computer to hardware, which I very rarely do), slowness or other problems. I know it sucks technically speaking, but as long as it works and I don't get any audible problems I'm not complaining.

[seanocean]

Speaking of standardization, does anybody know which way a midi port is supposed to face? Seems like if they would agree on the language they could agree that the port should be the same way as well. It's weird, roland has the 5 pins down, sometimes korg has them sideways, and most of the time DSI has them up.

Odd part of gear history that seems to be a mystery.. or maybe there's an answer in engineering that i don't know of.