USB MIDI without a computer?

[OregonJim]

Hi all,

If one has two keyboards with only MIDI-over-USB ports (i.e. no 5-pin MIDI), is there a way to connect them together so that MIDI out from one device goes to MIDI in on the other device WITHOUT using a computer/DAW in between? Generally, two USB devices can't talk to each other, only to a host (computer).

If, as I suspect, the answer is no, then are there any devices on the market that can be placed between them to accomplish the connection (again, not a computer)?

Seems to me that the industry-wide switch to USB-MIDI has crippled much of the functionality that MIDI was designed for in the first place. Now it seems you need a dedicated computer in the mix to do what was once a simple task...

[andyg]

This question has come up before on various forums. As far as I recall, the consensus on this is that it can't be done, but I've never tried it running from the 'To Host' port on one keyboard to the 'To Device' port on another. In theory, it should work as data being transmitted and received would be no different whether transmitted/received to/from a DAW or another keyboard.

Anyone tried it?

[KC9UDX]

I avoid USB MIDI like the plague. Every time I've used it, I've been unable to deal with the seemingly random latency. I don't know how anyone can make use of it.

[OregonJim]

Well, after several hours of digging, I've answered my own question - "yes and no".

Yes, there are 2 or three commercial devices out there that can act as a USB host to connect 2 USB-MIDI keyboards together; problem is, all devices have to be class-compliant. Plus they're rather expensive and kludgey. And no, you can't just connect the USB ports of two keyboards together.

andyg - most of the device ports on keyboards are for data/file storage applications only - I don't think MIDI or audio data streams can be directed to the device port (at least on the half dozen or so keyboards I've played with). In theory, you are correct - however, the keyboard firmware has to be written to support redirecting the streams. Anyway, one of the keyboards I want to use has only a host port (and isn't class compliant)...

KC9UDX has it right - USB-MIDI is to be avoided like the plague. I'll not buy another keyboard without standard 5-pin MIDI ports, that's for sure!

[KC9UDX]

[AllenAnalog]

As far as I'm concerned, manufacturers who deviate from the connector specified in a published standard, whichever one it is, should be taken out to wood shed and...

When you are working on stage, in a studio or in any other environment where equipment is set up and taken down with some frequency, using connectors with a new function that matches a historic use (like microphones and headphones) is a recipe for someone plugging in non-compatible equipment and potentially letting out the smoke.

When you see a MIDI 5-pin DIN connector or cable on a musical instrument you immediately know what it is for.

And this isn't the only performance-related equipment that has been bastardized by manufacturers using connectors that don't meet the standard. The DMX-512 standard for lighting control explicitly specifies a 5-pin XLR connector. Even though all five pins are not used, it was chosen to differentiate it from other (audio) equipment being used in the same venue. The some genius manufacturers decided to use a 3-pin XLR, just like is used on microphones and balanced audio lines. Hey, we're only using 3 pins, why have a 5-pin connector?

Even worse, other manufacturers decided to put DC power on the unused pins of the 5-pin XLR connector but that use is not in the standard so all bets were off as to what the voltage and polarity were. So once again you ran the risk of blowing stuff up and the universal plug-n-play that was part of using a standard was lost.

I find that I have to be much more careful about looking at the photos of the connector panels on equipment these days to avoid buying equipment with non-standard connectors used for well-defined standards-based functions.

USB use was fine when the functions that were supported through that connector were ancillary and not for MIDI. But once DAWs started becoming popular, some manufacturers wanted to put everything on the USB connector and save the $3 for the MIDI connector and circuitry. It just breeds confusion, lack of compatibility and, as the OP noted, frustration.

[OregonJim]

All I wanted to do was slave one keyboard to another - the most basic of all MIDI tasks. Unfortunately, this basic feature no longer works with the vast majority of currently manufactured keyboards that sport USB-MIDI jacks. Unless, of course, you add a computer into the mix. I'm not going to do that while playing live - the list of negatives is too long to list... Anyway, Merry Christmas to all, despite the frustrations! :)

[Momboc]

What keyboards are you trying to connect?

[andyg]

Yamahas fill transmit and receive MIDI data via USB, and a few will transmit and receive audio. But that won't help if they won't talk directly. :(

[twnelson]

As a computer hardware & software engineer, I feel compelled to inject my perspective on the MIDI-over-USB topic to clear up potential misconceptions.

The underlying cause of latency issues with MIDI-USB is rarely the hardware, as even the lowest speed USB connection (12 Mb/sec) is more than capable of transporting up to the full 16 channels of MIDI data simultaneously with consistently less than 1 msec of latency on a properly designed interface. I know because I've tested this as part of the development of my own MIDI-USB input and output boards for console conversions, using the Linux OS to enable direct performance measurements with minimal software interference.

Without going into the gory technical details, my experience has pointed to poorly designed and written USB software components in the most popular contemporary PC operating systems as the primary source of these latency problems. In some cases, the latency was so high and inconsistent that I had to question whether the vendor even cared about MIDI-over-USB, especially when handling MIDI data streams between multiple sources and destinations.

On Windows, the only way I could get reasonable latencies was to use the JACK Audio Connection Kit (essentially a software audio/MIDI patch panel) with ASIO drivers to 'plumb' everything together for testing. Mac OS was somewhat better, although it still took a lot of tuning to get the latency down.

--- Tom

[OregonJim]

What keyboards are you trying to connect?

I have several - for example, a Roland RD-2000 and a Yamaha PSR-EW400. The Yamaha only has USB-MIDI. The Roland has both USB-MIDI and 5-pin DIN. I'm looking for a generic solution, not a device-specific one.

I refuse to use a computer to act as a USB host for this simple task. But, the more I think about it, a Raspberry Pi might be the ideal solution. Some fairly simple Python code could get two USB-MIDI devices to talk to each other. Yes, it is technically a "computer", but it eliminates most of the drawbacks of using a conventional laptop in live situations (knocking/tripping over, being stolen, extra stands/cables/power supplies/space, constant OS upgrades breaking things, random latency issues, crashes, etc, etc...).

[John Kinkennon]

Check out the USB audio interfaces with MIDI via USB and the MIDI merging devices from MOTU (as an example) for an up to date perspective on MIDI over USB. The old 5-pin connector is a dinosaur as MIDI via USB is at least 200 times faster than the old 31250 bit rate that allowed a throughput of one MIDI transport per millisecond, give or take. There is really no excuse for latency over a couple of milliseconds with current, updated, PCs or Macs.

I've run wireless MIDI between Raspberry Pi Zero's with promising results as well. My experience is with organ consoles running Hauptwerk. I design my own encoders and decoders and my current console connects four keyboards, pedals, stops and pistons, plus organ status and control information, all on a single USB cable. That one cable and power goes to a remote audio interface and speaker power control relays. No huge snake to remote this console.

EDIT: OregonJim, I see you are living nearby. Look up my contact info on the web site below and come visit sometime.

[John Kinkennon]

To get back to the original question, the Raspberry Pi approach might work well. The keyboard USB would have to be class compliant -- no custom drivers. With the $10 price tag I'd set up one Pi Zero for each keyboard with an OTG adapter so each Raspberry Pi can present a MIDI host port. Then each PI converts the MIDI packets to TTL in and out at either 31.25k or 115k. Cross-connected they should work.

This is kind of the reverse of the usual USB-MIDI adapter cable that connects to computer host ports. In this case the connection is to a pair of device ports which is harder. Seems like someone should have this on the market...

EDIT: Here's the Kenton USB host device -- pricey for two of these: http://www.kentonuk.com/products/ite...usb-host.shtml

[KC9UDX]

I run all my Pis headless. I can VNC in from a "smartphone" if I have to. But usually I use an Amiga 2000. I don't normally need to do that, and if I used one for converting USB to proper MIDI, I don't think I'd have to.

[KOC62]

There may be other solutions, like using a Teensy3x processor that is much smaller than a Raspberry Pi. This hardware could certainly do it. However, I'm not fully conversant with what exactly this midi - usb interface is to accomplish. I would need to learn the technical details for me to explore this midi-usb interface further.
e.g. http://codeandlife.com/2016/01/17/mi...ith-teensy-lc/

Peter.