Can Wireless MIDI Be As Fast As Wired MIDI?

http://www.youtube.com/watch?v=ABPu3h-Y768

Can wireless MIDI be as fast as wired MIDI?

That’s one question raised by the new PUC wireless MIDI interface. The PUC is a wireless CoreMIDI interface that connects any MIDI hardware device to CoreMIDI-compatible apps. 

This video, via JamStikVideo, captures some of the preliminary latency tests Zivix is doing to demonstrate the MIDI transmission times over WiFi on a MacBook Pro (i7) and an iPad 2 (iOS6) using their new WiFi MIDI interface, the PUC.

The PUC is being developed as a crowd-funded IndieGoGo project.

Let us know what you think of this comparison. And if you’ve got other concerns or other things you’d like to see demonstrated with the PUC, leave a comment! For a crowd-funded project like this to be a success, the developers will need to address the types of questions that Synthtopia readers bring up.

24 thoughts on “Can Wireless MIDI Be As Fast As Wired MIDI?

  1. Beware: The figure of merit for a real-time system is not the *average* case latency, but the rather the worst-case. And the worse-case on Wifi can be extremely bad indeed under congested network environments.

    In the case of MIDI, the worse-case events are an issue. These are events that occur significantly (in music terms) late or if they’re using MIDI over RTP, not at all.

    A Wifi-based MIDI interface is likely to work acceptably in many environments, but you just don’t know when (or more importantly, where) it will start falling over.

    1. What about when you “create a network” and connect through that? Doesn’t that get rid of the problem of a congested network in your area? I’ve tried MIDI through wifi by creating a network specifically for that purpose, but the latency has always been unbearable. I would love to be able to control my softsynths on my daw with a couple apps I have in my iOS (like thumb jam)…

      1. No. The issue is radio spectrum and interference – and the Hidden Node problem – which tends to cause collisions. I have about 15 access points in range at home, and no doubt many more that could interfere.

        Clearly not having devices on ad-hoc wireless network will help though.

    2. My worry isn’t the latency itself but the inconsistency of the results. Wired midi at 10 ms every time, but the puc was different every time. This could potentially be an issue. You can’t have your latency jumping around like that. The range of variance was pretty small, however, this was a simple test. How much does it vary when you start piling up notes and controller data.

  2. Ugh, no. Just… no. I’m not even going to get into all the ways this video is just wrong, I’ll only say this: The issue with MIDI is NOT latency, it is JITTER. This means (RMS) average *deviation* in delay times, and by their own admission, the Fast Track over USB smokes the PUC. (zero jitter! wow, I gotta get me one of those!)

    But screw it, this isn’t really a product that anyone reading synthtopia would seriously use as a reliable MIDI device in the first place. Which makes me wonder what the point of such a video is, if not simply to confuse people.

  3. why aren’t they doing a more scientific test? i.e.many more than 3 experiments. under different cpu loads etc. the results are not very useful.

  4. The gap between the MIDI signal and the returning audio result appears to be largely dependent on what device and synth you are running. Wireless MIDI and cabled MIDI are basically the same thing, delay-wise.

    5ms or 20ms from keystrike to attack is virtually instant, considering the keybed triggers a sensor, the sensor feeds a converter, which sends it out of a MIDI cable or USB cable, into a wireless interface, into a computer’s WIFI card, which routes it into software, which reacts by creating a signal, and converting that into an analog waveform. It’s a lot of steps, and they are done fast!

  5. @Paul van Nugteren

    USB MIDI devices often build multi-event packets and also have to wait for the host to give them bus access. About 1ms latency is typical for USB interfaces – and this is entirely reasonable when one considers the wire-time for NoteOn on traditional serial midi is ~1ms too.

    It’s fairly obviously that the soft-synth isn’t the same between iPad and MacBook Pro. The iPad is absolutely capable of going lower than the figures we’re seeing here – and in fact is pretty much unlimited in this regard other than by CPU performance. For a Mac, a 32-sample buffer they claim to be using is a buffer latency of less than 1ms, so I’d expect we’d be looking at about 3ms max. Both midi and audio render are real-time threads on both platforms (or at least ought to be if the soft-synth author knows what they’re doing) and should unaffected by other system load (unless it’s real-time as well).

    1. They could have placed some rusty old satellite dishes, Russian microwave ovens, and MRI machines in the way, see if they could get a few more ms of delay. You know, real world situation.

  6. m-audio soundcards have notoriously bad midi drivers.

    why not use an RME? or ploytec? even the Midiman Midisport has better drivers than that soundcard.

    not to mention the evidence of jitter with these tests.

  7. If I may respond to some of the comments here without getting too deep into the weeds… 1) Sorry if you’re upset about the audio on the intro – we whipped this up in less than a day and the point was to put the results we’ve been seeing in our testing, not win an emmy for recording quality. 2) Yes, we went fast and dirty on the video because – well, most viewers only watch 90 seconds so we can’t dive into the minutia of our test rigging, the number of reps, the various apps, the various iPads, the plug-ins, etc. We wanted to provide our average results for average users, that’s why we’re not using premium hardware for MIDI interfaces or multiple machines. Besides, the tests we didn’t show are yielding very similar results per device category. 3) Yes, latency is one factor (and the one that most popularly gets brought up) but you’re correct in stating that jitter is another component to the equation. However, when dealing in the sub 10ms range we’re really talking very tiny increments of time, and I’ve yet to meet anyone who can tell me they perceive a difference of 2ms in this note-on and the next. If you can then I don’t know of any MIDI interface that can please you because ultimately they all deviate at some point or another. 4) iOS and OSX are very different worlds for developers, and we’ve been running a considerable amount of testing across the board in regards to iOS6, the beta versions of iOS7, and the general release of iOS7. Like lots of iOS audio developers have issued statements about the audio handling of iOS7.0 not quite being fully baked yet, we’re seeing similar issues on the MIDI side and are attempting to gain some visibility as to what’s going on with the powers that be. Everything tests with higher audio latency figures for iOS in general; hardware, WiFi, Bluetooth. We’ve tested some proprietary apps where we can adjust some parameters with some success, but there’s some variables that have come in late that we’re working to diagnose and address. 5) The video demonstrates that latencies are extremely low on a MacBook i7 running Logic Pro X w/ an EXS plug-in. We hope that indicates what our transmission speed is capable of, given good audio handling abilities. 6) We’re working on more lab videos, distance performance is in the cue. 8) Once again, the PUC might not be for you if your timing discrimination is sample accurate 100% of the time, but if that was the case you’re probably not busy in the iOS world anyway (which is where the PUC is primarily targeted and design towards – it solves the “one more specialty adapter set in my already cluttered gig bag” problem.) The fact is that it’s a simple solution for a common problem, it’s relatively inexpensive, and it gets the job done within the same window of jitter and latency as the wire in iOS, and so comparable in OSX that I don’t know many musicians who could tell the difference without seeing the waveforms in a DAW.

    I enjoy Synthtopia a ton, it’s a regular “go-to” for information on audio & MIDI related tech for me. Thanks for your feedback, I appreciate it – we’re all looking to make and improve great product ideas for musicians and those who want to be musicians!

    Christopher Heille
    Music Product Specialist, Zivix LLC

    1. Christopher

      Thanks for responding to reader questions – it helps readers understand the technology and put your testing in perspective.

      As you have news, please let us know!

  8. I could very well be mistaken but I feel this was rigged. If you notice in the video right before they switch to the wired connection the guy says “we’ve just changed our grid division on the oscilloscope to 5 ms” to me this sounds like the measurement interval, which if that is true, the real reading was anywhere between 5 and 10ms for wired, probably beating the puc. this video made me question the legitimacy of their latency test and the quality of the product even more. plus, were they using ASIO drivers for the fast track??

    and regardless of latency other people have also said it. jitter is a serious problem. have you ever tried to use TouchOSC over a wireless connection? not every message gets transmitted. things gets lost in space.

    maybe I’m just biased because I like wires but this product seems sketchy.

  9. Hey Tim – we changed the interval division between tests because it illustrates better on the scope. And if you like your wires (believe me, I do too when the application demands it) but there’s a bunch of times I hate the wires… like when one of the iPad mini’s took a spill this week on a wired connection. Snap, crackle, iFixit.com search – you get the picture.

    Christopher Heille
    Music Product Specialist
    Zivix LLC

  10. The wired MIDI device you show in the lab test video is an M-Audio Fastrack Pro. I used to own it and got rid of it quickly because it is USB 1.1, is it not? Do you have wired MIDI tests with USB 2.0/3.0, Firewire, Lightning, and Thunderbolt MIDI devices?

    Theoretically USB 1.1 plenty fast enough for a MIDI signal but that’s not always the way things work in reality. For example, my USB 2.0 memory sticks are faster when I run them on a USB 3.0 hub vs. a USB 2.0 hub. On the surface that doesn’t make sense but it makes perfect sense when you consider that the overall USB 3.0 circuitry is better than the USB 2.0 circuitry.

  11. Fashion photography is considered as tricky by people but this is the one which requires
    fotografia matrimonio to be very attentive and also very good in each and every photograph
    he is clicking. They capture the real moments not the choreographing preconceived moments for last couple of years.
    Wedding photo journalism is the motto and style for this company.

Leave a Reply

Your email address will not be published. Required fields are marked *