USB Class Compliance: For Discussion of Issues & Bugs

Dmcghee wrote:Yes, the MPC records the way it was designed to record. I only hope Akai implement 32 audio tracks of disk streaming in stand alone.

Ok let’s break it down to sections with examples..
Let’s say you have a 32 sounds drum program, can all be played and controlled simultaneously on the X32?Can you record it’s motorized faders action as the mix progresses?
On the inputs side let’s say you have a studio with a bunch of synths, drum machines, modulars etc. , can you record 32 input channels simultaneously on the MPC via X32?

stewes wrote:

Monotremata wrote:
MOTU as well. Their "class compliance" only extends to MacOS and iOS, doesnt work on Windows or Linux. Although SOME MOTU interfaces do seemingly have Linux drivers created by the community, just like the old eMagic AMT8/Unitor 8 MIDI Interfaces (thats why they work on the MPC with no drivers). My 828MK2 fired right up and worked just perfect under Ubuntu, and MOTU themselves HATE Linux, too bad it was Firewire or it wouldve worked with the MPC too.

Hey, wow...is this really true? An old Emagic Amt 8 or Unitor 8 mkII is working connected to an MPC with firmware 2.10? Also a Motu 828mkII USB is also working with MPC? That would be sooo cool!

I have another question: If you connect your Behringer UMC1820 to the MPC and activate it in the menu are the onboard audio-connectors on the MPCs back are switched off then? What is with Midi and the CV/gate connecters? Are they also switched off if connected to UMC1820?

Unfortunately I can confirm the MOTU 828 mkII USB does not work. It might be class compliant for Mac, but certainly not Windows or Linux. The MOTU does think it's connected to a computer and gets locked to 44.1, but no sign of it on my MPC Live.

I'm looking for a solution to get 24 channels of ADAT I/O in and out of the MPC so I can hook up all the converters in my studio. The RME Digiface USB is a bust - no class compliance at all. The MOTU LP32 should work in theory, but based on reports of MOTU AVB stuff not working, it'll probably be the same. Only other possibility I've found is the Cymatic Audio Utrack24, which apparently works as a class compliant interface on Linux. They're quite cheap - I'm gonna try to test one. If it works could be a great solution for relatively high channel count with the bonus of a live recorder thrown in for good measure.

Fred187 wrote:Unfortunately I can confirm the MOTU 828 mkII USB does not work. It might be class compliant for Mac, but certainly not Windows or Linux. The MOTU does think it's connected to a computer and gets locked to 44.1, but no sign of it on my MPC Live.

I'm looking for a solution to get 24 channels of ADAT I/O in and out of the MPC so I can hook up all the converters in my studio. The RME Digiface USB is a bust - no class compliance at all. The MOTU LP32 should work in theory, but based on reports of MOTU AVB stuff not working, it'll probably be the same. Only other possibility I've found is the Cymatic Audio Utrack24, which apparently works as a class compliant interface on Linux. They're quite cheap - I'm gonna try to test one. If it works could be a great solution for relatively high channel count with the bonus of a live recorder thrown in for good measure.

Hold up a second, did you get a message that it didn't recognize the 828 at all? If the MOTU switched its clock and didn't give you an error that thing might be working. Go into the Preferences, Audio Devices and check the drop down menu and see if the MOTU is available. Switch it over if it is and you might be good to go.

The MPC doesn't indicate you are using an interface at all, other than in the Audio Devices prefs. The outputs still show up as plain old Output 1, Output 2, Output 3, etc. You just have to go in to the prefs and switch it over if you didn't already.

Here's what the Focusrite Scarlett 18i20 audio dropouts look like:



In the test, I made the Odyssey plugin produce a constant steady waveform (by pressing down a pad and going to Preferences, which leaves the note unreleased and playing indefinitely), recorded the output with a Zoom H4n recorder, and loaded the wav file into Ableton Live for analysis.

There is a similar dropout every 40+ seconds, and each dropout lasts about 21-22 milliseconds, which is very close to 44100 / 2048. There was some variance in the interval between dropouts:

0:29:518
1:10:109 (+40.591)
1:50:459 (+40.350)
2:30:692 (+40.233)
3:11:193 (+40.501)
3:52:141 (+40.948)


In my Mac, the same audio interface works without any glitches like that.

MPC One version 2.10.0.
Focusrite Scarlett 18i20 version 6.1b.
Connected directly with the Focusrite supplied USB cable, no USB hub in between.
The MPC's "32 Inputs/Outputs" option was NOT enabled.
No MIDI devices were connected to either the Scarlett or the MPC.
Playback was not on. The MPC was in a stopped state. No MIDI I/O should have been happening, because nothing was connected and nothing was playing, except the one stuck Odyssey note.

I seriously hope Akai fix their stuff. The Scarlett 18i20 is an affordable and fairly popular audio interface. The Class Compliant "working/not working" list thread cannot be trusted, because people are apparently not actually testing the interfaces.

HouseWithoutMouse wrote:Here's what the Focusrite Scarlett 18i20 audio dropouts look like:



In the test, I made the Odyssey plugin produce a constant steady waveform (by pressing down a pad and going to Preferences, which leaves the note unreleased and playing indefinitely), recorded the output with a Zoom H4n recorder, and loaded the wav file into Ableton Live for analysis.

There is a similar dropout every 40+ seconds, and each dropout lasts about 21-22 milliseconds, which is very close to 44100 / 2048. There was some variance in the interval between dropouts:

0:29:518
1:10:109 (+40.591)
1:50:459 (+40.350)
2:30:692 (+40.233)
3:11:193 (+40.501)
3:52:141 (+40.948)


In my Mac, the same audio interface works without any glitches like that.

MPC One version 2.10.0.
Focusrite Scarlett 18i20 version 6.1b.
Connected directly with the Focusrite supplied USB cable, no USB hub in between.
The MPC's "32 Inputs/Outputs" option was NOT enabled.
No MIDI devices were connected to either the Scarlett or the MPC.
Playback was not on. The MPC was in a stopped state. No MIDI I/O should have been happening, because nothing was connected and nothing was playing, except the one stuck Odyssey note.

I seriously hope Akai fix their stuff. The Scarlett 18i20 is an affordable and fairly popular audio interface. The Class Compliant "working/not working" list thread cannot be trusted, because people are apparently not actually testing the interfaces.

There isn't anything for Akai to fix, its 'built in' to the OS itself. Its up to the manufacturers to make their devices comply with it. Some manufacturers like MOTU only make them class compliant on a Mac, RME goes the whole distance, so does Presonus apparently. Other folks here have the 1820 working, try checking the 32 in/out button because the interface has 18/20 waiting to go, maybe that will make a difference?

What exactly is "class compliant on a Mac"? I thought that the "class" in "class compliant" means a USB Device Class, which is defined by the USB standard. Stuff like this https://www.usb.org/sites/default/files/audio10.pdf
No idea what documents were actually used, and how detailed and sufficient they are, but it's not defined by Apple. Are you suggesting that Akai users should expect each audio interface manufacturer to tailor their products for the particular quirks of the Akai MPC series? Instead of expecting Akai to investigate what the differences in operation are between hosts where the interfaces actually work, vs those with glitches? Or are you suggesting that Akai are not able to change how their products operate regarding handling of USB audio interfaces?

I say, other people do not have it working. They claim it's working, but they just haven't tested properly. Or at least I have not seen credible evidence, just someone saying "works" in the list thread. I have no reason to assume that proper testing has taken place, quite the contrary. I took the MPC One to the store and tested it before buying the 18i20, and I didn't notice the dropouts there. I just switched through all the outputs and they seemed to work. Only after actually using it for half an hour or so, I realized something might not be right and I started investigating further.

Btw, the same dropouts happen in the "32 Inputs/Outputs" mode.

I think this thread needs to be re-done from scratch, with detailed instructions on what exactly has to work, and how exactly that has to be tested, and that evidence has to be provided. Or if it's just about user experiences, everyone who claims that something "works", should provide specific details of what exactly has been tested, for what kind of work the interface has been used, for how long, etc. It's not enough of a test if the interface shows up in the selection list, or if some sound comes out.

Class compliancy means that the device (allegedly) complies with the USB standard's specifications of how a device of the audio Device Class must operate. With standards-compliant devices, it is enough for hardware (and software) manufacturers to make their products comply with the interoperability standard, and not with each and every individual product. In theory. But practice has shown that there are differences in how the standard is interpreted by manufacturers, and so, things may not work in practice, even though class-compliancy is advertised.

I'm slightly pissed off right now, I'm afraid I have no other real option but to return the Focusrite Scarlett 18i20 for a refund. (unless miraculously Focusrite or Akai react in some way)

It doesn't feel realistic to expect that all audio interface manufacturers will test and tailor their products to work with Akai MPC boxes. Apple is big enough for things like that, Akai is not.

I sent a support request to Akai about this, asking if there's something to be done, or if I should return either the 18i20 or the MPC One for a refund.

HouseWithoutMouse wrote:I think this thread needs to be re-done from scratch, with detailed instructions on what exactly has to work, and how exactly that has to be tested, and that evidence has to be provided.

You're asking too much of people, very few will have the time or inclination to do this.

Personally I feel it is Akai should be doing this detailed testing and then publishing an official list of compatible USB devices. Ultimately it's their product and their marketing that is claiming USB class compliance in the MPC. Fairly pointless having this feature unless there's a decent selection of compatible devices to choose from.

Anyway, let's keep this on topic - for the moment, people please carry on giving feedback where you can, and try to indicate what kind of testing you've done on this.

I'm going to test tonight quickly to see of my Motu828es will show up.. even if it does work I do need a interface connected to another interface what I really want is to have in standalone 16/32 outputs to your daw without using the software which doesn't make sense in the first place..

While waiting for Akai and Focusrite to respond, I did some more testing. This time I synchronized the Focusrite Scarlett 18i20 to an external word clock via the coaxial/RCA SPDIF input, and recorded over 21 minutes of sound output from the same test project



The 18i20's front panel shows a green lock light, and the control panel shows it's locking to the external clock all right, when I connect it to the Mac instead of the MPC One.

Now the dropouts are spaced at 5+ minute intervals, but they are the same length, around 21 milliseconds. Here are the time positions of the dropouts in this external-wordclock test

3:23:663
8:38:790 (+315.127 sec)
14:17:523 (+338.733 sec)
19:52:324 (+334.801 sec)

What could cause this? Clearly the MPC One isn't waiting in a locked state for the USB audio clock during the dropout. The Odyssey's oscillator proceeds normally, just the output is silent.

Good point about Akai (rigorously) testing and publishing the results. That would be ideal.

For now though, maybe things should be approached kind of as a traffic light system: the reds definitely don't work (TR-8S for example), amber might do but not tested to destruction (Axe IO, maybe 18i20?), green have had more pronounced stress testing and appear to perform well (Model 12, given they are on the akai vids?).

So okay finally downloaded and installed the update .. I don't have time tonight to dive into all the features.. I did however try my audio interface Motu 828es first problem like others have mentioned it need 44.1k bitrate okay changed now I can see and use 32 outs but guess what like others reported less 6 bars drops outs and clicks and glitches.. I won't spend anytime finding out why I don't think this feature is full tested and I'll just go back to using my X the way I was before.. only way this will work of they open up the outputs in standalone using the audio interface built-in the MPC then send the outs to your audio interface.. oh well still love my MPC..

HouseWithoutMouse wrote:While waiting for Akai and Focusrite to respond, I did some more testing. This time I synchronized the Focusrite Scarlett 18i20 to an external word clock via the coaxial/RCA SPDIF input, and recorded over 21 minutes of sound output from the same test project



The 18i20's front panel shows a green lock light, and the control panel shows it's locking to the external clock all right, when I connect it to the Mac instead of the MPC One.

Now the dropouts are spaced at 5+ minute intervals, but they are the same length, around 21 milliseconds. Here are the time positions of the dropouts in this external-wordclock test

3:23:663
8:38:790 (+315.127 sec)
14:17:523 (+338.733 sec)
19:52:324 (+334.801 sec)

What could cause this? Clearly the MPC One isn't waiting in a locked state for the USB audio clock during the dropout. The Odyssey's oscillator proceeds normally, just the output is silent.

Are you sure its not some effect of the sampled signal? Like perhaps your source has a 5 minute time window of oscillation and after that it cuts out and resets its position?

A digital oscillator is finite. it may take a long long time (depending on resolution) but eventually it will have to either reset the oscillation loop or halt.

How many other devices have you tried sampling for 20 min durations? Try sampling the tv now that you know the positions for the drop outs and see if they exist in your sampled audio from the tv if not than it may be some issue with the oscillator's duration.

It looks like the signal is clearly interrupted, theres no denying that but there are ways to isolate if its the source or the card. If its the card, are you using external power? powering it off usb? (if powering off usb maybe its browning out, the mpc's usb ports are only .5 amps (500 miliamps) not enough current to power hungry things.

- The recorded signal is from MPC's Odyssey synth with a stuck-on note. The Odyssey synth doesn't get to know where its generated audio is going to.
- With other audio interfaces, for example the MPC internal audio or Steinberg UR44, there are no dropouts.
- The interval between the dropouts varies slightly. When the 18i20 has its own internal clock as master clock, it's about 40 seconds, but when it's locked to an external word clock coming from an old Echo Audiofire 8 (which I didn't mention), the interval is over 5 minutes.