The MIDI Forum

Hmm???

I've just downloaded the manual, and had a quick look through it.

Basically, it mentions 'midi', but I don't think it's 'midi' as we know and love it??

As you imply (but omit to say explicitly) there is no midi information in the manual, no midi implementation chart, no discussion of using sysex etc, etc.

It looks to me (from my QUICK look) that this unit does NOT receive midi data in the usual way, and that therefore you cannot use midi to control anything. 'Midi' seems to be referred to ONLY regarding certain links with certain other devices, i.e. for Garage Bond, or with a mobile phone.

Furthermore, this keyboard does NOT seem to have any midi connectors, only bluetooth (although that MIGHT involve some midi facility, but seems not to).

The older devices you mention were designed/constructed to do things with midi, i.e. to be fully/totally controllable via midi from other devices. The Go:Keys seems to be designed for a quite different market, maybe as an add-on for Garage Band?

So - sorry - but this keyboard may NOT do what you are hoping.

Geoff

The USB is also a MIDI connection and is recognized in Jack control and can be used with Qsynth / FluidSynth.
Also Rosegarden sequencer plays Midi files and in STUDIO setup it is found as a valid output device.

Roland calls it compliant so no drivers are needed.
The extension controls the LOOP MIX function as expected.

The source code is vague as to where it derives the values it uses.

Roland GO:KEYS ScratchX Extension has this stuff at bottom of screen that points to itself and this domain documentation for WEB-MIDI.
Roland is not listed (yet) in the list of manufacturers.


"GO:KEYS ScratchX Extension (source code) uses a web technology called Web MIDI API. For details, click here."

There are extensive references to MIDI so I would expect a table what is and is not recognized.

Geoff,
Congratulations on being the top poster of all time having only registered July 2017.


Geoff
Registered on Saturday, 29 July 2017
2 Posts 103 Replies

Your response shows me there is a real need for a quality based MIDI community.
I should have done my research but felt the association name and work to date was worthy
of asking this question and expecting an answer from Roland.

I will be persistent and help those few that are here that "Basically, it mentions 'midi', but I don't think it's 'midi' as we know and love it??
"

Not the case !
MIDI is to music what ASCII is to the NSA database !

Found this blog entry that may have been posted for fun last year !
DOES MIDI HAVE AN IMAGE PROBLEM?

Aha...

Well, if RoseGarden will play midi files via the device, then I'd expect that other software should be able to do the same, so therefore the device DOES have at least some of the normal midi facilities.

Your note about 'compliant' and drivers is odd. Usually, the drivers may be needed to allow a particular machine to SEND midi information, because the hardware of the machine is nor normal, i.e. regarding ports used, IRQs, memory addresses, etc. Or it's needed to allow software to work with different midi cards, or select between, say, MPU-401 or USB or other options. This would not normally have any relevance to the connected device which is going to receive the midi data. Usually, a specific piece of software may need a driver, but with the driver, it can connect to just about anything. That's certainly the case with my various midi systems.

There are usually two aspects to the midi system information provided for many devices.

The first comprises the instructions for all the midi commands supported, and most of this is fairly normal and generally applicable. Some things can vary, i.e. the extra bank selecting commands.

The second aspect relates to a sort of 'memory map' of the device, and how you might use SYSEX commands to access that memory directly to do a range of specialised things, incl potentially construct/change sounds directly in memory. This latter will be dependant on how the device is constructed (in firmware) and this might not be possible - I've got devices where you can change nearly everything, and others where you can change nothing in this way.

Yes, any midi device OUGHT to have a 'Midi Implementation Chart', detailing what can be sent and received, channels, modes, controllers implemented, etc. This is a standard format. I think it's part of the midi specification that it OUGHT to be provided. EVERY device that I've got has one, even if it's limited because that's ALL the device will do midi-wise! Oh, I tell a lie! I have a Quadraverb unit, which DOES allow some degree of midi control, and this does NOT seem to have a MIC - however the midi facilities are VERY limited compared to most other devices that DO have a MIC!!

Geoff

Lee,

Thanks for the comment.

There are a number of things that I'd love to persue, but we should stay 'on topic' so here is not really the place.

Regarding - briefly - the image of midi, I'd say that if there is a problem, it's because people just don't understand midi. I've always found midi quite wonderful, partly because I fully realise that I'm using only a few percent of what CAN be done with it, if only I had more time/expertise/musical abilty.

Yes, there are a LOT of really CRAP midi files, but there are the occasional GREAT ones as well, and I keep the latter in mind as to what COULD be done with midi. Anyway, I KNOW that if a midi file is no good, then I merely have to FIX it!

Geoff

Lee,

Also OT, totally, but Sunday night I'll be watching Superbowl! Been interested in the game since 197?, and interested in the Pats since Steve Grogan was QB, but I'd be watching anyway regardless of who's playing, just even more so as the Pats are there - AGAIN!!

geoff

Geoff,
Watched last moments of Super Bowl.
Enough said.

As far as MIDI, I agree and feel after learning it in the early days and building 25 pin serial to 5 pin DIN Z80 based box then porting "Ghost Buster" to play on Juno-106 using DEC Vax 11/780 mainframe I feel I understand MIDI more than many.

I have a Yamaha PSR-160 and couple Kurweil K1000 devices so I was interested in a USB General MIDI 1.0 to make fair sounding presentations.
I now see and downloaded GM 2.0 and WebMidi so pleased with these discoveries.

Now want to find what NEW tech Go:Keys can do.
The code on Github reveals (SCRATCH experimental) the MIDI to "LOOP MIX" commands.
I also got the Scratch 3.0 HTML5 rewrite running offline as fas as it is developed.
MUSIC extension uses MIDI Note values to play MP3 ASSETS that are modified for tones.

Attached is screen showing conflicting NOTE blocks.

Some discussions talk about Scratch 1.4 compatibility which was last time MIDI was part of code.

Got to go so find me on GITHUB or SCRATCH forums and we can delve into off topic pursuits.

Lee T. Davy
Cool Games Network

Since it has been so quiet and cold I decided to try to extract the implementation manually.
I contact Roland USA and Japan and got some responses but the Japan R&D Go Keys
team never participated after initially responding.
I am now trying to find blank implementation chart in Libre Office compatible form.
I am using Linux Ubuntu Studio 17.10 with qjackctl connecting qsynth with internal audio
GAIN 0 and message screen displaying output in MIDI codes.

I attached the screen capture for displaying MIDI codes extracted.

Qsynth CHANNELS shows activity during LOOP MIX that shows parts sending NOTE ON
OFF and MIDI CLOCK.
Also attached QSYNTH LOG file is small with just BANK select then LOOP MIX to log
activity when buttons pressed.
No keys were pressed or program changes selected.

I will check back later as I progress with this implementation chart.
also saw a new EXPERIMENTAL ScatchX Extension
described as

add note on/off block for test

together with sample SBX that animates on the beat.

I uploaded the activity with nothing other than ScratchX Sample in the log.

I decided to record using the Go Keys while it played the ScratchX Sample_fish.sbx and then upload as a backup the song using procedure as described in user manual.

Attached is the directory including the Linux directory

Backup Sample Fish

.

GO-61

is the top level that appears when the Go Keys is put in backup mode and
BACKUP is mounted as a USB DEVICE.

Inside is a playable MIDI FILE SONG01.MID that contains the stored MIDI that can be played by the Go Keys or other MIDI device or GM software device.

It sounds terrible using QSYNTH as expected and experienced if GAIN is not set to 0 during use of Go KEYS.

Hi,

The file you attached to the previous post with the name "GoKeys_backup SAMPLE.zip", after downloading it appears not as a ZIP file, but a TARball:

$ file GoKeys_backup\ SAMPLE.zip 

GoKeys_backup SAMPLE.zip: POSIX tar archive (GNU)

The two capture samples (qsynth logs) of the other post, also named as ".zip" are also TAR files.

Anyway, I've played the file "Backup Sample Fish/GO-61/BACKUP/SONG01.MID" using QSynth+"General User GS v1.471.sf2" (http://schristiancollins.com/generaluser.php) and it sounds quite normal to me. Somewhat cacophonic, but nothing out of the ordinary.

As you are obviously using Linux, maybe it would be useful for your investigation my program "kmidimon" which is available in most Linux distributions. It is a MIDI monitor in real time, with event filters, MIDI recording and playback. It also helps analyzing MIDI files and also other file formats.

Regards,
Pedro

Following up Pedro's comments above, I've now downloaded the midi file - I used JZIP to unpack the file (worked fine regardless of 'zip' or 'tar' type file). I also noted the DUMP file of the .mid file, which is a very similar format to process I have (DECODE/ENCODE).

I tried playing the file on my XP machine, using the VanBasco player and VirtualMIDISynth with the 'Timbres of Heaven' SF file. To my ear, it was somewhat better than a 'cacophony', although it seemed not quite right, I suspect that at least one of the patches being used was not quite right. But mostly fairly close. The setup I was using does support GS to some extent. Overall, reasonably OK, just needs a little adjustment to get it right, like I suggested, maybe just one or two of the patches slightly out?

Looking at the details of the file as detailed in the DUMP file, I note there are quite a few instances of an undefined controller being used. Not sure if this controller IS defined within GS2? But I'd lean towards this NOT being the problem Lee's been having - any wrong controller would be disregarded anyway?

Overall, I'd say there is nothing wrong with the data in the midi file, and the only question may be that the patches (Program Change) commands will be very specific to the receiving machine. Maybe Lee's machine is less GS2 compliant than the setups that Pedro and I have been using?

Anyway, the file IS GM2, and NOT GM. All the patches are selected via Bank Select followed by PC.

Lee - can you be more specific about what is WRONG with the file as you hear it playing?

Geoff

Since the previous posting, I've studied the midi file somewhat, and is seems somewhat odd. Probably, little of this has any bearing on how it sounds, mind you.

I checked all the Program Change instructions, and determined that the file contains Bank Select instructions, followed by PC instruction. The PC instruction alone, however, seems to use the 'correct' patch number for GM (as opposed to GM2) so I believe that if there are NO banks to select, then the file may play properly. If ALL the banks can be selected, then it may work also. If SOME ONLY of the Banks can be activated, then things will go funny.

I've now edited my copy of the .mid file to remove all the Bank Select instructions, and it does sound better.

I've also removed all the NRPN instructions - I've no idea what they are doing, but they will be totally specific to a specific machine (the GO-61 ?) and they certainly don't do anything with any device I've got. I don't think they would cause me any problem, but I cannot be sure what they might do with a device that recognises them?

It seems odd that the file includes some set-up for Channels 12 to 15, including PC instructions, but zero notes? Wonder what happened there?

One of the reasons the file seems odd when playing is that Channel 16 (Elec Piano) is just too loud (compared to the other channels). So I've altered that to make it less quiet, and this makes the file play a little better. This may be accidental, one of the known problems with playing midi files can be that the relative volume of the same patch on different devices can be different, enough to mess up a balance that sounded 'spot-on' when the file was created on the original set-up.

It'also seems odd that there are quite a few NRPN commands for Channel 16 long before there are ANY notes for Channel 16.

The file may well be a 'work-in-progress' that went awry, so apologies for my comments if that's the case

Geoff

Yesterday I got a message from Roland Japan stating that there is not any active development going on for the Go Keys so I am now deciding on an approach to deciphering the implementation.
"Our engineer who is expert in Scratch X, he is very busy with other project.
Please wait.

At this moment we don’t have the MIDI implementation chart for you."

Scratch X is still not changing but evolving to Scratch 3 using HTML5 instead of Flash for Scratch X.
If there is any interested here, it is in development at https://github.com/LLK
MIDI is eluded to in the specs but I have not seen how it will be jacked.


I am looking for a MIDI Implementation Chart template for Libre Office.Then I will try to fill in the blanks.

Today, I will try to use kmidimon instead of muted qsynth.

I forgot how I created the uploaded files so I will resurrect that effort and explain the approach.
The bank select is part of the implementation in that each sound bank button select sends a separate MIDI channel allowing instruments to be played in parallel rather than all on channel 1.

Since I only have the Juno 106 and Kurzweil K1000 PX to compare I really am also learning about conforming to MIDI standards and if Roland follows the rules.

I suspect the Go Keys is a sound canvas on steroids if I understand the implementation and market expectations.

I am also learning to navigate this forum and it is difficult to follow threads so will be trying to subscribe or bookmark the topic in conversation.

Hello Lee,

Sorry to hear you're having such a problem getting the Midi Implementation Chart. As I think I noted before, I THOUGHT that most midi devices would come with one. So how come Roland cannot just send you a copy of one for the specific keyboard, I don't understand.

I do NOT follow, at all, one of your comments above. Maybe it's something peculiar to the specific keyboard, or to something you were doing, but bank select should NOT have anything to do with channel select, and nothing to do with 'allowing instruments to be played in parallel'.

Of course, Bank select (like any other CC), is specific to a channel, but it's a channel YOU define, not 'automatic'.

You could easily use multiple bank select (and program change) in a single channel, although it would be a total pain in the ass and no sense at all in most cases. No problem if a specific channel starts as one instrument, then changes to something else later. Setting a specific channel to a specific instrument is just a massive convenience, otherwise you'd end up with as many bank select/program change commands as note commands in your piece. Is it just me, but I hate it when some system starts doing something automatically that ought NOT to be automatic, that I should have control over (like how to use channels).

Geoff

I have only owned the Go keys a few days...The Go Keys responds via my DAW using LSB 0 and MSB 0, 1, 2,,3. (these are the Go Key sounds) GM2 are LSB 121 and MSB 1,2,3.
I send this info via my Code 61 controller and the sound patch shows in the Go Keys screen.

I believe the loop record uses the channels 12-15.. When I get a chance I will look into this..

Thanks - some clarification there.

When I was looking at Lee's midi file - mentioned above - I wondered how come the data was in ch 12-15. Looks like Lee was using the Loop Record facility, and that's what that does!

As for the instrument select, yes, I could see in the midi file that all the sounds seemed to be being selected using LSB/MSB (which is usually JUST for the Bank Select), then followed by a PC (Program Change) standard controller to select the actual instrument. Are you saying that the instruments can be selected via LSB/MSB ONLY? The data in the midi file certainly seemed a little odd??

Geoff

I received the preliminary MIDI chart but can't share publically.
The implementation is using a lot of NPRN .

It follows tightly how the ScratchX experimental code was being defined.
There is a MIDI Clock when ever there is a loop mix and all 16 channels are concurrently active with low 7 banks each a button on control panel plus drums on 10.
The other 8 channels 8-16 are loop mix with NPRN selecting the phrase 0-10 and part not NOTE ON as I expected.
Channels are similar to live analogy on 5 octave keyboard split by Bass, Part (Melody A) , Part (Melody B) , Drums.
The values can be seen at the GITHUB Extension

I am not sure what PartX1-4 is in the chart for channels 12 to 15.

I have been using receivemidi and sendmidi from Linux command line.
I will experiment more !


Maybe a way to get 8 parts instead of 4 plus drums plus 4 recorded for sequencing.

If you watch the videos and read the code you can see how chart was derived.
I plan to try to follow the Roland GO:KEYS ScratchX Extension and Youtube Tutorials.

One playlist is for SCRATCHX.

Another is Quick Start for learning to use the GO:KEYS

NPRN also selects Loop Mix KEY as 1 of 12 as done for real after pressing chord button.

A second version of the extension Extension2.js
shows additional menu choices for more of the instruments which are listed in the MANUALS.

Data reception and transmission is Note ON / OFF , Bank Select, Modulation , Program Change, Pitch Bend.
Also included are volume, panpot, expression, resonance, cutoff, decay, attack and release time

Transmission also has Real Time messages for clock, continue, and stop.

Hello, Lee.

As for the implementation, in the published "specs" in the official website, you can see "conforms to GM2" in the part of MIDI data format.

That means that you can actually re-create a midi implementation chart only from the published requisites for GM2 and include the ones for GM, so I think you didn't have to go through all that trouble to get the information you wanted.

If you want to write an editor for the unit all you need to do is write a standard editor for GM2. It would include all the recognized parameters like bank select, volume and pan, but also the ones for vibrato speed, rev send level and so on.

All RPN messages are defined.

There are Melody channels and rhythm channels defined in the specs. You can set any channel as melody channel by sending bank select messages 79H/xxH, where xx would be the actual bank number.
Similarly, channels 10 and 11 can be set as rhythm channels if you send the bank select message 78H/xxH.

If you say that the unit presents 4 rhythm channels, this means that it implements an optional part of the GM2 spec that allows any channel to be defined as rhythm channel by sending the bank select messages 78H/xxH through it.

About the notes, the unit must have true voice, i.e. should be able to play some sound when any note number arrives. (notes 0-127)

Note off messages are ignored on rhythm channels.

And there's more in there, but I assume you already did your homework and read it.

What you mention about NRPN messages being used to select phrases is not included in the GM2 spec, though.

Could you share the NRPN numbers for phrase selection and LOOP mix key, please?

Also, I don't really know what you mean by ScratchX. Isn't it a website?

Thanks.

Hugo,

Regarding Scratch, I suggest that you look at the first two posts in this thread from Lee, where he says something about Scratch/ScratchX, and provides some links to further information about it.

From that, I understand that it's a programming system, allowing access to the Web Midi API.

Geoff

ScratchX is an MIT Life Long Kindergarten (LLK) project to make programming tools to teach concepts to young students.
It is considered LEGO like programming and has an interest in LEGO hardware as well.

The underlying part is BLOCKLY from Google and ScratchX is an experimental version already in maintenance mode
while Scratch version 3 is developed to eliminate the Adobe Flash requirement.

It uses WebMIDI and will be easy to use.
Not all functionality will be available in Scratch 3.0 so regular MIDI tools will still work on the NPRN written
into the firmware.

Since I am not a programmer it was difficult to uncover the actual MIDI code in the ScratchX extension.
As a non-programmer I find the charts alow insight into what the device can do.
I now can use it more intuitively as documented in manuals and videos.

I like editing and changing things that I dslike when I hear the playback.
The builtin function requires a level of accuracy I don't achive in real time.

Alternatively the Go:Keys is able to be used in live situation but no where like a workstation.

I hope performers share their compositions by way of uploading the GO-61 backups to a sharing space.
No filenames are editable so all will be SONG01.MID etc
I tried putting filenames in the RESTORE GO-61 and although it worked it was prone to crashing the synth.
I also tried using Rosegarden to make a recording using the Go:Keys feature and it reassigned the sounds when played back.

I don't see the GM2 requisites as a valid argument for not having the MIDI Implementation Chart V2.0.
I also don't see the YouTube videos as a replacement for written instructions for how to create a live and layered sequence.

Eventually I see this product influencing young and old as well as home and professional musicians.
Again, it replaced a 20 year old Yamaha PSR-160 which had General Midi with batteries and speakers for portability.
I couldn't download songs to that one either to play along.
Maybe I will get that to work on the Go:Keys.
For now it will pair as speakers to stream music from the recorded songs on a phone or other Bluetooth device.

And now a word from the manufacturer.
https://www.roland.com/us/products/gokeys_go-61k/
"GO:KEYS is a fun, easy-to-use keyboard that lets you play along with your favorite music—streamed wirelessly through its powerful onboard speakers. Just connect your smartphone via Bluetooth®, pick a song, and play along!"

It is also shown in the ads using the preset patterns and sequences.
This is where the sharing of GO-61 BACKUP could begin.

(Hope this sheds some light on my reason for wasting my time getting the chart.)

I still really like this keyboard at 66 years old.
The kids can use the Yamaha for now and it's built in library of tunes.

Lee