Maybe try key mode ad Thru then hit the root first and follow by next notes for inversion. Thru plays notes of your held chord in the order you press keys. "Sort", as is the default, plays from lowest note up.

Since your original ARP is only a triad and the inversion must be interperated as a 6th chord, the chord intelligence must boil down what it thinks as G6 to GMaj.

Hi Jason, thanks for your response!

Since your original ARP is only a triad and the inversion must be interperated as a 6th chord, the chord intelligence must boil down what it thinks as G6 to GMaj.

I cannot see why voicing G B E must be interpreted as a G6 chord, or why Yamaha would choose to implement the chord intelligence like that. If I included the D as in G B D E then it would make sense to go for G6.

But you are right, it does seem like Montage interprets G B E as G major added sixth instead of an inverted Eminor triad.

I read in this article https://www.yamahasynth.com/montage-category/montage-smooth-it-over-a-motif-xf-conversion that Montage should be able to interpret all inversions of major and minor triads:

You do not always have to have the root note as the lowest note.
If you play a C major triad, in any inversion, the 'bass player' (arp) will recognise you want the bass line rooted in C.
If you play a C minor triad, in any inversion, it recognises and adjusts the bass phrase

That would suggest that it should be able to recognise G B E as an inverted Eminor triad?

I didn't see this article before posting - but quoting the article:

Regarding recognizing inversions ...

For example, Dminor7 (D-F-A-C) can easily be recognized in any inversion, except when the flat 3rd (F) is in the bass. When the flat 3rd is in the bass (F - A - C - D) the technology has to see this as an F6 chord.

The above details 7th chords - but I assume the same general rule applies to triads as well. A minor quality chord with the minor 3rd in the bass gets interpreted as a major sixth chord.

G B E is ambiguous and could be Emaj 1st inversion or G6 root position. The only way to get a handle on the chord spelled is context within the progression. This is not something the chord recognition does. It looks at what you're playing right now without any context of past or future chords. If you're given 3 notes G B E and have no context (as the chord recognition would not have) then either answer G6 or E- would be correct labels. The 5th is an optional chord tone for a 6th chord. It would have been better, for your use, if the chord decoder would favor simple maj/min forms over the more "exotic" 6th. But that does not seem to be the policy here - so you have to do something to make the system work for you (adjust).

I have investigated the MOXF "XS-style" arp chord recognition at quite some length just by experimenting and what the above reply says is correct, a minor chord in 1st inversion is recognised as a 6th chord on the lowest note played instead. It's the only triad inversion that does this - all the others work as you would expect

In fact, I've done quite a few investigations into how the chord recognition features work in the arps, right up to and including 13th chords. I can't say I've sussed out the algorithm 100% but the results of my investigations are below:

1. Major, minor, dimished and augmented triads are recognised as such in root position.

2. Major and minor triads can be inverted and are recognised correctly /except/ for the 1st inversion of a minor triad that is recognised as a 6th chord with the root as the lowest note played (so the third of the intended minor chords)

3. Dominant 7th chords are recognised correctly if all notes are present in root position and in all inversions.

4. Minor 7th chords are recognised correctly in root position, and 2nd and 3rd inversions. In 1st inversion they’re recognised as 6th chords on the lowest note (i.e., the minor 3rd of the 7th chord).

5. “Straight” (i.e., not altered) 9th chords are recognised correctly in root position if all notes are played; they’re also recognised if the root and/or fifth are omitted - e.g., E, Bb, and D is recognised as a C9.

6. Flattened or sharpened 9th chords need the fifth and root to be played to be recognised correctly.

7. 11th and 13th chords need the root to be played.

8. Recognition of 9th, 11th, and 13th chords in inversion - too difficult to work out (my ear isn’t that fantastic) - and is a bit hit-and-miss. Mostly miss. Plus, of course, these chords can often be interpreted many ways anyway.

I’d love to see a flowchart that shows how the MOXF works it out! It’s all very well saying the arps are “chord intelligent” but if you don’t know the rules on chord voicing it’s very difficult to second-guess what the recognition tool is going to do. It matters quite a lot for basslines!

Of course, manna from heaven would just be if Yamaha gave us details of their "intelligent chord recognition algorithm" that is so publicised. It's all very well it being an "intelligent chord recognition algorithm" if your rank-and-file educated musician has such a hard time working it out.

I hope that the above is of some help anyway.

David.

G B E is ambiguous and could be Emaj 1st inversion or G6 root position. The only way to get a handle on the chord spelled is context within the progression. This is not something the chord recognition does. It looks at what you're playing right now without any context of past or future chords. If you're given 3 notes G B E and have no context (as the chord recognition would not have) then either answer G6 or E- would be correct labels. The 5th is an optional chord tone for a 6th chord. It would have been better, for your use, if the chord decoder would favor simple maj/min forms over the more "exotic" 6th. But that does not seem to be the policy here - so you have to do something to make the system work for you (adjust).

I see. Thanks for the clarification 🙂

I guess I canjust play the chords in root position.

Hi David,

I’d love to see a flowchart that shows how the MOXF works it out! It’s all very well saying the arps are “chord intelligent” but if you don’t know the rules on chord voicing it’s very difficult to second-guess what the recognition tool is going to do. It matters quite a lot for basslines!

Of course, manna from heaven would just be if Yamaha gave us details of their "intelligent chord recognition algorithm" that is so publicised. It's all very well it being an "intelligent chord recognition algorithm" if your rank-and-file educated musician has such a hard time working it out.

I hope that the above is of some help anyway.

That is some really good information you have collected. Thanks!

I agree, it is really strange that Yamaha have not detailed how the chord recognition works. Would save me a ton of time.

It would also be nice if there was some method for tuning the chord detection in the Montage. For instance, to detect a first inversion triad rather than a 6 chord.

OT: There are some choices that Yamaha has made when deciding on the features of the Montage that I do not understand. For instance, you cannot send a midi file to the device via USB cable. You have to put it on a USB stick and move it physically to the Montage. Why not just add the feature in the Montage connect utility? So strange.

Hi!

I just did some more testing. I have issues with the second inversion of a major triad being detected as a wrong (6?) chord for my own user arp. Sigh.

Can one of you reproduce this?

Record four tones C3 E3 G3 E3
Convert to user arp "Orig Notes" Root note C3
Voicing C3 E3 G3 results in correct playback C3 E3 G3 E3
Voicing E3 G3 C4 results in correct playback E3 G3 C4 G3
Voicing G3 C4 E4 results in wrong playback G3 C4 D4 C4

😮 😮

Thanks in advance 🙂

Hi Andreas

Oh yes, I should have added: the investigations I did were all with the built-in "XS" style arps.

When I performed the same investigations with my own user arps I found that there was absolutely no way to get the user arps to perform the same chord "intelligence" rules as were found in the "XS" style arps, and believe me I spent hours and hours trying every possible setting.

So the behaviour you describe is, unfortunately I believe, "normal" behaviour for user arps. For user arps I think you just have to rely on root-position triads or seventh chords and just hope for the best when playing them (and, in general, endeavour to play them all in root position which of course as trained musicians we are taught NOT to do!). And then after recording dive into the sequencer and change the "wrong" arp notes manually. Which of course you could do when recording to the MOTIF/MOXF sequencer with ease, but for live performance on the Montage you well … errr …. can't do it cos it's designed as a machine for live performance obviously. Unless you're recording to a DAW - I don't really know if you're a performer or a studio musician.

I'm a bit of both and in general wouldn't trust a user arp in a live performance situation for all the reasons we've discussed already.

That flowchart showing the algorithms used for arp playback would be so darn useful, hey?

David.

Hi David,

I see. These are for live performances. We try to cover some tracks by Muse. They have a lot ofl analog mechanic-sounding arpeggiated synth lines. The problem is not so much remembering to play the root position chord instead of an inversion, although it feels strange. The problem it is that if I have an arp parttern that goes, lets for simplicity say C3 E3 G3, then I want to be able to voice E3 G3 C4 and get an arp phrase that goes E3 G3 C4. If I voice the chord in root position, I get a different result than I want, one that does not sound as the music I try to cover.

I think Matthew Bellamy, the main composer of Muse, uses some kind of ancient analog synthesizer (Juno-60 or something like that) to get these synth lines. I'm really amazed that my 2018 flagship Yamaha synth can't reproduce these the arp patterns.

I actually started out with just a laptop, an audio interface and a MIDI controller on stage. I arranged my live set in an Ableton Live session, and it worked OK. Making an arp that worked the way I wanted was totally doable. But the laptop was sort of slow to start up and sometimes the audio interface would not be detected. So I purchased a Montage to get a solution that would be self contained and ready to go instantly when powered on. But now it looks like I'm gonna need the laptop anyway if I want to get the arp to play back like I want it. Sigh.

Thanks for your input, it has helped me a lot!

Andreas

Hi Andreas

Glad I could be of help but sorry that we couldn't find a solution! Agree with you about not wanting to take a laptop onto stage!

Do you know of some (free) software called "one man band" that works with Yamaha style files (like found on arranger keyboards). There, the arps work absolutely perfectly. And it's freeware too. Plus you can reverse engineer how the chord recognition algorithm works because it's written in Visual Basic and they give you the source code. I'd love to see the source code of the Yamaha chord recognition and resulting arp generation!

Anyway, good luck with the performances and, you never know, someone from Yamaha might respond with a solution to your problem.

Good luck with the performing with the laptop on stage - but I agree, it's such a shame after buying a flagship synthesizer.

David.