By the way - this latency problem occured from the first to the most recent firmware version...

1. First of all, was this with a 5-pin DIN midi cable going directly from the midi out of the Nord to the MODX in?

2. On which MODX part was the sound loaded? Typically on Yamaha devices, the first parts get processed before the last parts, so you might find less latency when triggering part 1 vs part 16.

3. Which audio interface did you use to record both devices?

4. Were you recording audio via USB - or - analog audio into your soundcard? Please mention for both the Nord and MODX.

5. Which MODX sound were you using? Were you using an arp?

6. What is the latency like in the reverse situation whereby the the MODX is triggering the Nord directly via a 5-pin midi cable?


FWIW, a researcher picked 4 different midi modules to measure their midi-to-audio latency:



https://link.springer.com/article/10.3758/s13428-018-1042-7

(MIDI-Tech Pianobox, Roland SD50, Korg Micro X, Kawai CL25)

You can see that the latency can depend on the specific sound used and that the MODX's 5-6 ms is not unusual.

I notice that the sounds listed above with percussive attacks (drums, xylophone) have the lowest latency, and sounds that aren't typically perceived as having fast attacks (reed organ, choir) have the slowest, which is not too surprising, for the real instruments too...

I remember years ago experimenting with a Moog synth to see how slow the attack had to be until I could perceive it, and finding that I could hear anything greater that about 10 msec...it was interesting to me that 10 msec corresponds to a timing mentioned in the above referenced singer.com article: "Sensorimotor synchronization studies with period and phase perturbations have shown that interval changes of 10 ms can influence synchronization" (most of that article is way over my head)...

I'm sure that there are some people who can hear delays of less than ten milliseconds, but for most of us, these are not perceptible...and anyway, we probably subconsciously compensate for MIDI latency in our playing...

df

Hi Michel,

to answer your questions:

1. First of all, was this with a 5-pin DIN midi cable going directly from the midi out of the Nord to the MODX in?
The test from my original post was done using an external keyboard triggering both Nord and MODX, to eliminate the possibility of MIDI output latency from the measurement. To split the MIDI signal, I used an iConnectivity MIO-XL interface. The order of output ports did not make a difference. I tested with Midi over DIN and via USB and got very similar results.

2. On which MODX part was the sound loaded? Typically on Yamaha devices, the first parts get processed before the last parts, so you might find less latency when triggering part 1 vs part 16.
On the first part, the others were empty.

3. Which audio interface did you use to record both devices?
RME Fireface UFX

4. Were you recording audio via USB - or - analog audio into your soundcard? Please mention for both the Nord and MODX.
Analog audio in. The Left Output of the MODX and the Left Output of the NORD were recorded into the Left and Right channel of a stereo pair.

5. Which MODX sound were you using? Were you using an arp?
I tested using the AWM2 init patch and and the FM-X init patch. ARP was off.

6. What is the latency like in the reverse situation whereby the the MODX is triggering the Nord directly via a 5-pin midi cable?
Attached is another screenshot, where the MODX triggers the NORD and the NORD triggers the MODX (see the part names) via a direct DIN MIDI cable with nothing in between. The first both use simple waveforms (FM-X Init Patch), the second pair a piano sound (AWM2 Init Patch). The difference is obvious. To be clear, this is not the reverse situation of the original post, where I used an external keyboard to trigger both.

I'm not sure about the results from the table that you show. A delay of 5ms is noticeable, especially in a setup with multiple synths where only one is delayed like this. Anything beyond 10ms is awkward, unless you play slow pad sounds. After all - a low latency is an important reason why people still use hardware synths on stage instead of computers with plugins. And again - playing the MODX directly is fine, but triggering it via MIDI, which I like to do in my live setup, is not.

If I were to hazard a guess: the MODX is maybe just waiting too long for incoming MIDI bytes before process them in a batch. When looking at the USB descriptors, I noticed that the MODX's max packet size is 512 bytes, where most other devices use 64 bytes. I know that this is not directly the latency, but it seems as if it was made to chew on larger batches.

@Boris - thanks for following up! I'll do some further testing on my end with my Montage just to contribute more data.

5ms is somewhat noticeable with percussive sounds... but practically speaking in a live performance, not a huge deal (speaking as a live performer since '97).

You wrote: "I'm not sure about the results from the table that you show." I think you're underestimating the latency of many synths. Having personally used and repaired hundreds of synths, you'd evidently be surprised.

This is nothing compared to the Montage arp latency that I documented:

https://www.yamahasynth.com/ask-a-question/arp-timing-jitter-latency

I also confirmed this issue.
I have connected MODX both Mac & PC via USB, also connected to PC via RME interface onboard MIDI, Prodsonus Quantum 2626 onboard MIDI.
Roughly the same result.
There is no significant delay for MIDI output.
The issue is MODX MIDI input, which is about 5-10ms slower than my other synths, including MOTIF ES.

I wish Yamaha would fix this issue asap.
Hope this is not a hardware issue like MOTIF rack that had horrible MIDI issue. (See SOS article)

Also I'm wondering if Montage has the same issue.

Also, it's definitely noticeable, even to those that don't know it's possible to be there.

Daughter asked me about it, she could feel it in her little setup. Figured out a way around it, but her setup is super simple.... just computer and MODX.

Took me ages to isolate and find that it was this. It was the last place I looked for the delay.

For anyone driving more than one synth, this would probably be an unsolvable nuisance.

Right. I started noticing it as soon as I connected MODX6 to MO8 playing piano.
At first I thought it was because of key feeling, but it was so annoying that I measured the latency.
Many people here would notice it just with a MIDI cable connecting in and out directly, local off.

MIDI Jitter seems to be average.

This sounds like it could be very unfortunate for someone who buys a MODX6 or MODX7 and pairs it with some other weighted action keyboard underneath from which they want to play the MODX' pianos. Or any time someone connects a second keyboard to play instruments that are located in Parts 9 through 16. I'd be curious to see if many people are noticing this.