Manikin Schrittmacher MIDI Timing tests

[BaconTastesGood]

Since I'm now officially obsessive...

My Schrittmacher arrived today and I decided to run the same MIDI timing tests I did on my RS7K. I connected my MicroQ to MIDI OUT 1 and my Nord Lead 2 to MIDI OUT 2 and recorded (each independently, not both at once). This was, in hindsight, a bad idea since I can't factor out if there's a timing difference between the ports (I doubt it), but anyway, science and all that.

The results are surprising to me.

First, the test: 16th notes, 120BPM, only a single sequencer line, capturing a total of 64 notes (4 measures) into Reaper @ 48Khz. I then measure the sample offset note-to-note and also how far it has strayed from the fixed grid.

Looking at the graph, the timing with the MicroQ strays a lot occasionally.

The NL2 is extremely solid, I think the worst note-to-note error I saw was 43 samples (< 1ms). The MicroQ on the other hand spiked all the way up to 321 samples of error (6.68ms!)

This could be just a case of bad luck with the data, or the MicroQ may actually just be worse at handling incoming MIDI data -- which would be somewhat sad given that it's a rack unit.

Interesting factoid: Schrittmacher => NL2 has tighter timing than the RS7K's internal sample playback engine.

[cgren72]

this is interesting, but is the midi timing of the microQ ever actually noticeable when listening?

[BaconTastesGood]

this is interesting, but is the midi timing of the microQ ever actually noticeable when listening?

Yes, a million times yes. =( It's bad enough that I may end up selling it.

The 'test' for that is that sometimes I'll leave a sequence running and go off and do something else (I'm too lazy to turn it off, etc.) and then it'll have a MAJOR hiccup while playing the sequence -- I'm talking extremely audible to anyone, at least 20+ms if I had to guess -- where it'll push/pull rapidly, i.e. suddenly it lags then snaps ahead to compensate.

With the NL2, this NEVER happens. It's a solid sequence the whole time. Same with the Blofeld.

But the MicroQ definitely has issues responding to MIDI at that rate.

I'm probably going to code up something that will automatically process an audio file and figure out the amount of error automatically, that way it's trivial for me to capture one minute of audio and then analyze the data instead of doing it manually. Expect a lot more charts if I do

[cgren72]

I like charts, so this works out well. And like you said earlier, it is kind of weird that the midi timing on he microQ would be bad when it is rackmount.
And what will you code up something with? What program will you use for this?

[BaconTastesGood]

I'm a programmer, so I'll write something for scratch, shouldn't take long. I'm just going to load up mono WAVs, normalize them, then use specifiable window widths and threshold values to determine trigger levels and dump the output to a CSV, which I can then import to a spreadsheet.

[cgren72]

I'm a programmer

That explains it lol.

[cgren72]

The only programming I do is on Arduino, so im pretty much a "lightweight"

[nuromantix]

No disrespect but you really ought to consider using it to make some music instead!

[BaconTastesGood]

No disrespect but you really ought to consider using it to make some music instead!

I do that too

But at the same time, there's nothing more frustrating than trying to make music with uncooperative tools. If I had not known about the MIDI timing issues I would likely have been extremely frustrated when the MicroQ just decides to wander off on its own...

[cgren72]

No disrespect but you really ought to consider using it to make some music instead!

I do that too

But at the same time, there's nothing more frustrating than trying to make music with uncooperative tools. If I had not known about the MIDI timing issues I would likely have been extremely frustrated when the MicroQ just decides to wander off on its own...

I don't see anything wrong with knowing what is going on with a piece of gear that is being used to write music. If it saves frustration, It may certainly be worth it. Frustration with equipment is a huge music writing turn off. Most of my troubles are with sequencing, and I could see how it could be the same way for alot of people, so knowing which pieces can be sequenced the best is good knowledge.

[BaconTastesGood]

I don't see anything wrong with knowing what is going on with a piece of gear that is being used to write music. If it saves frustration, It may certainly be worth it. Frustration with equipment is a huge music writing turn off. Most of my troubles are with sequencing, and I could see how it could be the same way for alot of people, so knowing which pieces can be sequenced the best is good knowledge.

The reason I wrote this tool is because it took less than two hours and I can feed it an arbitrary WAV file and it will generate all the note deltas. This is a HUGE benefit for me since now when I think some gear is janky I can just record some audio for a minute, feed it in, and then immediately see if I'm imagining a problem or if it's there.

And I agree, sequencing is easily the #1 most aggravating thing, especially when dealing with a primarily MIDI based studio, so knowing which bits of kits are reliable and which bits aren't for timing will save me a lot of aggravation.

I'll be honest, I'm now kind of looking at the MicroQ somewhat askance for drums or bass lines...

[BaconTastesGood]

Won't bother posting a graph, but I ran the tests (now automated, woo!) on my Blofeld and on the MicroQ on port 2 (in case the port mattered).

Blofeld was as tight as the NL2. MicroQ was still pretty bad. I extended samples to 60 seconds and was finding swings of over 5ms (either direction, sometimes early, sometimes late, sometimes back to back), whereas the NL2 and Blofeld were tight < 1ms note to note error.

The grid error (which is a factor of the Schrittmacher) was constantly increasing and early, indicating clock drift between Reaper and the Schrittmacher, but that's to be expected (and tempo sync at each measure would make it irrelevant).

[inaheartbeat]

Interesting results. I wonder if the Q has the same issues as the micro Q. I know my Blofeld feels very tight on arpeggiated sequences so it doesn't surprise me to read your results. I should probably run similar tests on my Q and my Virus TI2. I have a feeling the Virus is not going to come out well. It always feels like the arpeggiator is "off" a bit on that.

[ninja6485]

Have you ruled out the possibility that your microQ is functioning improperly?

[BaconTastesGood]

Interesting results. I wonder if the Q has the same issues as the micro Q.

I'm guessing they share the same code so I would consider it likely, but I have no way of telling. The Blofeld is rock solid though.

I know my Blofeld feels very tight on arpeggiated sequences so it doesn't surprise me to read your results. I should probably run similar tests on my Q and my Virus TI2. I have a feeling the Virus is not going to come out well. It always feels like the arpeggiator is "off" a bit on that.

Internal sound generation (step sequencer/arps) should be significantly tighter than receiving MIDI, but I would be curious how it turns out.