EMI Timing Issues and MIDI Instability

[jimmyklane]

Hey everyone.

I’ve switched over to Reason from Nuendo. I’ve always had Reason, but had been using it as a ReWired sound source and driving my hardware from Nuendo.

Now that I’m Reason 10.1 as my only DAW, I’m finding that my synths are responding much differently than I’m used to. I have a very solid understanding of MIDI, and the electronics involved and know that each instrument and even each note will have slight inconsistencies. The problem is that Reason is putting my recorded tracks so far off the beat that I’m having to chop them up and manually move sections, and if it’s a continuous part then I’ll need to do several takes and comp them together which is ridiculous.

Has ANYBODY experienced this behavior? It’s worse with MIDI over USB, which never used to be a problem for me. I’ve since hardwired everything as well as kept the USB connections.
I’ve got a dedicated port for just about every instrument except for a few two-instrument thru chains which don’t give me problems.

My computer is completely optimized for audio and the ASIO drivers are up to date. The MIDI device drivers are also up to date. On Win7.

[madmacman]

Indeed, the EMI Midi Timing is utter crap. I usually use(d) it with my Modular (Midi-2-CV) interface, and the results weren't satisfying at all. Remedy: Since I usually recorded straight 16th mono patterns, I corrected the resulting clip with Audio Quantize, which solved the problem for me. But still...

[selig]

We talking latency, jitter, or both?

[madmacman]

We talking latency, jitter, or both?

To my understanding (constant) latency wouldn't be much of an issue. You could simply cut the audio clip at the very first transient and then the entire recording shifts to the correct position. So no, it must be a jitter issue with single notes playing odd.

[jimmyklane]

We talking latency, jitter, or both?

It’s jitter. Sometimes the first transient is actually BEFORE the first MIDI note in a clip!

I cannot figure out why Reason is performing so poorly, and Audio Quantize absolutely wrecks carefully crafted sounds.

[Ahornberg]

I think jitter is caused by the OS. On windows jitter is measureable. I cant speak for Mac.

[selig]

We talking latency, jitter, or both?

It’s jitter. Sometimes the first transient is actually BEFORE the first MIDI note in a clip!

I cannot figure out why Reason is performing so poorly, and Audio Quantize absolutely wrecks carefully crafted sounds.

Ouch, that sounds bad…and if I'm understanding you correctly, and this works fine in other apps but not Reason, there is good news and bad news.

The bad news is that it's Reason's fault, and they are slow to address this sort of thing, historically speaking. The good news is that it's Reason's fault, and it CAN be fixed if/when the powers that be decide to address it.

How bad is the jitter, and is it tempo dependent (if so, faster tempos can mitigate it, maybe?).

Also, this would suggest that it's entirely possible that MIDI coming INTO reason is just as susceptible to this jitter? Have you tested this per chance?

[jimmyklane]

It’s jitter. Sometimes the first transient is actually BEFORE the first MIDI note in a clip!

I cannot figure out why Reason is performing so poorly, and Audio Quantize absolutely wrecks carefully crafted sounds.

Ouch, that sounds bad…and if I'm understanding you correctly, and this works fine in other apps but not Reason, there is good news and bad news.

The bad news is that it's Reason's fault, and they are slow to address this sort of thing, historically speaking. The good news is that it's Reason's fault, and it CAN be fixed if/when the powers that be decide to address it.

How bad is the jitter, and is it tempo dependent (if so, faster tempos can mitigate it, maybe?).

Also, this would suggest that it's entirely possible that MIDI coming INTO reason is just as susceptible to this jitter? Have you tested this per chance?

MIDI that I play into Reason is pretty much dead on, assuming system latency is at or less than 128 samples. Something I have NOT tested is if it’s buffer dependent. I will often play in all the parts, and then record them all in as audio (separately or at once doesn’t seem to matter much) at 512 or even 1024 samples, which is another old habit, to avoid buffer under runs.

Changing playback tempo would only be an experiment, as eventually I’ve got to get it into the software as audio at the correct tempo. How it plays back via the live instrument and how it actually comes back on the track are often two different things!

[normen]

Wait, are you saying you're recording the real synths into reason and the recording is late / off the beat in comparison to the MIDI?

Maybe this is your issue, I was preparing to do a video about this. It's a bit hard to explain, I'll try to make water-tight re-readable sentences.

So the DAW is recording the audio in a way so that when you play back later you get what you heard through your speakers while recording. This means that when you have software monitoring enabled the live audio coming in during the recording is heard late (because of the processing time - obvs). This means it's also recorded late on the lane, so that when you play it back later it's the same. It kind of "expects" you to play live and adapt to the latency by listening - which most of the time you'd do as a musician

So if you for example hear your synths through an external mixer and then record in reason with just the faders pulled down Reason will still "think" that you're listening with that delay and your tracks will be off all the time. The solution in that case is to enable hardware monitoring in Reason. Then the recording will be smack on when it was played in "real time" relative to the synths played from the DAW. Because Reason then "knows" you hear the synths in actual real time.

[selig]

Ouch, that sounds bad…and if I'm understanding you correctly, and this works fine in other apps but not Reason, there is good news and bad news.

The bad news is that it's Reason's fault, and they are slow to address this sort of thing, historically speaking. The good news is that it's Reason's fault, and it CAN be fixed if/when the powers that be decide to address it.

How bad is the jitter, and is it tempo dependent (if so, faster tempos can mitigate it, maybe?).

Also, this would suggest that it's entirely possible that MIDI coming INTO reason is just as susceptible to this jitter? Have you tested this per chance?

MIDI that I play into Reason is pretty much dead on, assuming system latency is at or less than 128 samples. Something I have NOT tested is if it’s buffer dependent. I will often play in all the parts, and then record them all in as audio (separately or at once doesn’t seem to matter much) at 512 or even 1024 samples, which is another old habit, to avoid buffer under runs.

Changing playback tempo would only be an experiment, as eventually I’ve got to get it into the software as audio at the correct tempo. How it plays back via the live instrument and how it actually comes back on the track are often two different things!

OK, I was thinking the test would be to record an arpeggiator from a synth compared to using a Reason arpeggiator at the same sync rate and see how the timing compares. That way you know which part of the "trip" is causing which problem.

Would like to understand what's going on here, so any testing would be appreciated - then we can alert the Props so they know exactly what the issue is. I will try to replicate your tests and see if it's interface dependent, buffer dependent (as you mentioned), or "other".

[selig]

Wait, are you saying you're recording the real synths into reason and the recording is late / off the beat in comparison to the MIDI?

Maybe this is your issue, I was preparing to do a video about this. It's a bit hard to explain, I'll try to make water-tight re-readable sentences.

So the DAW is recording the audio in a way so that when you play back later you get what you heard through your speakers while recording. This means that when you have software monitoring enabled the live audio coming in during the recording is heard late (because of the processing time - obvs). This means it's also recorded late on the lane, so that when you play it back later it's the same. It kind of "expects" you to play live and adapt to the latency by listening - which most of the time you'd do as a musician

So if you for example hear your synths through an external mixer and then record in reason with just the faders pulled down Reason will still "think" that you're listening with that delay and your tracks will be off all the time. The solution in that case is to enable hardware monitoring in Reason. Then the recording will be smack on when it was played in "real time" relative to the synths played from the DAW. Because Reason then "knows" you hear the synths in actual real time.

That may be part of the problem which is why I asked if it was timing "latency" or timing "jitter" (or a combination of both). Seems he's seeing more of a jitter issue than a sync/latency "offset" issue, but I want to confirm first.

What tests would you suggest to help pinpoint where the issues are coming from?

[normen]

That may be part of the problem which is why I asked if it was timing "latency" or timing "jitter" (or a combination of both). Seems he's seeing more of a jitter issue than a sync/latency "offset" issue, but I want to confirm first.

What tests would you suggest to help pinpoint where the issues are coming from?

Right, he even said some notes are too early.

I'd try recording the MIDI directly into some other software, bypassing the hardware MIDI parts to test if it's actually coming out of Reason like that. After that into another computer or sequencer via the hardware. And then if that's not straight anymore I'd try with other hardware interface(s). That should give a pretty good idea where the timing issues occur. When recording in software you should probably also try it while playing through the hardware simultaneously and without.

Edit: I use Reason the other way around, playing it with MIDI, so I don't have much experience with the EMI.

[jimmyklane]

I’m going to run a battery of tests.

1: audio pulse at 120bpm from synth
2: audio pulse at 240bpm from synth
(these should be controls I think)
3: MIDI *FROM* Reason to synth, audio return @ 128
4: MIDI *FROM* Reason to synth, audio return @1024
5: Arp from internal tempo, MIDI to Reason
6: Arp from MIDI synced tempo, MIDI to Reason
7: Arp from Reason, audio return at 128 & 1024
8: Arp from Reason to Subtractor, Record to Track

[jimmyklane]

That may be part of the problem which is why I asked if it was timing "latency" or timing "jitter" (or a combination of both). Seems he's seeing more of a jitter issue than a sync/latency "offset" issue, but I want to confirm first.

What tests would you suggest to help pinpoint where the issues are coming from?

Right, he even said some notes are too early.

I'd try recording the MIDI directly into some other software, bypassing the hardware MIDI parts to test if it's actually coming out of Reason like that. After that into another computer or sequencer via the hardware. And then if that's not straight anymore I'd try with other hardware interface(s). That should give a pretty good idea where the timing issues occur. When recording in software you should probably also try it while playing through the hardware simultaneously and without.

I’ll try to see if I use a MIDI loop back, but then who do I blame if the MIDI is messed up? I’ve rarely had use for loopback, even though I’ve got one that I installed for VCV I’ve never actually used it

[jimmyklane]

Wait, are you saying you're recording the real synths into reason and the recording is late / off the beat in comparison to the MIDI?

Maybe this is your issue, I was preparing to do a video about this. It's a bit hard to explain, I'll try to make water-tight re-readable sentences.

So the DAW is recording the audio in a way so that when you play back later you get what you heard through your speakers while recording. This means that when you have software monitoring enabled the live audio coming in during the recording is heard late (because of the processing time - obvs). This means it's also recorded late on the lane, so that when you play it back later it's the same. It kind of "expects" you to play live and adapt to the latency by listening - which most of the time you'd do as a musician

So if you for example hear your synths through an external mixer and then record in reason with just the faders pulled down Reason will still "think" that you're listening with that delay and your tracks will be off all the time. The solution in that case is to enable hardware monitoring in Reason. Then the recording will be smack on when it was played in "real time" relative to the synths played from the DAW. Because Reason then "knows" you hear the synths in actual real time.

Ok, what I’m actually experiencing is MIDI *DRAWN* into the sequencer hitting the MIDI in on my synths, and sounding “wiggly”....like it’s got a swing on it, but not a steady swing either. Straight 16ths do not record as straight 16ths....if I compensate for the latency through the MIDI lines, the opticouplers, the return MIDI line, and system latency by sliding the track where the very first transient is on beat, the rest of the notes may well be off.

I’ll record an example tomorrow. It’s REALLY weird because not only is it somewhat variable, but it’s absolutely different for each MIDI port, synth, and USB device! If it were as simple as a 5ms latency issue, I would never have made a thread!!! I would KILL for something that simple to fix.

The USB on the BassStationII is the worst. Any synth hooked up to the MOTU interfaces seem solid enough, but it could be those synths, as I’ve got just about one port per synth these days, but that includes several USB to MIDI converters like Maschine and my MIDISport 2x2.

I’ll have to repatch and see if I get different results. If it comes down to a device or using the USB connections then I’ll simply use all DIN MIDI and call it done. They were never unstable in Nuendo, and only suffered from regular latency in Live.

[raymondh]

Hi guys

Fascinated by this thread, thought I'd do a test with my system.

Here are the results:
Windows 10 PC, i7, Focusrite 18i8 interface.
Roland System 8 and JX-8P both connected via DIN MIDI cables.

Test 1: 8th notes in piano roll played thru EMI to synth, and audio recorded to Audio track.

System 8 (condition parameter set to -128 to remove the artificial analog slop/tuning)
- latency 12ms (from the MIDI note played to the recorded audio)
- jitter 5ms (I don't know if it's jitter but there is a 5ms variation in the timing of the various notes recorded)

JX-8P
- latency 27ms
- jitter 9ms

Test 2: System 8, playing 8th notes with internal arpeggiator (internal clock, not sync'd to Reason)
- jitter 0 ms!

Conclusions
(1) There is unfortunately some slight MIDI timing issues out of Reason, or Windows, or the MIDI ports on my Focusrite interface (I don't have another DAW to isolate the issue)
(2) Any problems I have with the dear old JX-8P I just put down to character

Here is the Reason file: https://www.dropbox.com/s/8wbl8osth928c ... eason?dl=0

Over to you guys.....

[jimmyklane]

The latency of each synth is different. Unless you’re running analog via CV, you’ve got latency due to MIDI.

It’s the jitter. Ok, so you’ve got 5ms...my ears tell me more from my system, but even 5ms is noticeable inside a groove/pattern.

I’ll run what tests I can here, and post the file here

[normen]

The latency of each synth is different. Unless you’re running analog via CV, you’ve got latency due to MIDI.

It’s the jitter. Ok, so you’ve got 5ms...my ears tell me more from my system, but even 5ms is noticeable inside a groove/pattern.

I’ll run what tests I can here, and post the file here

Well what kind of data are you running over the midi connection? A normal DIN connection has almost 1ms of latency per note.

[jimmyklane]

The latency of each synth is different. Unless you’re running analog via CV, you’ve got latency due to MIDI.

It’s the jitter. Ok, so you’ve got 5ms...my ears tell me more from my system, but even 5ms is noticeable inside a groove/pattern.

I’ll run what tests I can here, and post the file here

Well what kind of data are you running over the midi connection? A normal DIN connection has almost 1ms of latency per note.

Note on/off, velocity, that’s it.

[jimmyklane]

Also, I’m aware that the optoisolator and associated circuitry adds some latency. Latency is not my problem. Jitter is my problem

[normen]

Also, I’m aware that the optoisolator and associated circuitry adds some latency. Latency is not my problem. Jitter is my problem

Yeah but a chord with 5 notes will have 5ms latency, one note only 1ms. I mean a certain amount of jitter is normal over a DIN connection, the time range you mentioned fit exactly to the usual "jitter" you had in the olden days. What you said before sounded more like 20-100ms range issues (to me).

[jimmyklane]

Also, I’m aware that the optoisolator and associated circuitry adds some latency. Latency is not my problem. Jitter is my problem

Yeah but a chord with 5 notes will have 5ms latency, one note only 1ms. I mean a certain amount of jitter is normal over a DIN connection, the time range you mentioned fit exactly to the usual "jitter" you had in the olden days. What you said before sounded more like 20-100ms range issues (to me).

It’s is more like 20-30ms....I wasn’t the one that did the testing and found 5ms.

[normen]

It’s is more like 20-30ms....I wasn’t the one that did the testing and found 5ms.

Ah alright, I got that mixed up. Yeah, thats a funny lot :/

[madmacman]

Hmmm, ok.

This thread encouraged me to do some more tests. Unfortunately absolutely non-scientific. Midi interfaces are a dirt cheap China cable or USB connections to either Blofeld, or Doepfer Dark Energy.

Audio Interface: Before (what I had in mind when I wrote my first reply) I had a Balance. Now for a week or so I have a new Yamaha mixer with USB interface. Recorded at 24bit/44.1KHz with a buffer of 512 samples.

My scenario: 1/8 notes with 1/8 pause in-between played @120bpm

At first glance: The recording in January was much worse than it was today. While not "perfect", the transients are acceptable close enough to the expected position.

Calculating latency: I cut the audio at the first transient and removed the trailing part. Then I looked at the position marker in the top line of the sequencer. All recordings roughly started around 65-70 ticks.

Now I do the maths: 240 ticks are a 16th note. So 60 ticks are a 1/64 which then translates my measurements into 8-9ms

Ok, no idea what I messed up back in January. The setting was different: more tracks, faster bpm, different hardware. But today I wouldn't complain. Provided no one proves me wrong.

[jimmyklane]

My specific issue is resolved at 64-128 samples. With a buffer that small, the MIDI comes in very clean and with very little jitter....less than 4ms for monosynths, and most of my polys are really dead on.

It’s when I attempt this at a full 1024 samples that everything gets messed up.

Can anybody replicate my findings?