Okay here is step two. I wasn't exactly sure here either how to record midi and audio data from my keyboard but I *think* I figured it out. Here is a screen shot. It definitely looks like they aren't sync'd, unless I set it up wrong.
 

I noticed here that clicking on the front of the notes in audio, you can see in the piano roll view that the midi started ahead. The first and fourth notes had the biggest chunk of midi data in front of them and the second and third were closer to being in sync

I messed with the manual offset but but didn't really notice anything, my midi data in the piano roll view is still ahead of the barline when I play in tempo.
I noticed in the other post that it said to unplug your outputs on audio interface and run one cable from the output to the input. I unplugged the outputs and ran a cable from the right input to the right output and still, everytime I hit measure, I get a different sample and ms. It seems to me like that must be a major problem. I can't use the data if it's always changing, right?

brundlefly, I just tried the suggestion in your first post. I put the manual offset back to 0, and above that I messed with the Timing Offset, right now I'm at 24msec and it seems to be pretty dang close. My midi data is lined up on beat and the audio data seems to be lined up too.
Do you think I should still try enabling IgnoreMIDIInTimeStamps in TTSSEQ.INI? And if so, how do I do that?

I haven't had time to review all your posts, but the thing I was looking for that is definitely a problem is that CEntrance says your Round-trip latency is 769 samples, which is pretty typical for a buffer setting of 256 - maybe just a tad high, indicating there's  little bit more "hidden" (unreported driver and USB port) latency than normal. But SONAR says RT is 1195 samples, which makes no sense for a buffer of 256.

The confusing thing is that having SONAR over-estimate the latency would tend to make it over-compensate the audio latency, so that it would be early relative to the MIDI, whereas it appears you MIDI is early. But with that big a discrepancy between actual and reported/detected latency, anything is possible. You could even be aligning the wrong MIDI events to audio transients. You might want to try checking just a single hit to be sure which way it's off.

I would try to get to the bottom of that RT discrepancy with Line 6/Cakewalk support before doing anything else. You may be able to get some semblance of normal behavior out of the system by adjusting offsets, but you shouldn't have to be entering values that big.

If you do want to try adjusting it, though, the first place to start is with a Manual Offset of 769-1195 = -426 (yes, negative 426 to counteract the over-compensation). And then retest Audio and MIDI sync with Timing Offset at 0.

Okay. I read a little bit more. Here are a few more thoughts;

Yes, when doing the round-trip test with CEntrance, it needs to have an analog audio loopback path from output to input to send the "ping" around and measure the total time (latency). Without that, it shouldn't even be able to report a number.

Normally, you'd expect the round-trip latency to be twice the buffer size (same buffer in and out) plus maybe 88 samples (1ms each way) for A/D/A conversion, plus maybe another 20-30 samples of driver/bus latency, depending on the interface. This would give you 512 + 88 + 30 = 630. Thats why I'm surprised your CEntrance screenshot shows 769. But if it's variable, that's  big problem in itself. It should be consistent to within a sample, if not exactly the same every time. And the difference between what CEntrance reports, and what SONAR reports should be the same at all buffer settings, allowing you to set the Manual Offset and forget it.

Regarding CEentrance reporting "769 samples / 17.44ms". That's just sloppy symbology. It should really be "769 samples = 17.44ms at 44.1kHz" (769 samples / 44100 samples/sec = .01744 sec. = 17.44ms).

Looking at your screenshots, it appears that the audio track is a soft synth track with Waveform Preview enabled. That's not what you want to be doing. You want to set up a track to record the analog input(s) of your Toneport, and connect the audio output(s) of your Novation to the interface and record its synth module output at the same time you record the MIDI driving it.

What brundlefly said!!  The toneport driver may have reported wrong information to Sonar.  Also, if Centrance is reporting different numbers every time, the driver is probably doing something weird, and you might never be able to get a consistent compensation. (Unless they're only different by a few samples).

Bill B

Ah crap I figured I was probably setting it up wrong. Even with nothing plugged in to any inputs or outputs of my toneport, CEntrance still gives a random sample and ms number when I hit measure. And no, they are off majorly, like several hundred samples each time and several ms each time.
Looks like I need a better quality interface.
Is it even worth it to set up an audio and midi track like you mentioned if I'm having these funky driver problems?

What's a decent interface to replace my toneport since it looks like it's probably the problem?

I found a focusrite saffire pro24 new for $223 so I may switch to that.

sorry to dig up an old thread, but i had this issue today. came here, did a search and this topic came up. turns out, none of these solutions were necessary for me. what was necessary in my case was to disable the plugins i had placed in the audio fx bin of one of my tracks. i had totally forgotten that i had put them there, and as soon as i disabled them (didn't have to delete them), i was able to playback my midi drums without them triggering ahead of the beat. wanted to add this to the thread in case it helps someone else in the future.