Intel. Harmony tracking: 7.12 vs. 7.08 & earlier

B

Bakerman

Axe-Master
Just loaded 7.12 (Ultra) and noticed the intelligent/custom shifters take significantly more time to change intervals for a given tracking setting. For example:

Set up a 3rd harmony in C major, tracking at 2.40 (just what I happened to use for the preset in which I noticed this), play B (shifter produces D) then hammer to C. There's a brief yet obvious E-flat note before the harmony shifts to E. Pull off to B and E-flat sounds again before changing to D. Do a moderately-paced B-C trill (16th notes at 150bpm or so) and it sounds pretty bad--the harmony seems to spend as much time at E-flat as at the right note. Turning tracking toward 10 reduces those glitches somewhat but adds latency as expected. As far as I can tell, the glitches are still more prominent than when using that 2.40 setting with firmware 7.08.
 
Bakerman said:
Just loaded 7.12 (Ultra) and noticed the intelligent/custom shifters take significantly more time to change intervals for a given tracking setting. For example:

Set up a 3rd harmony in C major, tracking at 2.40 (just what I happened to use for the preset in which I noticed this), play B (shifter produces D) then hammer to C. There's a brief yet obvious E-flat note before the harmony shifts to E. Pull off to B and E-flat sounds again before changing to D. Do a moderately-paced B-C trill (16th notes at 150bpm or so) and it sounds pretty bad--the harmony seems to spend as much time at E-flat as at the right note. Turning tracking toward 10 reduces those glitches somewhat but adds latency as expected. As far as I can tell, the glitches are still more prominent than when using that 2.40 setting with firmware 7.08.
Click to expand...
Is tracking set to stepped instead smooth?
 
I can't comment on the harmony tracking, but the qulaity of the harmony seems a lot better (less "gurgly").
 
After more experimenting I think this is a general pitch detection thing. Synth is latent moving to the next pitch and feels really awkward to play/hear, response to pitch as a modifier is slower, plus the harmonizer interval thing.

Clips for comparison of 7.08 and 7.12. First there's a synth into amp panned left and guitar/amp panned right. Eruption tapping has some octave-down glitches in both but overall 7.08 tracks far better. After that is a fixed harmony with pitch acting as bypass modifier for the pitch block--the harmony is in the right channel and gets muted when F#4 or higher is played. When I tap a high note after a low one, it takes more time for the harmony to be muted in 7.12, indicated by those staccato-sounding notes. Then there's some intelligent harmonizing. In the 7.12 clip tracking was at 2.40 for the first segment then turned to 10 at 1:02. In 7.08 clip it's at 2.40, 10 and 7.32. (Since 10 actually sounded worse than 2.40 for that one, I tried backing down a bit. In this case it's lower values that keep the out of key notes to near-zero length.)

http://dl.getdropbox.com/u/28651/712.mp3

http://dl.getdropbox.com/u/28651/708.mp3
 
Yeah, I can confirm that the pitch block in general (even in fixed harmony) adds a distinctly audible delay in 7.12 versus what we had in 7.07 (all parameters being equal).

Enough to convince me to switch back to 7.07 (other than the improved tuner in 7.12, I have no urgent need for any of the other new goodies).

Definitely smoother, but unacceptably slower.

Daniel
 
Wow, undeniably audible, hard to believe Cliff would let that get past QC, with them so focused on the editor I certainly won't be the one to bug them about that anytime soon though. I haven't used the harmonizer much but was going to build a patch next week, I'll just have to hold off until it is fixed to implement.
 
I guess I'll have to wait to update - I rely on the harmonizer for several tunes. Thanks for the heads up.
 
+1

One of my patches uses pitch block to bring my guitar half step up ... After the 7.12 upgrade it had quite a delay in it .. I had to adjust tracking thing almost to 0.5 smth to make it more playable which didn't sound as good as 7.08 did in my opinion...

Secretly hoping for a resolution...

On a side note - the tuner is so accurate now that I can definitely tell a big improvement...

Mik.
 
Everyone,
I know FA is busy with the editor but this is probably a big enough problem to warrant a firmware fix. I would gladly trade smoothness for accuracy. I think even if Cliff had to revert us back to the way it was before the improved smoothness I would bet most would be just fine with that. Would be good for as many of you who use the harmony tracking features to give your patches a good test drive and chime in here so Cliff and the team can be aware of it.
 
The pitch block itself didn't change. Anyone claiming to hear extra latency is imagining things. The only change was to the global pitch detector. To get faster tracking set the pitch source to Local Mono.
 
FractalAudio said:
The pitch block itself didn't change. Anyone claiming to hear extra latency is imagining things. The only change was to the global pitch detector. To get faster tracking set the pitch source to Local Mono.
Click to expand...
Bakerman says he believes its coming from the pitch detection in this quote.
Bakerman said:
After more experimenting I think this is a general pitch detection thing. Synth is latent moving to the next pitch and feels really awkward to play/hear, response to pitch as a modifier is slower, plus the harmonizer interval thing.
Click to expand...
Maybe Cliff can check the global pitch detector to see if there is an anomoly with in the algorithm itself.
Just wanted to say thanks for all the hard work Cliff and the team have been putting in here lately.
Fractal Audio Rules!
 
FractalAudio said:
The pitch block itself didn't change. Anyone claiming to hear extra latency is imagining things. The only change was to the global pitch detector. To get faster tracking set the pitch source to Local Mono.
Click to expand...

Did you listen to his example clips? That is a pretty dramatic difference.
 
What Cliff means is that there's no added latency of the kind you'd hear by increasing a voice's delay time. Pick a note and some picked note comes out of the pitch block as soon as it did before. It's only the intelligent/custom shifters that will stay at wrong intervals longer than before with pitch source as global. If local tracking works as quickly/consistently with a variety of tones that's good. I didn't really test it when global worked fine and am now away from the Axe-FX for a few days.

The synth and pitch-as-modifier operation are still compromised with slower global detection, though. Try the synth... the slow response is really obvious and quick notes in close range like C-D-E-D will give a terrible sounding portamento thing instead of a well-defined series of notes like before.
 
Bakerman said:
What Cliff means is that there's no added latency of the kind you'd hear by increasing a voice's delay time. Pick a note and some picked note comes out of the pitch block as soon as it did before. It's only the intelligent/custom shifters that will stay at wrong intervals longer than before with pitch source as global. If local tracking works as quickly/consistently with a variety of tones that's good. I didn't really test it when global worked fine and am now away from the Axe-FX for a few days.

The synth and pitch-as-modifier operation are still compromised with slower global detection, though. Try the synth... the slow response is really obvious and quick notes in close range like C-D-E-D will give a terrible sounding portamento thing instead of a well-defined series of notes like before.
Click to expand...


Correct. FWIW, I have checked the latency (global). It is not that.
 
I've returned the tracking speed to pre-7.12 performance. I'll have new firmware in a day or two.
 
FractalAudio said:
I've returned the tracking speed to pre-7.12 performance. I'll have new firmware in a day or two.
Click to expand...

Dude, you totally rule! I'm looking forward to catching up on firmware - I'm running 7.04 bits which is like prehistoric now :lol:

oldladydevilsign9ge.jpg
 
You must log in or register to reply here.
Back
Top Bottom