Bug? Pitch block delay, comes and goes

G

guitarnerdswe

Fractal Fanatic
I noticed something weird in 8.01. When using the fixed harmony in the pitch block, they pitched part of the signal has a delay which comes and goes. It might be alright, and then all of a sudden, there is a noticable slap back effect caused by the delay of the pitched signal. I'm talking like 40-50ms of delay, and then it corrects itself after a while, and then it comes back etc etc.

Anyone else notice this?
 
Bump for this. Funny no one noticed this, since I got it confirmed by Fractal. It's quite obvious, and nobody else noticed?
 
+1

I have the same or similar issue. I use the pitch block to tune down 1 or 2 steps. 100% Mix, 100% voice1, 0% voice2, no delays configured. Sometimes there is a noticeable delay, I'd even say more than 50 ms, but I cannot measure it.

I am using FW 9 beta, but had this issue before with FW 8 too.
 
I actually measured it and it was as big as 95ms at times. I even sent FA sound clips, and they confirmed the bug, yet nothing happened about fixing it.

And yes, it's still there in the beta 9 firmware. This is a HUGE issue, I can't believe it has been fixed and that people don't notice. Its almost 2 months with a block being unusable (I only use that mode), that's a long time.

Sorry for the tone, but I'm getting quite tired of this.
 
Last edited:
What is your track adjust set to? You'll want to use lower values for lower latency and higher values for more accurate tracking.

I'm pretty happy with it around 3 or so. But it's pretty subjective.
 
I have always used it at 0 for best latency (I only do small shifts), but the problem happens no matter where the track adjust is set. I'm not kidding with 95ms, I measured in a DAW. It's a big old slapback that nobody can miss. The annoying thing is that it comes and goes. I just tried it again. It was fine for 30 second, then it appeared again, and went away, and came back.
 
Last edited:
I just tried measuring this and got about 25-30ms on average. Cliff does say however that the latency will vary as the algorithm adjusts itself automatically to ensure accurate tracking at all times. So I am not saying that 95ms is impossible but I was not able to see it and it should definitely be the exception not the norm.

I wouldn't recommend setting tracking to zero either as I don't think it will ever track correctly at that value. Try it around 2 and let me know if that helps.
 
Hi!

No, sadly the track adjust value doesn't matter. The lowest latency is a bit higher, but it will still swing just as high from that value.

FWIW, I haven't seen any indication of this kind of problem in any of the other pitch block modes, only the fixed harmony. Also, it was never like this before the new pitch tracking came around v8. v7 and before have perfect tracking.
 
The reason for the re-design of the pitch tracking algorithm was for improved accuracy. The old algorithm was much worse when it came to tracking complex chords. It produced artifacts and the new one does not.

Attached is a screen cap showing the latency that I am seeing. I created a test patch that pans the wet signal all 100% to the L and the dry 100% to the R. Top is wet, bottom is dry. X-axis shows the time scale. You can see there's < 20 ms latency between wet and dry. I tried this experiment a few times with different notes and fixed vs intel harmony settings. Results were pretty consistent. Latency increases on more extreme settings (shifting by more than an octave for instance) but that's to be expected.

If you could please post your preset and some screen caps showing the 90ms latency when shifting by 2 steps we'll try to figure out what's causing it. Thanks.
 

Attachments

  • pitch latency.jpg
    pitch latency.jpg
    24.7 KB · Views: 27
  • Adam Pitch.syx
    6.3 KB · Views: 9
Here is a sample preset. It doesn't really matter, it will happen sooner or later on all presets using the fixed harmony block. That's the really tricky part. It might be fine for 1 minute of playing, than appear for 15 seconds and then go away again, and so on.

I've also attached 2 sound clips with 2 different track adjust values. I never "double-hit" any notes, all the double hits are the slapback by the pitch latency.



 

Attachments

  • LUKE 90 CLEAN(2).syx
    6.3 KB · Views: 7
I had noticed it right away, but then I also noticed the Track Adjust parameter on page 2 of the block and have been adjusting that as needed. (btw - I'm using this for bass.) For my 1/2 step down patch, Track Adjust is at 2.84, on my low B patch, Track Adjust is at 3.55 and for my low C patch, Track Adjust is at 4.29.
 
I've tried that too, but as you can hear in the clips, the track adjust didn't matter. The latency issue appeared anyway after a while. Besides, the lower the track adjust, the better the latency,
 
I would agree that you can still hear some of it, but the Track Adjust does help to remove some of the unwanted artifacts. At least in my situation. The odd thing is that it does not appear to be consistent, as it sounded better at a higher level when working with -4 (low C) versus a lower level on -5 (low B).
Funny thing is that I've had other bass players comment on how well the pitch shift was tracking in a live situation. So even though I can hear it when practicing, it is obviously getting hidden in the mix in a live performance.
 
Well, the tracking was great before v8 firmware. This is a new problem, and only in the fixed harmony mode. So if they are using old firmware or an standard/ultra, they wont have any
issues :) Writing from an iphone, sorry for the short answer.
 
I am not able to achieve the latency in your sound clip using the patch that you posted. I'm still getting about 15-30ms on average.

Did you try my patch?
 
Strange, I'm playing the patch I posted right now, and the probem is still there and very apparent. Here is another sound clip ( I just selected the patch and started playing). I also attached a picture of the 70ms latency the patch happened to have this time (look at the black bar under "selection"). Generally, when the problem appears, it's always 70ms or more.


8178926469_a49ae828bf_h.jpg


Anyway, I tried your patch and the problem is still there. I noticed some things though. First, as you can see in the picture the below, the pitched part doubles itself sometimes (the top part is only the pitch signal). In the picture there are 2 "bad" notes, and then 2 good notes.


8178926631_49e2028bdb_h.jpg




Second, the latency issue gets worse with smaller shifts for some reason. I tried you patch, and dailed away the octave, added some slight detune and the latency got to 70-95ms again, instead of the approx 30ms I got with just a low octave.
 
Sorry, I'm still not able to reproduce the 70ms+ latency with any settings and I'm definitely not seeing any of the weird "double" notes.

Has anybody else on the thread tried those patches?

Also, for slight detunes we'd suggest that you use the detune mode of the pitch block instead of the fixed shift.

Are you recording via USB and monitoring directly from the Axe-FX II?
 
I don't like the detune mode, you can't adjust the pitch tracking. Anyway, I should be able to use the fixed harmony for detune, I've done that for years and it has worked flawlessly. Plus there are detune parameters in the mode ;) And there are several factory presets (including the Micro-pitchshift and Detune slap presets) that uses the fixed harmony mode for detune.

You don't have use that small shifts for the bigger latencys to appear. You can do a step or 2 and it will be there too.
I'm not using the USB, I record thru analog outputs and into a Motu soundcard.

It's really strange that it doesn't happen to you. I used your exact patch and got those double notes in the pitched signal, and I can add a pitch block to any patch, add a little detune and the latency problem will be there. There obviously is a problem since you can see it and hear from what I posted. There most be something you can do?

EDIT: Ok, something is seriously wrong here. I just tried your preset again Adam, and when the track adjust is really high or low, there is alot of loud clicks. Especially with the track adjust low. That has never happened before.



First with track adjust at 6.16, then 0, then 10.
 
Last edited:
If values of 0 or 10 are producing artifacts then don't use those values. As I mentioned above, more moderate settings of 1-2 should produce good results with acceptable (for me at least) latency.

Perhaps Cliff can adjust the range of this parameter so that the user is not able to select values that produce artifacts.

I'm sorry I don't have any immediate solutions for you but I assure you we'll continue to look at the latency produced by the the pitch shifter and try to find ways to improve it.
 
You must log in or register to reply here.
Back
Top Bottom