Jump to content


Photo

Variax Vs. Axe Fx Pitch Shift?


  • Please log in to reply
41 replies to this topic

#1 rlintz

rlintz

    Just Startin'

  • Members
  • Pip
  • 2 posts

Posted 24 August 2013 - 07:21 PM

I'm really intrigued by the Variax - especially now there's a model with a Floyd.  I play through an Axe FX II and in a hard rock / metal band which requires a lot of various tunings (E, Eb, Drop-D, Drop-Db, C).  I'm able to get by with multiple guitars and occasional use of a fixed harmony pitch shift block via the Axe.  Problem is the slight latency when utilizing a fixed harmony pitch block.  My question is for those with an Axe FX II and a Variax, does the Variax represent an improvement to this approach?  How convincing are the tones - especially the nuances (palm muting, pinch harmonics, etc.) when playing via the modeled pups in various tunings?  I'm fortunate enough to have many guitars so the guitar modeling isn't as important to me as are the tunings.  The thought of bringing one guitar to band rehearsal's (instead of several and selecting the tuning on-the-fly would be very cool.

 

Thoughts?

 

Thanks,

 

R


  • 0

#2 Crusty_Old_Rocker

Crusty_Old_Rocker

    Uber Guru

  • Line 6 Expert
  • 5112 posts
  • LocationGold Coast - Australia

Posted 24 August 2013 - 09:34 PM

I can't help you as far as the Axe FX goes or the style of music you play.  There are some real differences with the modelled guitars and using native magnetic pickups that relate to the modelling being applied to the signal that comes from piezo pickups in the bridge.  However, as for latency, if it's there, I can't detect it and if you were to do the pitch shift via a blend of the original tuning via the magnetic pickups and the harmony tuning via the modelling, I would guess that would overcome any perceived shortcomings or limitations of the modelled tones coming from the piezo pickups.

 

The best thing to do would be to find someone with a JTV89 and invite them over to try it out with your setup.

 

Cheers,

 

Crusty


  • 0

#3 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 24 August 2013 - 09:35 PM

Totally different results. The Variax has a separate pickup for each string, while the Axe can only process the overall sound. Variax does a very good job at pitch shifting each string.


  • 1

#4 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 24 August 2013 - 11:48 PM

I'm really intrigued by the Variax - especially now there's a model with a Floyd.  I play through an Axe FX II and in a hard rock / metal band which requires a lot of various tunings (E, Eb, Drop-D, Drop-Db, C).  I'm able to get by with multiple guitars and occasional use of a fixed harmony pitch shift block via the Axe.  Problem is the slight latency when utilizing a fixed harmony pitch block.  My question is for those with an Axe FX II and a Variax, does the Variax represent an improvement to this approach?  How convincing are the tones - especially the nuances (palm muting, pinch harmonics, etc.) when playing via the modeled pups in various tunings?  I'm fortunate enough to have many guitars so the guitar modeling isn't as important to me as are the tunings.  The thought of bringing one guitar to band rehearsal's (instead of several and selecting the tuning on-the-fly would be very cool.

 

Thoughts?

 

Thanks,

 

R

 

Oz kinda nailed it on the head.

The Variax has pretty high quality pitch shifters, not the best, but pretty awesome.

Variax's pitch shifter mechanism has a bit of formant preserving in it's algorithm, or at least compared to a very basic pitch shifter, it does.

 

Since the Variax pitch shifts each string individually, the tuning will sound a LOT better than pitch shifting all 6 strings at once with 1 pitch shifter. 

 

As for latency, there will always be latency when it comes to pitch shifters, as it needs to read a bit of the input's wave before it can determine what the algorithm needs to do to make it sound like the correct pitch it's shifting to.

 

Latency never seemed to be much of a problem with me on the Variax, and I hate latency. 

 

 

Like I said before, the quality is pretty good. Downtuning doesn't sound like muddy garbage with cut off highs, it actually keeps a lot of the tonality when the pitch shifter is on. Like I said, I think it has some formant preserving going on in the algorithm, which helps make it sound natural.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#5 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 25 August 2013 - 10:11 AM

As for latency, there will always be latency when it comes to pitch shifters, as it needs to read a bit of the input's wave before it can determine what the algorithm needs to do to make it sound like the correct pitch it's shifting to.

 

Latency never seemed to be much of a problem with me on the Variax, and I hate latency. 

 

It shouldn't need to read a bit of the input wave in terms of general pitch shifting. I would imagine all you need to do is adjust by 12th root of 2 (approximately 6%) for each semi-tone. Sample at 44.1KHz, play back at 41.6kHz. Almost 0 latency. The only additional processing would be that you'd need to skip a sample 6% of the time, so you might want to blend that with the samples around it. That could introduce some overall latency, but it should be tiny and in no way audible (although it would produce overtones). Unless I'm missing something?


  • 0

#6 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 25 August 2013 - 11:27 AM

It shouldn't need to read a bit of the input wave in terms of general pitch shifting. I would imagine all you need to do is adjust by 12th root of 2 (approximately 6%) for each semi-tone. Sample at 44.1KHz, play back at 41.6kHz. Almost 0 latency. The only additional processing would be that you'd need to skip a sample 6% of the time, so you might want to blend that with the samples around it. That could introduce some overall latency, but it should be tiny and in no way audible (although it would produce overtones). Unless I'm missing something?

 

"Almost 0 latency" You just defeated the purpose of your argument. 

I said that pitch shifting requires at least SOME latency. "Almost 0 latency" is not 0 latency, and is -> some latency.

 

I've messed with pitch shifters that have latency adjusters on it. The less latency you put into it, the less accurate it will pitch shift. At extremely low latency, it just sounds like complete noise.

 

I'm sure there's different approaches to pitch shifting out there that kill latency inducing routines in the programming, but

you have to understand that there's a balance between latency and quality of pitch shifting. 

The Variax is something you'll want quality pitch shifting with.

 

 

In the end, I've tested a LOT of pitch shifters. All of them have latency. They need to read some of the wave to determine how to pitch shift it WITHOUT sounding horrible.

Not one of them had no latency. If no latency is possible, I'm sure one of them would have done that by now out of the bundle I have tried.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#7 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 25 August 2013 - 12:00 PM

"Almost 0 latency" You just defeated the purpose of your argument. 

I said that pitch shifting requires at least SOME latency. "Almost 0 latency" is not 0 latency, and is -> some latency.

 

I've messed with pitch shifters that have latency adjusters on it. The less latency you put into it, the less accurate it will pitch shift. At extremely low latency, it just sounds like complete noise.

 

I'm sure there's different approaches to pitch shifting out there that kill latency inducing routines in the programming, but

you have to understand that there's a balance between latency and quality of pitch shifting. 

The Variax is something you'll want quality pitch shifting with.

 

 

In the end, I've tested a LOT of pitch shifters. All of them have latency. They need to read some of the wave to determine how to pitch shift it WITHOUT sounding horrible.

Not one of them had no latency. If no latency is possible, I'm sure one of them would have done that by now out of the bundle I have tried.

 

Waddap Clay-man,

 

By almost 0 latency, I meant so small you can't possibly hear it. Nothing has 0 latency. Even a 1 inch wire. That's why I didn't say 0 latency. No-one should ever say 0 latency, as there's no such thing. As an example, moving the speaker closer to my ear reduces latency too.

 

I guess I'm not seeing where the complexity lies. The only complexity I can thing of is trying to smooth out the overtones/warble. That could introduce additional latency while waiting for samples, as well as latency due to processing time. In any case, whatever Line 6 does in the Variax, they do it well and I've never heard any latency when using alternate tunings (i.e.: it's so small, you can't hear it).


  • 0

#8 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 25 August 2013 - 12:13 PM

Waddap Clay-man,

 

By almost 0 latency, I meant so small you can't possibly hear it. Nothing has 0 latency. Even a 1 inch wire. That's why I didn't say 0 latency. No-one should ever say 0 latency, as there's no such thing. As an example, moving the speaker closer to my ear reduces latency too.

 

I guess I'm not seeing where the complexity lies. The only complexity I can thing of is trying to smooth out the overtones/warble. That could introduce additional latency while waiting for samples, as well as latency due to processing time. In any case, whatever Line 6 does in the Variax, they do it well and I've never heard any latency when using alternate tunings (i.e.: it's so small, you can't hear it).

 

Very true, but by latency I'm talking about processing latency, as in adding more time between if the effect was processed, or if it was bypassed.

 

There are 0 latency processes when it comes to DSP stuff, EXCLUDING D/A A/D conversion and all hardware latency. This is called real time processing.

Pitch shifting usually has latency of at least 3-5 ms minimum.

 

Do you understand what I'm saying?

 

 

Edit: The complexity is what would cause latency, but it's needed. Like I said, if there was no latency, it wouldn't know how to process the pitch and it would be absolutely warble noises. It HAS to have SOME processing latency to process the pitch without warble.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#9 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 25 August 2013 - 12:43 PM

Very true, but by latency I'm talking about processing latency, as in adding more time between if the effect was processed, or if it was bypassed.

 

There are 0 latency processes when it comes to DSP stuff, EXCLUDING D/A A/D conversion and all hardware latency. This is called real time processing.

Pitch shifting usually has latency of at least 3-5 ms minimum.

 

Do you understand what I'm saying?

 

I do, yes. One of the problems in the guitar world is people get hung-up on latency. Then they start hearing things that aren't there (I'm not saying this is you at all. Merely that people hear the word latency and immediately they can't play anymore). So, I'm just making the point that everything has latency, and it's only an issue if it truly becomes audible.


  • 1

#10 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 25 August 2013 - 12:50 PM

Yessir,

 

 

I do, yes. One of the problems in the guitar world is people get hung-up on latency. Then they start hearing things that aren't there (I'm not saying this is you at all. Merely that people hear the word latency and immediately they can't play anymore). So, I'm just making the point that everything has latency, and it's only an issue if it truly becomes audible.

Latency is something you'll want the minimum, but it's something you can't get rid of. It all depends on the routine you're doing and how fast it's being processed.

 

Pitch shifting does a routine that requires latency, which doesn't matter how fast the processor is, because a processor can only read the current form of the wave, hence why it needs latency to read some of the wave, and apply the algorithm to pitch shift the input without making it sound like crap.

 

Latency shouldn't be of concern though unless it is actually noticeable, like you said. 6-8 ms on Reaper and the latency of Variax on alt tunings, and I still don't really notice any latency.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#11 johnnyayyy

johnnyayyy

    Power User

  • Members
  • PipPipPipPip
  • 1375 posts

Posted 25 August 2013 - 01:16 PM

You guys sound like you know what you are talking about. I will add this:

 

I was noodling around with my JTV two days ago and was playing the 12 string acoustics, I tried turning the Alt Tune knob down a full step and I guarantee you I could feel and hear the latency between what my fingers were doing and what was coming out of my headphones. Switching back to "Standard" on the alt tune knob the latency was gone, or rather decreased to the point I could no longer hear or feel it.

 

I have read of others complaining about the latency they experienced with the Variax but had never noticed it myself until now - likely because I have not used the alt tunings much in the past.

 

I think maybe it would be safe to say that whatever latency is there will be noticeable and possibly problematic for some and not noticeable or bothersome for others, but it is definitely there.

 

I do not know whether the amount of latency I experienced with the JTV would be a create a problem for me when playing in time with others, but I think it might. Will test soon...


  • 0

#12 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 25 August 2013 - 01:25 PM

You guys sound like you know what you are talking about. I will add this:

 

I was noodling around with my JTV two days ago and was playing the 12 string acoustics, I tried turning the Alt Tune knob down a full step and I guarantee you I could feel and hear the latency between what my fingers were doing and what was coming out of my headphones. Switching back to "Standard" on the alt tune knob the latency was gone, or rather decreased to the point I could no longer hear or feel it.

 

I have read of others complaining about the latency they experienced with the Variax but had never noticed it myself until now - likely because I have not used the alt tunings much in the past.

 

I think maybe it would be safe to say that whatever latency is there will be noticeable and possibly problematic for some and not noticeable or bothersome for others, but it is definitely there.

 

I do not know whether the amount of latency I experienced with the JTV would be a create a problem for me when playing in time with others, but I think it might. Will test soon...

 

if Standard is your physical tuning (E standard), that means the pitch shifter is OFF, which means there's no latency going on at all, because its... well... off. It doesn't need to add latency if it's not actually shifting the strings.

 

I believe it does add latency to all strings if just 1 is tuned different, but this is so you don't have delay between the strings your playing.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#13 johnnyayyy

johnnyayyy

    Power User

  • Members
  • PipPipPipPip
  • 1375 posts

Posted 25 August 2013 - 01:44 PM

if Standard is your physical tuning (E standard), that means the pitch shifter is OFF, which means there's no latency going on at all, because its... well... off. It doesn't need to add latency if it's not actually shifting the strings.

 

I believe it does add latency to all strings if just 1 is tuned different, but this is so you don't have delay between the strings your playing.

 

 

Yes, that is what I meant.

 

I definitely felt latency with the Alt tuning knob set to 1 step down, latency was gone when I turned the alt tuning off, therefore I deduce the alt tuning process is adding a bit of detectable latency.

 

An easy test would be to make a recording of two tracks in protools or similar, one track with a microphone placed near the JTV strings to record the guitar acoustically and one track with the JTV alt tuning knob engaged output  plugged straight into the recorder.

 

Magnify the waveforms and measure the time difference between the start of a note on the mic'ed track and the start of a note on the DI alt tuned track. I am guessing the latency might become greater as you go lower and lower on the alt tune knob...?

 

I have a JTV 89, just remembered the alt tune knob has different settings than on JTV69 and 59 so your results may vary - I think I will go measure this right now, I am kind of curious exactly how long a delay I am experiencing and if the delay gets longer as the notes go lower


  • 0

#14 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 25 August 2013 - 02:16 PM

I was noodling around with my JTV two days ago and was playing the 12 string acoustics, I tried turning the Alt Tune knob down a full step and I guarantee you I could feel and hear the latency between what my fingers were doing and what was coming out of my headphones. Switching back to "Standard" on the alt tune knob the latency was gone, or rather decreased to the point I could no longer hear or feel it.

 

I have read of others complaining about the latency they experienced with the Variax but had never noticed it myself until now - likely because I have not used the alt tunings much in the past.

 

I have never noticed any latency with alt tuning, even on my Variax 700.

 

That being said, (1) I don't doubt what you're hearing and (2) I gather the 12-string modelling requires significantly more processing than the 6-string modelling. Hence, alt-tuning of 12 strings models in the 700 were not available from what I recall. So, if you're going to have latency issues, they're going to be most apparent in the 12-string models. I don't tend to use alternate tunings on my 12-strings so have never noticed it. My guitars are overseas at the moment, so I can't test it. Hopefully it's not that big an issue it throws out people's timing. I'd like to know how many ms delay from picking to hearing makes it noticeable. I'd be guessing somewhere around 25ms from my audio days.


  • 0

#15 Charlie_Watt

Charlie_Watt

    Power User

  • Members
  • PipPipPipPip
  • 1196 posts

Posted 25 August 2013 - 03:29 PM

All music has latency.  Sound travels 1 foot in 1 msec so if you are 5 feet away from your amp speaker, you have 5 msec of latency just due to the sound path length.  A pitch shifter by it's nature has to have some latency.  It has to measure the frequency of the note before it can shift it.  Physics!


  • 0

#16 johnnyayyy

johnnyayyy

    Power User

  • Members
  • PipPipPipPip
  • 1375 posts

Posted 25 August 2013 - 03:42 PM

I have never noticed any latency with alt tuning, even on my Variax 700.

That being said, (1) I don't doubt what you're hearing and (2) I gather the 12-string modelling requires significantly more processing than the 6-string modelling. Hence, alt-tuning of 12 strings models in the 700 were not available from what I recall. So, if you're going to have latency issues, they're going to be most apparent in the 12-string models. I don't tend to use alternate tunings on my 12-strings so have never noticed it. My guitars are overseas at the moment, so I can't test it. Hopefully it's not that big an issue it throws out people's timing. I'd like to know how many ms delay from picking to hearing makes it noticeable. I'd be guessing somewhere around 25ms from my audio days.



I just did some testing, looks like on the T-Model setting there is around 2ms delay between initial waveform peaks when comparing mic'ed track to JTV direct output track, probably not enough that anyone would ever be able to detect.

http://farm3.staticf...47ea58915_b.jpg

Turning the alt tune knob down did increase the latency on the T-Model but not by much, changed to around 10ms

http://farm6.staticf...e0c40c614_b.jpg

Switching to 12 string acoustic in standard tuning the latency is around 5ms

http://farm8.staticf...8d819f60a_b.jpg

Alt tuning knob tuned to C changes latency to around 17ms

http://farm4.staticf...e738afe7d_b.jpg

I think I remember reading that around 10ms is where latency becomes noticeable for most, so things feeling weird to me on the detuned 12 string at 17ms latency between striking a note and hearing it in my headphones is probably not just my imagination.

Whether it will be enough to throw off my timing or I can compensate for it mentally and physical is a different question - when I bought my first ProTools rig I wasted a year recording and rerecording, thinking I had lost all sense of rhythm because I simply could not play in time with other players tracks for overdubbing, after a year or so of frustration I found the problem was the latency in my DAW/computer, got a different interface for recording with near zero latency, and suddenly I was able to play the way I had for all the years I spent doing analog recording before I moved to ProTools. I don't know how much latency that rig had but it was plenty to make it impossible for me to play...

Will do some tests with the 12 string this week and see whether I am able to use them.

 

EDIT: I just realized I needed to add 3 ms to the latency figures above to allow for the 3 foot distance from the JTV to the mic when I made the comparison recordings.

 

Tele = 5ms total latency, 12 string acoustic = 8ms latency, and 12 string acoustic alt tuned down to C = 20ms latency.


  • 1

#17 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 25 August 2013 - 04:04 PM

All music has latency.  Sound travels 1 foot in 1 msec so if you are 5 feet away from your amp speaker, you have 5 msec of latency just due to the sound path length.  A pitch shifter by it's nature has to have some latency.  It has to measure the frequency of the note before it can shift it.  Physics!

 

I know there's physical latency, but we're talking about process/digital latency.

This is latency that happens during the input->processing->output stage.

 

There's 2 types of latency with this kind of latency

 

1) Over all performance latency. This is the buffer/latency you set on your DAW. This is the amount of time you give your processor to preform an effect or task. The lower the latency, the harder and faster your processor has to work. If your processor can't process the effect in fast enough, you will have crackling and/or stops in the audio.

 

2) There's latency which is induced by an effect. It sets a determined latency because that latency is deemed an appropriate amount of time for the effect to work correctly.

An effect that causes latency most likely needs to read some of the input in order to apply the effect correctly.

This latency has NOTHING to do with how fast or strong the processor is. You can have a fast processor and the latency will still be the same, as it's a set latency by the effect, not the overall buffer of the processor.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#18 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 25 August 2013 - 04:39 PM

All music has latency.  Sound travels 1 foot in 1 msec so if you are 5 feet away from your amp speaker, you have 5 msec of latency just due to the sound path length.  A pitch shifter by it's nature has to have some latency.  It has to measure the frequency of the note before it can shift it.  Physics!

 

Actually Charlie, it doesn't necessarily have to measure the frequency of the note before shifting. Fortunately, frequencies are geometric. Therefore, you just sample the note at one frequency, and play the samples back at a different frequency. As long as your sample rate is high enough, the pitch is magically shifted by that %. Approximately 6% per semi-tone. But, you have to drop the same percentage of the samples, and this introduces artifacts. This is when you might want to get some sort of feel for the surrounding samples in order to reduce the artifacts.


  • 0

#19 Charlie_Watt

Charlie_Watt

    Power User

  • Members
  • PipPipPipPip
  • 1196 posts

Posted 25 August 2013 - 05:16 PM

Yes but it still has to have a sample of the pitch it's going to shift so there has to be some digital delay.


  • 0

#20 rlintz

rlintz

    Just Startin'

  • Members
  • Pip
  • 2 posts

Posted 25 August 2013 - 05:49 PM

Uhhh... So, is anyone using their Variax in a live setting switching between various tunings in a hard rock/metal setting? Are you satisfied with the tracking, palm muting, and other playing nuances? How big is the difference between the Variax and a rack full of differently tuned guitars?

Thanks guys, ,

R
  • 0

#21 johnnyayyy

johnnyayyy

    Power User

  • Members
  • PipPipPipPip
  • 1375 posts

Posted 25 August 2013 - 06:12 PM

Uhhh... So, is anyone using their Variax in a live setting switching between various tunings in a hard rock/metal setting? Are you satisfied with the tracking, palm muting, and other playing nuances? How big is the difference between the Variax and a rack full of differently tuned guitars?

Thanks guys, ,

R

 

 

Not 100% on topic and if you start a new thread asking this question you will get more responses, but... Palm muting on the models is not quite 100% tho I would say it is greatly improved over the old Variax, maybe 90-95% correct now - some things you will definitely still need the magnetic pickups to pull off effectively but there they are and easy to switch to any time you need them.

 

I recommend finding a JTV at a dealer near you and trying it out before you buy to see if it works for you, or if this is not possible you could always buy a JTV and return it within the 45 day period most online dealers allow.

 

Tracking is good on  the 6 string models, not perfect and a little slow w some noticeable latency on the 12 string models in my experience/opinion.


  • 0

#22 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 25 August 2013 - 07:04 PM

Yes but it still has to have a sample of the pitch it's going to shift so there has to be some digital delay.

 

No, that's my point, it doesn't need a sample of the "pitch". It just needs to sample the amplitude at twice the highest frequency it wants to reproduce, e.g. 44.1KHz, then play it back at 44.1KHz - 6%, dropping every 17th sample. The delay is effectively 0.

 

The overtones and warbling is caused by the dropping of the samples, so you need to do some nice stuff for that, but changing the frequency by itself doesn't inherently require "sampling the pitch". It can be done without any knowledge whatsoever of what notes are being played.


  • 0

#23 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 25 August 2013 - 10:05 PM

No, that's my point, it doesn't need a sample of the "pitch". It just needs to sample the amplitude at twice the highest frequency it wants to reproduce, e.g. 44.1KHz, then play it back at 44.1KHz - 6%, dropping every 17th sample. The delay is effectively 0.

 

The overtones and warbling is caused by the dropping of the samples, so you need to do some nice stuff for that, but changing the frequency by itself doesn't inherently require "sampling the pitch". It can be done without any knowledge whatsoever of what notes are being played.

 

I know it doesn't need the pitch, but it still needs data and time to process everything correctly.

 

First half is the pitch shifter on 38 ms latency, second half is on 3.5 ms latency

it can't even track chords on 3.5 ms.

 


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#24 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 25 August 2013 - 10:21 PM

I'm just trying to separate the two concepts. Charlie said you need a sample of the "pitch" so I was clarifying that.

 

To re-iterate,

 

1) you can pitch shift without determining the frequencies being played as described above. In fact, it's very easy, but doesn't necessarily sound very good.

 

2) to compensate for the artifacts that are created if you do the most basic pitch-shifting, which has essentially 0 latency, you need to introduce computing to try to determine how to minimize those artifacts. Even then, you're probably not looking at the freqeuncies of the notes being played. More like the general shape of the nearby samples to try to work out how to adjust for the dropped samples with minimal artifacts. Now, that data does include the shaping caused by the frequencies being generated, but you're not inherently looking at the frequencies themselves.


  • 0

#25 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 25 August 2013 - 11:07 PM

I'm just trying to separate the two concepts. Charlie said you need a sample of the "pitch" so I was clarifying that.

 

To re-iterate,

 

1) you can pitch shift without determining the frequencies being played as described above. In fact, it's very easy, but doesn't necessarily sound very good.

 

2) to compensate for the artifacts that are created if you do the most basic pitch-shifting, which has essentially 0 latency, you need to introduce computing to try to determine how to minimize those artifacts. Even then, you're probably not looking at the freqeuncies of the notes being played. More like the general shape of the nearby samples to try to work out how to adjust for the dropped samples with minimal artifacts. Now, that data does include the shaping caused by the frequencies being generated, but you're not inherently looking at the frequencies themselves.

 

It definitely does need some detail in order to minimize artifacts indeed. It's why polyphonic pitch shifting can get a bit warbly but still not sound like complete trash, because the more notes you have in the input, the more complicated it is to work with the wave.

 

Not only does having a pitch shifter assigned to each string allow you to tune each string individually, but it helps the Variax pitch shift with better results.

 

Either way, the latency is used as a fallback to have time to tailor the signal into a good quality pitch shift.

The best VST pitch shifter I've seen is called Pitchwheel, and that needs 46 ms to do high quality pitch shifting. It sounds remarkable.

Reaper's ReaPITCH also is high quality and has high latency.

 

The less latency, the less natural and professional the pitch shifting will sound.

 

For the Variax, you need a good balance between latency and quality. Fast enough to not ruin your guitar playing, but good enough quality for it to pass as capo/detuning substitute.

I say the Variax does a nice job.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#26 Charlie_Watt

Charlie_Watt

    Power User

  • Members
  • PipPipPipPip
  • 1196 posts

Posted 26 August 2013 - 05:37 AM

Yes - the pitch shift is achieved by resampling but resampling requires digital filtering too.  I was not implying that there would be lots of latency in pitch shifting but there is some - and I doubt that anyone can really detect it.  More likely what folks notice is some artifacts if the filtering isn't done properly.  Today's processors are plenty fast to be able to do complex digital filtering without adding noticeable latency.


  • 0

#27 johnnyayyy

johnnyayyy

    Power User

  • Members
  • PipPipPipPip
  • 1375 posts

Posted 27 August 2013 - 05:07 PM

I was not implying that there would be lots of latency in pitch shifting but there is some - and I doubt that anyone can really detect it.  More likely what folks notice is some artifacts if the filtering isn't done properly.

 

Really...? Have you played the 12 string acoustics in HD 2.0 alt tuned down 3 or 4 half steps? I would be surprised if anyone tried this and did not feel the notes coming out of their headphones waaaaaaayyyyyy after they are played. It feels like I am playing in slow motion, if playing without a timing reference I keep feeling myself slowing down waiting for the sound in my headphones to catch up to what my fingers are doing.... unless that is what you mean when you say " filtering artifacts" -  I always thought that time delay I am experiencing (around 17ms between the time I strike a note and it is played back by my JTV as shown in the ProTools screen caps above) ) was what everyone means when they say "latency". Am I misunderstanding what you are saying?


  • 0

#28 johnnyayyy

johnnyayyy

    Power User

  • Members
  • PipPipPipPip
  • 1375 posts

Posted 27 August 2013 - 05:27 PM

Yes - the pitch shift is achieved by resampling but resampling requires digital filtering too.  I was not implying that there would be lots of latency in pitch shifting but there is some - and I doubt that anyone can really detect it.  More likely what folks notice is some artifacts if the filtering isn't done properly.  Today's processors are plenty fast to be able to do complex digital filtering without adding noticeable latency.

 

Also, I am not 100% sure that what is going on in the 12 string emulation is a simple pitch shift - glitches in the 2.0 software show the processing is behaving differently than a regular pitch shifter.

 

If you have not already, try this:

 

Physically tune your D string one step down to C. Select the Chime 2 Rickenbacker 12 string and play the string open a few times until you hear what sounds like a car crashing. Hear that hugely out of tune clash? The string is sounding a C but the JTV is generating a C the same as the string tuning with the simulated octave string sounding a C#.

 

Now slowly turn the D string tuner sharper and you will hear the lower not rise while the higher octave note stays the same UNTIL both notes are in tune, similar to what a "smart" harmonizer does when you play a note that is not in the scale you have set the harmonizer to follow.

 

Try it! Then please explain what is going on. According to Line 6 it is impossible for the JTV to generate an alt tuned note more than one octave above the note it is tracking, but this experiment says otherwise. According to Line 6 support the JTV is calibrated to operate with the guitar tuned to standard tuning only, and tuning the string to anything other than standard tuning is not supported.

 

Another L6 support person tells me that harmonics are confusing the alt tuning and the problem is possibly due to the guitar's setup (buzzing fret rattle or bad intonation), I thought this might be the case and had already checked the intonation and raised the action so sky-high at the nut that the string could not possible be hitting any of the frets when played open but this does nothing to reduce the amount of random pitch generation.

 

Explanation?

 

ancient-aliens-i-dont-know-therefore-ali


  • 0

#29 Charlie_Watt

Charlie_Watt

    Power User

  • Members
  • PipPipPipPip
  • 1196 posts

Posted 27 August 2013 - 05:53 PM

I played around just now with the 12 string models and standard tuning and I do not notice any latency when playing.  Maybe it's me.  One thing I did notice is a big improvement in the 12 strings when I added some damping above the nut.  Much cleaner sounding to me.  One thing to consider - a real 12 string has a little latency on the string pairs when strumming them.  Of course this would depend on the direction of the strumming.  Overall, I am very pleased with the 12 string sounds - much better than on my old 500 model.


  • 0

#30 johnnyayyy

johnnyayyy

    Power User

  • Members
  • PipPipPipPip
  • 1375 posts

Posted 27 August 2013 - 06:25 PM

I played around just now with the 12 string models and standard tuning and I do not notice any latency when playing.


Me too, did not notice any latency in the 12 strings with standard tuning and measured only around 5ms. When switching to alt tunings I noticed a big latency and measured around 17ms, or an additional 12ms when compared to standard tuning.

Measurements were made picking single notes with a standard plectrum, no strumming involved.

And while the friction from the extra six strings on a real 12 string might slow down picking or strumming if strength and angle of attack were the same as on a 6 string, the notes would certainly not take longer to get to my ear just because there were 12 strings - this is what I am hearing on the 12s with alt tuning, the alt tuning is causing a longer delaybetween the time a note is struck and it reaching my ears.
  • 0

#31 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 27 August 2013 - 09:09 PM

 the alt tuning is causing a longer delaybetween the time a note is struck and it reaching my ears.

 

So, move your ears. :)


  • 0

#32 johnnyayyy

johnnyayyy

    Power User

  • Members
  • PipPipPipPip
  • 1375 posts

Posted 27 August 2013 - 09:57 PM

So, move your ears. :)

 

Hmmm, the signal path at the moment is JTV, Headphone amp, headphones, ears... so the ears should go before the headphones or before the preamp...?

 

JTV 1/4" out DI to ears? :lol: 


  • 0

#33 ozbadman

ozbadman

    Power User

  • Members
  • PipPipPipPip
  • 1689 posts

Posted 27 August 2013 - 10:57 PM

JTV 1/4" out DI to ears? :lol:

 

      


  • 0

#34 DocSmith43

DocSmith43

    Just Startin'

  • Members
  • Pip
  • 2 posts

Posted 08 September 2014 - 05:57 AM

Hi - New to the forum and JTV 59 - but my band plays tunes that require me to switch a lot from Banjo to reso guitars to 12 strings acoustics to electrics, etc.... (haven't found a use for the Sitar yet!! :) )

 

Any rate, the delay in plucking and then hearing a single note line played on the Banjo or reso models is quite noticeable and does throw my timing off. Trying to compensate by playing ahead of the beat slightly but not so easy! 

 

Johnnyayyy's comment made me LOL! I'm not sure if Line 6 is working on that??

But is there anything> something else that can be done in perhaps WB that will reduce this perceived delay/latency?  

Thanks in advance,

 

DocSmith43


  • 0

#35 stevekc

stevekc

    Iknowathingortwo

  • Members
  • PipPip
  • 287 posts

Posted 08 September 2014 - 08:38 AM

My 2 cents - DSP created Alt Tunings require effort and this adds latency.

And as DocSmith43 noted above, the  Banjo or Reso models also yield a delayed output that depending on your experience and ears  - is either acceptable or not.

 

It boils down to who has the fastest hexaphonic Guitar Modeling DSP  system in 2014  - and IMHO the Boss GP-10 is the current king

 

http://www.vguitarfo...p?topic=10782.0

tn1_94921.jpg

 

 


  • 0

#36 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 08 September 2014 - 02:54 PM

My 2 cents - DSP created Alt Tunings require effort and this adds latency.

And as DocSmith43 noted above, the  Banjo or Reso models also yield a delayed output that depending on your experience and ears  - is either acceptable or not.

 

It boils down to who has the fastest hexaphonic Guitar Modeling DSP  system in 2014  - and IMHO the Boss GP-10 is the current king

 

http://www.vguitarfo...p?topic=10782.0

tn1_94921.jpg

 

 

 

Incorrect.

DSP will not determine how fast the latency is on a pitch shifter.

 

Latency adds chance for a more quality sounding pitch shifting function. The more latency, the more of a sample the pitch shifting mechanism has to work with.

 

Good quality sounding pitch shifters will have some latency. Low latency will warble and sound chopped up. It's just facts.

 

 

Imagine a word suggester on your phone. It can't guess which word you're trying to type until it has enough letters to guess. Same applies with a wave form. It can't pitch shift something properly without enough data to smooth out the waveform correctly.

 

Has nothing to do with DSP, and besides, the GP-10 sounds bad imo.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#37 stevekc

stevekc

    Iknowathingortwo

  • Members
  • PipPip
  • 287 posts

Posted 08 September 2014 - 02:57 PM

 

 

DSP will not determine how fast the latency is on a pitch shifter.

 

I own them all ( VG-8 , VG-88, VG-99, GR-55,  Variax,Tyler Variax ,

 

and the Boss GP-10 has the lowest latency for Alt Tunings and Guitar Modeling in 2014.

We agree to disagree

 

http://www.researchg...the_time_domain

 

 

http://ieeexplore.ie...snumber=4589950

 

http://diuf.unifr.ch...cean/Ocean.html


  • 0

#38 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 08 September 2014 - 08:54 PM

I own them all ( VG-8 , VG-88, VG-99, GR-55,  Variax,Tyler Variax ,

 

and the Boss GP-10 has the lowest latency for Alt Tunings and Guitar Modeling in 2014.

We agree to disagree

 

http://www.researchg...the_time_domain

 

 

http://ieeexplore.ie...snumber=4589950

 

http://diuf.unifr.ch...cean/Ocean.html

 

Once again, it will not determine how fast the latency is.

There's no disagreeing.

 

What determines how fast the latency is is how the algorithm was programmed. It's just fact.

 

Most signal based pitch shifters will splice the audio signal and resample it based on which pitch you shift it to.

 

You shift it 12 semitones down, the signal will be resampled half as slow, which will make it sound an octave lower, but in order for the signal to not fall behind, given a frequency rate, it will jump forward to keep in time with the signal, basically keeping the sound in sync.

 

For shifting up 12 semitones, it'll resample twice as fast, and now, it'll  splice the signal up to basically sound an octave up, but since resampling will make it twice as fast, it stutters the audio so it keeps in time.

 

 

This is just how pitch shifting works. The algorithm will be used to determine how smooth this function sounds, how it overlaps those splices together to make it sound smooth. This is why it needs latency, so it has enough of the signal to work with to know how to smooth the sound out to make it sound natural and not warble.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#39 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 08 September 2014 - 08:58 PM

I own them all ( VG-8 , VG-88, VG-99, GR-55,  Variax,Tyler Variax ,

 

and the Boss GP-10 has the lowest latency for Alt Tunings and Guitar Modeling in 2014.

We agree to disagree

 

http://www.researchg...the_time_domain

 

 

http://ieeexplore.ie...snumber=4589950

 

http://diuf.unifr.ch...cean/Ocean.html

 

Oh by the way, one of your links, the third one, proves my point. Lower latency will make the sound choppier and warblier.

 

Has nothing to do with DSP, but with quality of programming (the algorithm)


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 


#40 clay-man

clay-man

    Gear Head

  • Members
  • PipPipPip
  • 910 posts

Posted 09 September 2014 - 07:28 AM

I can't fathom how BOSS has the "most strongest dsp on the market" and still manage to make their amp modeling sound like dated garbage.


  • 0

For a minute there, I lost myself.

Radiohead_bear-728286%5B1%5D.png 





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users