Poly Capo/Shift has too many artifacts an sounds bad

[cruisinon2]

8 minutes ago, codamedia said:

Maybe we should turn the heat down in here!

 

That would be a first...;)

[codamedia]

9 minutes ago, SaschaFranck said:

Well, I posted an example. I did so before, too. I have no idea how people could still say everything is fine and dandy.

 

You make the assumption and the accusation (whether you realize it or not) that people can't hear it because their ears aren't good enough (eg: untrained ear) to hear it! 

When someone can't hear it... guess what?  Sometimes they CAN'T HEAR IT! 

[qwerty42]

5 minutes ago, SaschaFranck said:

 

Maybe you're right, maybe not 100% right, though.

When I mentioned the polystuff to be quite a warbling mess for the first time in the main 3.0 release thread, I was told it wasn't so and I should possibly tune my guitar properly. So I've been sending chords made of pure sine waves (it really doesn't get much more in tune) through the polystuff and the warbling was there in full glory as soon as it wasn't simple fifths or triads anymore. And yet people keep saying it's all great and it must be operator error or whatever.

 

 

Well, I posted an example. I did so before, too. I have no idea how people could still say everything is fine and dandy. But then, of course, I could stay out of this discussion entirely - just that it makes no sense. In case someone notices the less-than-shiny pitch shifting quality of the polystuff, I'll support him/her, simply because that's how it is.

 

Do you have perfect pitch? I mean that genuinely and am curious. Some people have a *much* higher pitch acuity/sensitivity than others, which makes small pitch differences and dissonances glaringly obvious compared to the majority of people. I'm wondering if maybe you fall into that group and that's why to you it's unusable but to others it's hardly noticeable. For some people fretted instruments in general are annoying to listen to because the notes all across the fretboard don't have ideal intonation just because of the simplifications required to build it. Likewise, some people hear no real difference with 'squiggly frets' (frets bent into weird shapes to correct the intonation) while to others they are a gamechanger.

[SaschaFranck]

1 minute ago, qwerty42 said:

Do you have perfect pitch? I mean that genuinely and am curious.

 

No. And I'm not talking about anything exactly pitch related here. Just a warbling effect. Something like a weirdo modulation effect, just that it's not meant to be weird.

[SaschaFranck]

4 minutes ago, codamedia said:

 

You make the assumption and the accusation (whether you realize it or not) that people can't hear it because their ears aren't good enough (eg: untrained ear) to hear it! 

When someone can't hear it... guess what?  Sometimes they CAN'T HEAR IT! 

 

That's all fine - but how is that an accusation then?

[qwerty42]

FWIW, turning it up much louder with my crappy earbuds and listening more closely, I do hear the slight warbles now. But I also hear warbles just from the small dissonances in the first two bars, when the pitch block is still bypassed. What I'm hearing, personally (and I don't expect this to apply to everyone!) is that those 'normal' warbles that are due to the inherent dissonances of those notes are accentuated a little bit once the pitch block is engaged. And yes I also do hear it affecting the background note ringing out, if I intentionally focus on that.

To my ear -- and again, I'm not saying this makes anyone else right or wrong, it's just my personal observation -- I don't think I would ever notice that in a mix. If anything, I'd assume it was just a very subtle chorus effect.

I think this sound clip would make for a very interesting poll -- not meant to prove anyone right or wrong, but just to get an idea of people's average acuity for something like this. I think this might be like the blue/gold dress image and the laurel/yanny audio clip (if you're not familiar with either of those you should definitely look them up), where two people can hear entirely different things.

[codamedia]

10 minutes ago, SaschaFranck said:

That's all fine - but how is that an accusation then?

 

Because you say things like this.... :) 

 

24 minutes ago, SaschaFranck said:

I have no idea how people could still say everything is fine and dandy.

Can you not recognize the condescending tone in what you say?

[cruisinon2]

17 minutes ago, SaschaFranck said:

In case someone notices the less-than-shiny pitch shifting quality of the polystuff, I'll support him/her, simply because that's how it is.

 

 

Harrumph! I'm glad that's settled... saves me the trouble of recording my own clips and climbing Mt. Parnassus to seek divine judgement from the Oracle of Delphi...

 

Now that that's over, let's move onto something less contentious, like politics...;)

[caledoneus]

My guess is you are hearing the notes from the guitar "acoustically" and that is what is making it sound bad to your ears.  

 

[PierM]

Isn't really a matter of "ear" something others aren't. It's a matter of not making a fuss out of subtle glitches, and embrace the art of the constructive criticism instead the abuse of hyperboles. That's it for me at least.

 

I ear the subtle warbles in that mp3 example, but in that context of isolation and hyper listening to single notes, doesn't really tell me anything crazy. Or at least, doesn't sounds like a red flag or a showstopper to me, at all. Not even close.

 

I do think Poly is perfect? Nope. I said for example that a problem I have with it, it's the "pulse" tracking I do have, while playing a phrase under open notes. This can be annoying for my style, and I guess it can be improved, but still, it's a totally usable effect. When using big intervals it's WAY better than what I can achieve with my Variax, for example. For fast phraseggio is still good for me, as there is no time for warble analysis, notes are short, attack on high strings it's very good (can be improved on lower strings), and tone is in general better than those I get from other devices. Not perfect, but really good so far, especially if you are not using it as an alternative to a real detune. In that case, of course, it will never, ever, sound as good as a real detune. On the other hand, if you use the effect to reach sound environments which are not possible for a guitar, like to mimic other string instrument but with a much higher pitch (also with contribution of IRs and the new Acoustic sim), or to mimic some old synth pads, or to just experiment something isn't just a sequence of guitar chords trapped in a standard, then I'd say that Helix poly is good enough, with some range for great improvement.

 

Peace. :)

[SaschaFranck]

31 minutes ago, codamedia said:

Can you not recognize the condescending tone in what you say?

 

No. It's condescending to tell me I should tune my guitar because otherwise PolyPitch would be working just great.

[qwerty42]

5 minutes ago, SaschaFranck said:

 

No. It's condescending to tell me I should tune my guitar because otherwise PolyPitch would be working just great.

I don't see that anywhere in this thread...

And, I mean, considering the current state-of-the-art, it works pretty well, no? What out there works better, for realtime, low-latency processing, in a portable device? If you expect a digital pitch modulation to sound perfect and 100% transparent for this kind of purpose, I'm afraid you're still going to have to wait a number of years for that. I'm sure it will happen eventually, but it's the kind of thing that will probably require some clever machine learning algorithms to 'know' what perfect should and shouldn't sound like, to augment the already-highly-complicated process of frequency separation and individual note shifting for a combined set of notes. This is very challenging stuff in terms of how it's actually accomplished algorithmically.

[SaschaFranck]

8 minutes ago, qwerty42 said:

What out there works better, for realtime, low-latency processing, in a portable device?

 

Kemper, Axe FX, Eventide, HOG (at least regarding tracking and latency). Just to name these.

[codamedia]

20 minutes ago, SaschaFranck said:

No. It's condescending to tell me I should tune my guitar because otherwise PolyPitch would be working just great.

 

Let's make this clear for the sake of this thread.... I did not say that! Not in this thread, and not anywhere! 

[SaschaFranck]

7 minutes ago, codamedia said:

Let's make this clear for the sake of this thread.... I did not say that! Not in this thread, and not anywhere! 

 

Nah, I never said you did. But I have been told that PP was fine and that I was doing something wrong (or whomever else sharing my opinion) more than once.

[PierM]

16 minutes ago, SaschaFranck said:

 

Kemper, Axe FX, Eventide, HOG (at least regarding tracking and latency). Just to name these.

 

Could you post the results of your tests with those devices. Especially curious to listen that generic "eventide".... since all the eventide machines I owned in the past, they were TERRIBLE in tracking...and NONE of them actually do/did a real polyphonic pitch shifting.

 

Please, don't tell me you don't own any of those...:)

[SaschaFranck]

Just now, PierM said:

Please, don't tell me you don't own any of those...:)

 

Why not? I don't. But I have played through them more than once.

[SaschaFranck]

5 minutes ago, PierM said:

Could you post the results of your tests with those devices.

 

Oh btw, here's Bill Ruppert with the Kemper. Sorry, it's not me. But it should clearly demonstrate that the Kempers pitch shifting is in a completely different league.

 

[PierM]

11 minutes ago, SaschaFranck said:

 

Why not? I don't. But I have played through them more than once.

 

What "eventide" did you played that had REAL polyphonic pitch shifting? I'm asking that because Eventide never had, and has, a real polyphonic pitch tracker algorithm (pitchFlex isnt a poly tracker, same goes for PitchFactor algos and so on backward in time), so I'm curious to see if we are comparing bananas with apples here or what. This is kind of funny, as we were always complaining about the lack of a proper polyphonic algo in the Eventide forums, LOL!

[SaschaFranck]

7 minutes ago, PierM said:

What "eventide" did you played that had REAL polyphonic pitch shifting?

 

I might actually be wrong about the Eventide. I have played through their H9s quite often (two pedalboard-nerdy good friends of mine own the max version), but I might not have tried polyphonic things carefully.

[PierM]

1 minute ago, SaschaFranck said:

 

I might actually be wrong about the Eventide. I have played through their H9s quite often (two pedalboard-nerdy good friends of mine own the max version), but I might not have tried polyphonic things carefully.

 

H9 DOES NOT have Polyphonic Pitch Detection, so we are indeed talking apples and oranges here.

 

 

[SaschaFranck]

11 minutes ago, PierM said:

H9 DOES NOT have Polyphonic Pitch Detection, so we are indeed talking apples and oranges here.

 

14 minutes ago, SaschaFranck said:

I might actually be wrong about the Eventide.

 

[SaschaFranck]

Besides, it's not about pitch "detection".

[cruisinon2]

14 minutes ago, PierM said:

 

H9 DOES NOT have Polyphonic Pitch Detection, so we are indeed talking apples and oranges here.

 

 

 

What happened to the bananas? Personally, I think they're a pain in the a$$...I like them, but they're not cost effective. Once they hit that perfect ripe spot, you've got like 17 minutes to eat 'em, or they get disgusting... it's a shorter half-life than Taco Bell.

[SaschaFranck]

15 minutes ago, cruisinon2 said:

Once they hit that perfect ripe spot, you've got like 17 minutes to eat 'em, or they get disgusting...

 

That's what mixers are for. Add kiwi, some frozen raspberries, a dash of honey, fill up with oat milk - there's the most excellent smoothie ever.

[HonestOpinion]

3 hours ago, codamedia said:

 

Bars 3 and beyond have some very noticeable artifacts compared to bars 1 & 2.

 

 

The guitar sounds pretty much the same.... but there are added artifacts that would drive me crazy on a soft passage such as this.  I hear soft thuds, dropouts, and a little warbling. Don't focus on the upper ringing notes... it's the lower notes and what is behind everything where I notice it most.

 

Where I might differ in opinion from @SaschaFranck is that in a louder environment when set in a mix I could probably live with it for a song or two, but certainly not in a solo setting.

 

Not that it can't bear some improvement but I have to say I can't see what I hear on @SaschFranck's recording being a huge problem in a live setting. Heck, at least in the clubs, given the sounds of conversation, glasses clinking, chairs moving, servers taking orders, etc.. I don't even see it being much of a problem in a solo setting unless you guys are getting much more polite crowds than I have experienced or are playing in an auditorium you can hear a pin drop in with a rapt audience. Even then I think it could be passable depending on what is being played. Btw, with the exception of a few socially distanced porch jams all my gigs post-COVID have been "solo", as in, just me :-)

 

This is the first iteration of poly in the HX line and it took a while to get here so although everyone hoped it would be perfect out of the gate I think it is only reasonable to give Line6 a little time to refine it.  It is pretty decent right now and a paradigm shift for HX with some serious potential. Code optimization is actually at the absolute top of my list for poly enhancements so that it chews up less DSP and allows more of the blocks I use in my signal path to remain. Yet another reason I would love to see a modular approach to modelers where I could just throw some more/faster DSP at this. 

 

Glad to see people make specific suggestions on where it can be improved though especially when accompanied by audio that demonstrates areas of concern. Those observations are what will hopefully help drive its future development. It is probably an overstatement to call it unusable(not implying you were saying that) but standards differ between players. Perhaps some will have to wait(could be a long time) to use it live until it hits their threshold for performance use.  For me I think I could find places to use it now, as it is.

[donkelley]

16 hours ago, theElevators said:

I have been using the pitch wham effect to transpose my guitar.  It works almost just as good as the capo. It was already in full polyphony, as far as I'm concerned.  There are some glitchy sounds when the note is about to die.  The 3.0 poly capo is simply making it slightly better and slightly more reliable.  

 

 

That video is an excellent example of it working well, and that's 2.92 or something?  Amazing :-)

 

It's really in knowing how to setup a patch with pitch shifting properly, and how to play cleanly.  Compared to my days of playing with my old kramer pitchrider guitar to midi interface back around 1990 1991 in live shows, playing with the shifting capabilities in any modern system is just a welcome breath of fresh air.  And on the helix 3.0 fw, you have engineer-level control over performance versus latency and accuracy, and so forth.... far more so than on any other system I've ever used.

 

It's truly excellent.

 

Note in the video that you can hear his acoustic guitar sound through the mic - but recorded it would sound amazing.

 

Also - gotta love his guitar :-)  Brian May all the way

[SaschaFranck]

9 minutes ago, HonestOpinion said:

I can't see what I hear on @SaschFranck's recording being a huge problem in a live setting

 

I could post some faster chord progressions, if you wanted. Get's way more obvious. Really, there's no way I would ever want such a sound live - and I'm usually not exactly a hairsplitting guy when it comes to sounds. But I haven't tweaked all my things music for years just to completely give up on all of that by activating PolyPitch.

[donkelley]

1 hour ago, SaschaFranck said:

Besides, it's not about pitch "detection".

Depends on the business logic used.  But yes for the basic old fashioned pitch change logic as found in the pre-polyphonic systems (such as 2.92fw on helix), I'm certain it's using a variation of the standard logic that isn't aware of your pitch... recognition isn't part of it.

 

In poly pitch shifting though, from what I've read about it there is pitch identification as part of how it works - even if it doesn't use that logic to do intelligent pitch shifting, it is part of how it maintains identify of each note being played before shifting each note using more traditional methods.

 

That's how poly shifting works, apparently?  According to my understanding (it's been a few years though since I worked with a guy from digitech who talked to me about the concepts used) - and is the only way it allows complex chords to be shifted.  I'm only relaying my memories of what I was told by someone involved in coding this sort of stuff though.  It's 3rd hand info.

 

It's not new tech and has been around for years, but is highly patented by various manufacturers, including digitech which, even after the buyout, prevented line6 from using their algos for the fw3.0 poly shifting stuff.

 

I fully admit that some companies have done a better job than in fw 3.0 for complex chords, but fw 3.0 is far better than what the non-poly stuff would do in 2.92 from my own extensive tests.  Since this is the first release of poly done by line6 engineers from scratch themselves, I'm excited to see it improve over time.

 

But it's already plenty good enough for me - on my guitars - how I play - for my needs.  I have heard bad examples out there, but for me it tracks quite well.

[HonestOpinion]

2 minutes ago, SaschaFranck said:

 

I could post some faster chord progressions, if you wanted. Get's way more obvious. Really, there's no way I would ever want such a sound live - and I'm usually not exactly a hairsplitting guy when it comes to sounds. But I haven't tweaked all my things music for years just to completely give up on all of that by activating PolyPitch.

 

I can respect that. Even if I could I would never try to impose my standards for what is an acceptable tone or effect on another player. Just saying I think a lot of players will find this usable now at least for some applications. By all means post more examples and even open a ticket with Line6 with specific issues if it helps make Poly better.

[SaschaFranck]

26 minutes ago, donkelley said:

In poly pitch shifting though, from what I've read about it there is pitch identification as part of how it works - even if it doesn't use that logic to do intelligent pitch shifting, it is part of how it maintains identify of each note being played before shifting each note using more traditional methods.

 

No. That way, polyphonic pitch shifting would only work on very limited sources. But it actually works on everything you throw at it (which is also why tuning is irrelevant). In case you're curious, install anything pitch shifting from Serato (for instance "Sample", it's got a 2 week demo and it sure is impressive as far as realtime polyphonic pitch shifting is concerned).

 

 

[guitarno]

6 hours ago, SaschaFranck said:

 

Here's a sound example. 2 bars of an Fmaj7 chord, no pitching, 2 bars with PP on but at 0, 2 bars with PP on, transposed down by 2 semitones. Even with no transposition, just with PP switched on, the sound is downright unuseable. PP positioned as the first thing in the signal chain, medium picking strength. Guitar used is a Tom Anderson but it's like that (or worse) with each and every guitar I checked with.

HX_PolyPitch_01.mp3 393.56 kB · 23 downloads

I hope you're being sarcastic. That sounds pretty good to me throughout the whole clip. Yeah, it's not perfect, but unusable? I think that's a bit of an extreme judgement! It does have some coloration but for a live performance, it's plenty usable. 

 

[qwerty42]

59 minutes ago, SaschaFranck said:

 

No. That way, polyphonic pitch shifting would only work on very limited sources. But it actually works on everything you throw at it (which is also why tuning is irrelevant). In case you're curious, install anything pitch shifting from Serato (for instance "Sample", it's got a 2 week demo and it sure is impressive as far as realtime polyphonic pitch shifting is concerned).

 

 

Tuning might be irrelevant based on its implementation, but it still has to do pitch detection to isolate the discrete frequencies it is shifting by the desired intervals. It's far from trivial to do so.

Also, have you tried the Poly Capo instead of the Poly Pitch block? I have only tried the former but it has an adjustable stability vs. speed setting. Might give you better results, or maybe it works exactly the same as the Poly Pitch. I don't know, just a thought.

[SaschaFranck]

34 minutes ago, guitarno said:

I hope you're being sarcastic. That sounds pretty good to me throughout the whole clip.

 

Each to their own, but no, I'm not sarcastic.

 

18 minutes ago, qwerty42 said:

Tuning might be irrelevant based on its implementation, but it still has to do pitch detection to isolate the discrete frequencies it is shifting by the desired intervals.

 

No, there's no pitch detection happening. That'd be impossible in realtime on polyphonic material, at least given these day's technology and CPU resources. Have a look at Melodyne DNA, which actually *is* detecting individual pitches in polyphonic material. It's far from being a realtime process.

The only thing sort of pitch related in the entire process is harmonics. In general, the more harmonics, the worse the outcome. You can check that for yourself, the results are getting noticeably better in case you're slapping a rigid low cut in front of the pitch block. Quite obviously, that pretty much defeats its main purpose as you'll end up with a pretty muffled sound. And btw, I think this is the very reason some people consider the Drop pedal to be rather muffled. All these polyphonic algorithms work a lot better with less harmonic content present.

It's also the reason why things get more problematic the more complexed your chords are. Complexed chords create even more complexed harmonics. And these are the ones throwing off most pitch shifters easily.

 

Fwiw, there's actually some tools being pretty great at realtime pitch detection, namely something like Jam Origins MIDI Guitar converter, which works amazingly great. But then, I'd take any bet that there's quite some overtone (and maybe lowcut) filtering happening behind the scenes so it doesn't have to deal with any frequencies not relevant for the conversion process - but these are the very frequencies you want to keep when using polyphonic pitch shifting, hence any such filtering would not exactly be helpful.

 

18 minutes ago, qwerty42 said:

Also, have you tried the Poly Capo instead of the Poly Pitch block?

 

Of course. Seems to be pretty much exactly the same algorithm, just offering a different parameter set.

 

 

[donkelley]

1 hour ago, SaschaFranck said:

 

No. That way, polyphonic pitch shifting would only work on very limited sources. But it actually works on everything you throw at it (which is also why tuning is irrelevant). In case you're curious, install anything pitch shifting from Serato (for instance "Sample", it's got a 2 week demo and it sure is impressive as far as realtime polyphonic pitch shifting is concerned).

 

 

That sounds logical...and of course pitch bends, vibrato, whammy effects work which backs up your points.  However I have also found that having my guitar tuned accurately to A440 makes it work with less artifacts in some systems, and I'm not talking about harmonists here, I'm just talking poly shifting systems.  However what those given systems are doing behind the scenes I do not know - it seems plausible to me that, based on my findings of tuning to concert pitch, those are examples that work as my former coworker explained. 

 

Either way, as I mentioned, I'm basing that description with the caveat that it's old 3rd party information from a guy who was a lead developer at digitech for their pitch based products.  My conversations with him were roughly a dozen years ago at that, and even then he hadn't worked at digitech for many years.  So both his info, and my own memory, are very old :-)

 

 

[SaschaFranck]

Just now, donkelley said:

However I have also found that having my guitar tuned accurately to A440 makes it work with less artifacts in some systems, and I'm not talking about harmonists here, I'm just talking poly shifting systems.

 

I really don't think it's got much to do with 440 or whatever you use as concert pitch - I think it's much more about very accurate tuning which will cause less critical overtones (but then, the tempered system is causing all sorts of overtone havoc anyway).

Whatever, should be easy to test in a sequencer, using the finetuning option of whatever synth. Perhaps I'll give that a try later on.

[donkelley]

3 hours ago, SaschaFranck said:

 

I really don't think it's got much to do with 440 or whatever you use as concert pitch - I think it's much more about very accurate tuning which will cause less critical overtones (but then, the tempered system is causing all sorts of overtone havoc anyway).

Whatever, should be easy to test in a sequencer, using the finetuning option of whatever synth. Perhaps I'll give that a try later on.

That makes a lot of sense.  Just tuning your strings super accurately relative to each other, not whether it's at 440 or not.

[cruisinon2]

Intonation, gents... intonation. You can tune up with the best strobe tuner in the known universe all day long, and if the guitar's intonation is not spot on, it will wreak havoc with pitch shifting effects.

[donkelley]

2 hours ago, cruisinon2 said:

Intonation, gents... intonation. You can tune up with the best strobe tuner in the known universe all day long, and if the guitar's intonation is not spot on, it will wreak havoc with pitch shifting effects.

Totally

[emagli]

18 hours ago, DunedinDragon said:

Especially the Auto EQ parameter.  That was what was initially affecting my opinion of it sounding artificial.  But like you I'm not moving the capo by a huge amount.

 

Totally agreed. The auto EQ yields a very strange attack. I also turn it off, and use a tilt eq to compensate high end loss if necessary.