Geek question to L6 DSP engineers: Do any of the EQ block types use FIR algorithms? any chance of simple zero-phase filters in future ?

[d0stenning]

Digital EQs that model analog circuit designs have phase shifts that vary with the frequency.  Some types of digital DSP filter exhibit zero phase shift so add less coloration or alteration to the input signal. Do any of the EQ types do this ? 

Could it be done in the future - say just for simple "single-pole low Q" type high-end  or low-end rolloff etc for example ? 

 

This is something that a real ANALOG pedal could never do. 

[cruisinon2]

They won't even tell you the color of the tile in the crappers at corporate HQ...L6 generally considers just about everything "proprietary info". You've got better odds of being struck by lightning in a subterranean cavern, than getting them to divulge the "good stuff".

[silverhead]

They won't even tell you the color of the tile in the crappers at corporate HQ........

Been there. Red and black, of course.

[cruisinon2]

Been there. Red and black, of course.

Now if you told me they've got red kidney bean shaped urinals, then I'd be impressed...;)

[silverhead]

Now if you told me they've got red kidney bean shaped urinals, then I'd be impressed...;)

I think they have that at Fractal......

 

(Sorry for the hopefully humorous interlude. Not wanting to detract from the OP's legitimate question, but potty jokes can make you feel 12 yrs old again.)

[cruisinon2]

I think they have that at Fractal......

 

(Sorry for the hopefully humorous interlude. Not wanting to detract from the OP's legitimate question, but potty jokes can make you feel 12 yrs old again.)

Lol...that would make sense

[d0stenning]

They won't even tell you the color of the tile in the crappers at corporate HQ...L6 generally considers just about everything "proprietary info". You've got better odds of being struck by lightning in a subterranean cavern, than getting them to divulge the "good stuff".

 There's absolutely NOTHING repeat NOTHING  proprietary about the subject of this OP.  The principles and algorithms behind the kinds of DSP EQ used out there are widely available and taught in every university's electronic or DSP related engineering courses. 

 

Its merely about the types of EQ's available. To give you an idea  one can implement such EQ's loosely speaking in two ways: 

 

1)  emulate a circuit component by component by modelling an electronic circuit that already exists - usually applicable for old gear that is loved and sounds great. 

 

2) use pure DSP algorithms to implement the EQ directly. Often in ways one could never do it by analog circuitry. The different ways of implementing such digital filters have been around since the concept of DSP was first out there - say from the late 70s onwards.    

 

There's absolutely nothing "secret" about L6 or anyone els deciding to use any of the latter techniques to implement EQ or high/low pass filtering.

[cruisinon2]

There's absolutely NOTHING repeat NOTHING proprietary about the subject of this OP. The principles and algorithms behind the kinds of DSP EQ used out there are widely available and taught in every university's electronic or DSP related engineering courses.

 

Its merely about the types of EQ's available. To give you an idea one can implement such EQ's loosely speaking in two ways:

 

1) emulate a circuit component by component by modelling an electronic circuit that already exists - usually applicable for old gear that is loved and sounds great.

 

2) use pure DSP algorithms to implement the EQ directly. Often in ways one could never do it by analog circuitry. The different ways of implementing such digital filters have been around since the concept of DSP was first out there - say from the late 70s onwards.

 

There's absolutely nothing "secret" about L6 or anyone els deciding to use any of the latter techniques to implement EQ or high/low pass filtering.

I'm prepared to accept all that as Gospel. Since all of this is Greek to me anyway, I'm really in no position to refute anything you've said...however, that's not really the point.

 

L6 has a long and glorious history of refusing to tell you what time it is, never mind discussing how they actually go about doing what they do. Whether or not something is actually "proprietary" or not, is basically irrelevant. They're simply not going to tell you.

 

This forum is replete with examples of official refusals to discuss any number of things that could hardly be considered state secrets, yet tight-lipped they remain, usually blaming it on lawyers. My personal favorite is the refusal to divulge the recommended pickup height for the Variax guitars. Basic operating instructions so that a customer can get optimal results from a product...denied! Awesome. 👠Not exactly 007 kinda sh*t, but confidential it remains nonetheless.

 

This post will self-destruct in five minutes...;)

[ricstudioc]

They won't even tell you the color of the tile in the crappers at corporate HQ...

I'd assume they're assignable per user......

[Lachdanan0121]

Hopefully one day humanity will realize the barriers, limitations, and waste that comes with proprietary. Although it prolly won't be anytime soon, I mean look at how many stll think they need to rely on superstition. Sorry for my mini-rant, I am quite sore on the subject of proprietary. Carry on.

[d0stenning]

<Hopefully one day humanity will realize the barriers, limitations, and waste that comes with proprietary.> 

 

As an independent developer I have no problem with "proprietary" and such utopian idealisms might suit communist angels but not us earthly flawed human folk. 

 

But nothing in this question touches upon anything of any originality that would be proprietary. Its more akin to asking "do you have effect model X" ?

[spikey]

Been there. Red and black, of course.

 

LOL!

 

Wait.... Before or after you used them?

[cruisinon2]

LOL!

 

Wait.... Before or after you used them?

What was for lunch?

[Lachdanan0121]

<As an independent developer I have no problem with "proprietary" and such utopian idealisms might suit communist angels but not us earthly flawed human folk. 

 

But nothing in this question touches upon anything of any originality that would be proprietary. Its more akin to asking "do you have effect model X" ?>

 

I was responding to the others mentioning proprietary.

 

Nice, on the typical rhetoric of shock words for convolution in your first statement.  :rolleyes:

 

Moving on. However, I do hope you find the information you are looking for. I didn't mean to derail...

[jeremyn]

The most significant drawback to zero/constant phase (symmetrical) filters when it comes to live audio is that they introduce additional latency into the signal chain, proportional to the length of the filter. Longer filters with detailed low end frequency / full bandwidth response introduce 5 to 20ms of delay - which is huge, especially when you add it to all the other non-avoidable latencies in a digital signal chain. Super short (low latency) zero phase FIRs have low frequency bands that are too wide for a general purpose audio EQ. So basically you have to live with the extra delay in the signal if you want good low frequency control.

 

Also, symmetrical filters introduce a 'ping' or pre-delay to the signal that can sound a bit unnatural - but I suppose this might be a benefit for someone trying to go for something 'different'. However, FIRs don't have to be zero phase / symmetrical, they can implement any type of filter response with linear or non-linear properties (ie. they can emulate a completely analog filter with close to identical properties including near zero latency).

 

A linear phase (ie. minimal phase distortion) filter isn't exclusive to the digital domain, it's just much easier (less effort and a lot less componentry) to make high order filters with DSP (and be able to change them on the fly). A good analog Bessel filter can be made with a constant group delay and minimal phase distortion if that's what is desired - and it'll have extra latency, just like the FIR - but, will take a lot more space, parts, and still require tuning during production to compensate for component tolerances.

 

Likewise, just as an FIR can be used for zero phase symmetrical filters, they can also be used to implement typical analog filters including minimum phase, or any of the standard models where frequency response is more important than phase response or group delay.

 

A skilled DSP engineer has enormous control over filter design and implementation. The hardest part is selecting the appropriate characteristics, which is a bit of an art when pertaining to the human perception of 'tone'. What may seem ideal in a text book sense, may not have a particularly desirable sound over and above a more common less ideal approach.

 

(Edit) Obviously there are specific uses for zero phase filters outside of live audio general purpose EQ, such as the decimation and reconstruction filters that are used for sigma delta A-D/D-A converters and sample rate conversion. And where latency is not an issue, they allow very detailed (and easy) frequency control using FFTs. Also, an FIR can be used to implement very complex non zero-phase filters such as the impulse response block in the Helix.

[d0stenning]

thanks for that  jnysen.  

 

Pretty much answers the question as i see it. 

[jeremyn]

Also, if you really wanted to, you could load a symmetric filter kernel as an impulse response to achieve a zero phase filter with latency equal to half the filter length (approx 10ms for the 1024 point IR, and 21ms for 2048 points). But, it would be a fixed filter and you wouldn't be able to change it on the fly.

[mdmayfield]

I'm glad jnysen mentioned something about this above - there's nothing inherently bad about the phase shift that happens in a traditional IIR / minimum phase EQ, and nothing necessarily always good about avoiding phase shift by using an FIR / linear phase EQ.

 

One of my favorite sites to talk about audio stuff is HydrogenAudio, because as part of their Terms Of Service any claims about subjective audio quality must be backed up by double-blind test data. We all tend to hear not with our ears, but with our eyes and expectations - even when we're trying very hard to be objective.

 

A nice quote from an acoustics expert in a thread on there which is appropriate here:

 

To me it's a no-brainer to never use a linear phase filter. As far as I can tell, linear phase equalizers were developed in response to the "phase shift is damaging" nonsense spewed endlessly by misinformed audio magazine writers. The myth that phase shift is a problem was repeated so many times over the years that audio gear makers came up with a way to avoid phase shift, just to use as a marketing claim. The problem is [Linear Phase] filters are audibly worse than minimum phase! You can't always hear the pre-ringing, but often you can depending on the frequency and other settings.

 

 

That's from here: https://hydrogenaud.io/index.php/topic,103316.0.html

 

[d0stenning]

thanks for all that  mdmayfield - my days as an audiophile are kind of over - both in terms of the kinds of hi-fi I own these days and in my day to day interests. 

 

But I do retain a geekish interest in these things.  Plus..... I am an organiser for talks in a local Skeptics In The Pub  and double-blind tests and other scientific methodologies etc ( repeatability etc )are of great interest and often crop up in guest speaker talks and discourse.    

[arislaf]

I'm prepared to accept all that as Gospel. Since all of this is Greek to me anyway, I'm really in no position to refute anything you've said...however, that's not really the point.

 

L6 has a long and glorious history of refusing to tell you what time it is, never mind discussing how they actually go about doing what they do. Whether or not something is actually "proprietary" or not, is basically irrelevant.

 

This forum is replete with examples of official refusals to discuss any number of things that could hardly be considered state secrets, yet tight-lipped they remain, usually blaming it on lawyers. My personal favorite is the refusal to divulge the recommended pickup height for the Variax guitars. Basic operating instructions so that a customer can get optimal results from a product...denied! Awesome. Not exactly 007 kinda sh*t, but confidential it remains nonetheless.

 

This post will self-destruct in five minutes... ;)

Don't worry I am Greek and still don't get it :P

[davidb7170]

Good one! 😎

[psarkissian]

"There's absolutely NOTHING repeat NOTHING  proprietary about the subject of this OP.  The principles and algorithms behind the kinds of DSP EQ used out there are widely available and taught in every university's electronic or DSP related engineering courses"---

 

... and Julius Orion Smith III teaches DSP of digital waveguides, Karplus/Strong and modeling at Stanford. And has done some cool papers for the AES journal (JAES). my pal Josh Reiss from QMUL also teaches this stuff about digital audio. What they teach isn't proprietary, but what Line 6 does with it,... is.    ---Even I don't know how they do it.

 

A filter with zero phase with less coloration, Smith's book has a good chapter on stuff like that (Bessel or Allpass filters), and my pal Josh Reiss from QMUL has a book on audio effects where he gets into Allpass filters. Another colleague, Stanley Lipschitz from Uni of Waterloo, ON, Canada, also did a cool paper on Allpass filters (a more analog approach). All cool stuff.

 

When you try this yourself, remember to keep the filter poles and zeros inside the unit circle,

and keep the reins tight on the gain stages, so it won't self oscillate. A general rule of thumb

when it comes to filters (analog or digital).

 

Might want to do an Idea Scale request entry for the "EQ block" you're looking for.