UltraRes IRs for Torpedo products?

[dapeegoo]

Now that the Axe-FX are capable of using the UltraRes IRs, is there a way for Torpedo products to be compatible with those longer wav files?

 

[richardt4520]

Same question here. I wouldn't mind less memory space if it brings more detail.

 

[Rogue]

Not sure if the two notes will play these ultra res IRs or not.

I've thought the Two notes IRs have sounded better than standard impulse responses. From what I've heard of the ultra res IRs, they are slightly better than the normal ones.

I'd like to compare them them to the two notes stock cabs, but I'm not getting the feeling the are going to be better.

I'm just looking looking forward to some new two notes cabs.

 

[guillaume_pille]

Just for my personal edification: what kind of benefit are you aiming at with the so called "UltraRes IRs"?

You can do the experiment with any free IR on the market, with a free plugin which lets you tweak the length of the IR.

By the way, the Remote will automatically conform any kind of IR to the Torpedo format.

 

[guillaume_pille]

OK apparently it's an another Axe FX concept, well this is what I found interesting:

http://forum.fractalaudio.com/axe-fx-ii-recordings/81387-ultrares-vs-normalres.html

Basically, longer IR will bring you 2 things:

- more room effect
- better resolution

What exactly is better resolution? Well, it's not easy to define it (it depends on the sampling frequency and the size of the IR in number of points) and, more importantly, there is NO PROOF of any audio benefit of a higher resolution.

I made this answer to a customer once, I think it can be interesting for you too:

The impact of the IR length can be seen on the frequency resolution. This resolution is important mainly in the low frequencies, because having 10Hz resolution from 10KhZ to 48Khz is totally useless.

We at Two Notes decided that a 50Hz resolution was enough to accurately reproduce a speaker sound, after extensive tests with users. Actually, increasing the resolution is not directly linked to the sound quality.

Please read this article that explains a little more what this is about, the Two Notes products are the last row of the table at the end of the article (frequency 96KHz, 2048 coefficients

http://www.four-audio.com/en/techni...pulse-responses-and-frequency-resolution.html

We could easily improve resolution by choosing a lower frequency response, but as we are dealing with distorted signals more than often, I prefer to save the harmonic content rather than increasing resolution. That was again confirmed by tests.

There are other techniques using multi rates systems to increase resolution in low frequency, but frankly our tests didn’t lead to significant results. That kind of technique may have a marketing interest though (complicated names = better quality, as everybody knows ).

IR length has not only an effect on resolution, but also on the reverb quantity in the sound. Actually, the longer is the IR, the longer will be the reverb.

We decided that the studio/room reverb can be easily added to the sound by an external processor and focus on the cabinet sound, rather than having an important reverb impossible to get rid of

So, my conclusion:

Is it possible that some can be interested in having more room: YES

Is the increasing the resolution leading to an audio benefit thing supported by either audio tests or mathematical approach: NO

Is it possible to play those IRs in the Torpedo products: YES but they will be truncated

is it possible to play them at full length in a Torpedo Product : NO at this time, and if that was something to follow we can do it but there is 100% chances the latency will increase accordingly.

Always remember that you should be alerted when some TV/image marketing concepts are introduced in the audio world (HD, High-Res and so on).

Ear and eye does not work the same way AT ALL. The HD system in TV was introduced to bring better details FOR BIG SCREENS BUSINESS. The benefit of 4K definition for small sized-TV is really hard to find, and the benefits people find in small screens are usually side effect of the progress in the technology (refresh rate, vision angle etc...).

 

[guillaume_pille]

Oh, and the most important :

some people will prefer different length on the IRs. Some will prefer longer IRs, some shorter IRs. The tests on the Fractal forum is a proof of that, and we made the same tests 7 years ago.

What you can't prove is that the HD system is simply better. Saying that is simply and purely marketing.

 

[Raven]

Personally, I'd rather add my reverb & room sounds separately using dedicated software than having to deal with having an excess of it built into the speaker/mic IRs. I'm with you on this Guillaume, one thing to say about Fractal is that they're very good at marketing & hype, I've never found any of their statements to hold any water tonally though, it always just sounds like a modeller to me regardless of whatever new-fangled terminology they want to routinely work into the platform.

I'd really pay this 'UltraRes' thing no heed, their old speaker simulation was terrible, this might be marginally better but I don't think it'll come close to what we already have in the Torpedo line of products. ...just my two penneth, no offence intended to any Axe-FX users (each to their own .

 

[guillaume_pille]

I'd like to add a little something:

I am not saying that having the possibility to change the IR length is not interesting. You can try that with our Torpedo BlendIR or many free convolution plugins, and it's an interesting parameter to play with.

But fixing a length and calling it HD. Come on... BTW I'm a big fan of the Axe FX marketing and I already told it to Cliff, but that is not how I want to market the Torpedo products.

 

[Rogue]

I thought the older one sounded slightly better than the new one in that clip. So did most people, apparently. Others I thought the newer one sounded slightly better.

The responses in that thread make me laugh. But it's AxeFX people, and they are a unique breed.

I'm not sure what an "ultra res" is, but from the clips I have heard, the difference is very slight and I don't think will make up the difference I get from a regular IR and what I get from the TN stuff. That said, some people prefer 3rd party IRs over the TN stock ones. To each his own, I reckon.

 

[Rogue]

But fixing a length and calling it HD. Come on...

Is that it? Is that what an "ultra res" is? Just a longer IR? Many moons ago, threads would go on and on for pages and pages on the Fractal forum about how the length of the IR used in the AxeFX was perfectly adequate. Times have changed.

However, Cliff could have put out a new IR that was made in a room someone farted in and call it something cool and those guys would feign all over it. It's what they do.

 

[Raven]

BTW I'm a big fan of the Axe FX marketing and I already told it to Cliff, but that is not how I want to market the Torpedo products.

I guess you have to respect their success; however, the Two Notes way of honesty & respect to your customers is much more appealing to me.

...Plus, I like tubes!

 

[richardt4520]

The more I looked around on the subject after posting, it's starting to look like marketing hype to me, anyway. The differences can be construed as the way the output signal was measured, from what I'm seeing. Thanks for the clarification, Guillaume!

 

[guillaume_pille]

Well there are a few parameters when recording IRs:

- during the measurement: the type of stimulus (test signal) you use to do the measurement, the bandwidth you're working with

- after the measurement: the exact point when the IR starts, it's length... You can play with these parameters in BlendIR

Well, there's not that much more you can do with IRs....

 

[Jay Mitchell]

I've done some instrumented testing of "Ultrares" IRs in the Axe-Fx II. I posted about the results here: https://www.thegearpage.net/board/showpo ... stcount=14 . Here is a quote from that pose:

"The data in an "Ultrares" IR contains less information than the original IR from which it was derived.....To produce an Ultrares IR, the tail beyond approximately 20ms is decimated: lowpass filtered, then downsampled. Based on the frequency content of this tail - it cuts off very sharply at 2500 Hz - the decimation ratio is 8:1. IOW, the sampling rate used by the algorithm that processes the tail is 6kHz rather than the system rate of 48kHz. The demand on CPU goes down approximately as the square of the decimation ratio. This process has been in use for decades, for exactly this purpose (reducing cpu demand). Here's a page with an overview of decimation: http://www.dspguru.com/dsp/faqs/multirate/decimation.

 

[guillaume_pille]

Thank you Jay for this explanation.

Well, it looks a little like the multi-rate system I wrote about earlier. Thus, apart from getting some more room by having a longer IR (the IR is longer, right?) the whole resolution thing is not IMHO the most interesting path. And a 2,5KHz low pass filter on the tail? Well.. Can't really have an opinion on that.

Anyway, I have to correct what I wrote previously: it's not possible to load that in the Torpedo products. Well, you can, but it won't work as expected.

Actually it could work BETTER!

(just kidding, it's a subjective matter)

Just to be perfectly understood: I'm not the one who said it's bad to try and offer alternative options and progress of any kind (even if it's mainly marketing) in our field of expertise. Competition is good and makes us improve the products!

 

[Jay Mitchell]

Thank you Jay for this explanation.

You are welcome. It was clear from Cliff's original description here: http://forum.fractalaudio.com/axe-fx-ii ... rking.html that he was contemplating a decimation scheme. I saw exactly what I was expecting to see when I acquired IRs of "Ultrares" cabs in my Axe-Fx II.

Well, it looks a little like the multi-rate system I wrote about earlier. Thus, apart from getting some more room by having a longer IR (the IR is longer, right?)

It appears to be around 200ms total, with the first 20ms being full range and everything after that being lowpassed at 2.5kHz.

the whole resolution thing is not IMHO the most interesting path.

Agreed. Regarding resolution in simulation of guitar cabs, I came to the same conclusion you did many years ago. You need an IR about 20ms long to capture a convincing representation of a cab. Longer than that does not help, and any resulting inclusion of room reflections will make matters worse, not better.

And a 2,5KHz low pass filter on the tail? Well.. Can't really have an opinion on that.

I've done some A/B listening to a long IR that includes source and environmental effects with and without the lowpass filtering of the tail. The difference is not subtle and IMO does not favor the "Ultrares" concept. I think it's important to define your goal with an IR-processing device. If you want to do convolution reverb, 200ms is more than an order of magnitude too short. If you want to do simulations of guitar cabs, it's about an order of magnitude too long.

I'm not the one who said it's bad to try and offer alternative options and progress of any kind (even if it's mainly marketing) in our field of expertise.

My only reason for speaking up is that Cliff originally offered a technical argument in favor of a need for longer IRs and that his argument is mistaken. If it had been presented as yet another effect, there's really nothing to argue....

 

[guillaume_pille]

I may have to read it another time, but... Jay... did we just agree on something??

Anyway, talking about the length of IRs, it's quite easy for anybody to do their own tests and decide what length is good enough. Well, doing the test is easy, but deciding is not that easy. I agree with you somewhere around 20ms the first reflexions of the room tends to arrive and add something one might consider as unnecessary perturbations. But I guess it is a matter of taste.

An interesting thing is to shorten the IR to the point it sounds like a Pod speaker simulation.

 

[Jay Mitchell]

Anyway, talking about the length of IRs, it's quite easy for anybody to do their own tests and decide what length is good enough.

Perhaps not. In order to make an informed decision, you must be able to change the length of the IR as the only variable. If you do not have an appropriate test space, the limitation on maximum length will be the point at which the first reflection appears. Once the IR is that long, you are no longer listening to just the cab.

Well, doing the test is easy, but deciding is not that easy.

Based on my prior knowledge of the limitations imposed by human hearing capacity and many years of testing and listening to sound systems, I had a very good idea of the minimum length that would be required for an IR to satisfactorily represent the sound of a guitar cab. My original hypothesis was borne out by experiments, and I was not surprised by anything I heard during the decision-making process.

I agree with you somewhere around 20ms the first reflexions of the room tends to arrive

Fortunately, I have not been subjected to that limitation. In my facility, I can acquire IRs longer than 20ms with no room reflections in them. I still conclude that they need be no longer than approximately 20ms to faithfully represent a guitar cab. That does not mean that the process of achieving faithful representation is easy, however. It simply means that the resulting IRs need be no longer than 20ms.

An interesting thing is to shorten the IR to the point it sounds like a Pod speaker simulation.

I may have to try that at some point. So far, I've tended to experiment by going in the opposite direction.

 

[Raven]

I may have to read it another time, but... Jay... did we just agree on something??

Anyway, talking about the length of IRs, it's quite easy for anybody to do their own tests and decide what length is good enough. Well, doing the test is easy, but deciding is not that easy. I agree with you somewhere around 20ms the first reflexions of the room tends to arrive and add something one might consider as unnecessary perturbations. But I guess it is a matter of taste.

An interesting thing is to shorten the IR to the point it sounds like a Pod speaker simulation.

Well, however you guys are doing it at the moment seems to work wonders; you know what they say, if it ain't broke...

 

[guillaume_pille]

Perhaps not. In order to make an informed decision, you must be able to change the length of the IR as the only variable. If you do not have an appropriate test space, the limitation on maximum length will be the point at which the first reflection appears. Once the IR is that long, you are no longer listening to just the cab.

Absolutely, and that is why I was talking about a matter of taste. If someone likes his coffee with too much water for my taste, well, I won't try to convince him he's wrong. Unless his trying to lecture me using market-matical BS. ^^ And even then, you can't question his taste.