IR 500ms vs 18ms?

samtheman

Member
From another forum.... discuss :confused:


"I don't know that app so I wouldn't be the right guy to ask. I'm going to burn a bridge here and advise you that the IR player plug-in makes a big difference in the sound of the IR. Pete Thorn told me about mixIR 2, made by Redwirez. It costs 50 bucks, but it's way worth it.

Most IR players (especially the free ones like Wall Of Sound III and NadIR) truncate the sample. Many of the best IR's are 500 ms, so when the player reduces it down to 18 ms like Wall Of Sound does, you loose bass and and overall sound quality.

I never would've believed this, but I heard it with my own ears, and the difference isn't subtile at all. So now the Two Notes guys can hate me, but they need to re-design their software to play longer samples."
 
That's a very interesting topic, actually it's not that obvious to hear the differences and to figure out what exactly you hear.

The length will play on several parameters:

- the reverb length (at 18ms we deliberately get rid of the reverb, only the early reflexion stay)
- the cutoff frequency (at 18ms it's around 50Hz)
- the resolution (that parameter is kinda hard to handle but I'll explain later), resolution is also around 50Hz

Increasing the length will improve both the resolution and lower the cutoff frequency, so the questions are mainly:

1- what happens between 20 and 50 and is there any guitar sound here, and/or do we want it?
2- Is there so many accidents in the lowest part of a cabinet spectrum to need a better resolution(improving resolution can show very abrupt holes in the spectrum)? And if there are, can I hear them?

In our studies, the answer to both questions, after extensive blind tests, were no. Especially in a studio environment, where most sound engineer will get rid of the mess that is a mix of low end resonances between the cabinet and the ground and the room, where it'll be hard to struck a guitar of bass note that low.

Moreover, it's not like all frequencies below 50Hz disappear, they are just an approximation (it's a first order low cut).

But, I don't know, sound monitoring systems make progress, and the need of the guitarist evolve, so why not provide an improved version?

What about asking the gentleman you quote to come here and provide an obvious example of what he claims, and I actually can share a little blind test we did for our friends at Celestion? I won't certainly hate anyone for helping us improve our products. ;)
 
This is BS ("the best IR's are 500 ms")!

18ms +- is more than enough time for the filter to define the cab's/speaker's/mic's frequency response. What 18ms doesn't do is add the room tone which is actually the preferred method for most who then use a room sim to create the room tone independently.

The reason that's preferred vs having the room tone baked into the IR is that long IR's don't allow you to adjust the room tone except via a logarithmic increase in the number of IR's you have to manage (not to mention the CPU hit the extra IR length causes as well).

With a separate IR and room sim such as is provided by Two Notes, you can pick your speaker/cab/mic and room independently, and therefore it's much faster and more efficient than dealing with a multitude of different IR's with different rooms and mixes etc.


samtheman":aj2a965t said:
From another forum.... discuss :confused:


"I don't know that app so I wouldn't be the right guy to ask. I'm going to burn a bridge here and advise you that the IR player plug-in makes a big difference in the sound of the IR. Pete Thorn told me about mixIR 2, made by Redwirez. It costs 50 bucks, but it's way worth it.

Most IR players (especially the free ones like Wall Of Sound III and NadIR) truncate the sample. Many of the best IR's are 500 ms, so when the player reduces it down to 18 ms like Wall Of Sound does, you loose bass and and overall sound quality.

I never would've believed this, but I heard it with my own ears, and the difference isn't subtile at all. So now the Two Notes guys can hate me, but they need to re-design their software to play longer samples."
 
Fun fact: if the 500ms IR actually has very little information in the 20-50Hz, which can happen with a small sized cabinet, the 18ms approximation in that area may give MORE bass.

Do we know what it the length of the IRs in the Kemper, I feel like nobody bust their balls with that? Pete Thorn is using one, he must have asked them?
 
Here is some testing regarding the KPA cab sims...

http://dallashodgson.info/articles/KemperCabAnalysis/whitepaper.htm

guillaume_pille":o2g2pda9 said:
Fun fact: if the 500ms IR actually has very little information in the 20-50Hz, which can happen with a small sized cabinet, the 18ms approximation in that area may give MORE bass.

Do we know what it the length of the IRs in the Kemper, I feel like nobody bust their balls with that? Pete Thorn is using one, he must have asked them?
 
I did some testing yesterday. Close mic'd it's not night and day, but the 500 ms do sound fatter with more bottom end. Room mics is really night and day. With 18 ms you really don't get the much of the room sound.

I always thought the the VB-101/WOS sounded a little strange with mics set at max distance ... I love my Two notes gear and I think allowing for 500 ms ir's would be a great upgrade.

Is this possible to do with a firmware/software update?
 
Hi SunKing,

what tool did you use to do your test? What 500ms file did you use and how did you truncate it?

With 18 ms you really don't get the much of the room sound.

You definitely get the early reflexions, not the reverb itself, again, on purpose.

Is this possible to do with a firmware/software update?

That would mean adding a LOT of latency and/or lowering the sampling frequency. On WoS it's no big deal, your CPU does the job and if it's powerful enough you can still get a low latency, but the hardware units have some limitations due to the onboard DSP.
 
I have also made a lot of test with different ir loaders, both hardware and software, and the once that has a longer tail sounds more full and more natural to me as well. I red that you put a limit to 18ms in the live unit to get latency down but the logidy has stereo cab ir at a latency of 1,5 ms and can play irs up to 1,5 seconds. And when I try the same impulses in both the live unit and the logidy (that has a longer tail) the logidy reacts smoother, more natural and with larger headroom. The live is a bit brighter and when now that I have started to make my own impulses with tone matching software the live doesn't really get the same tone out of them that the ir loaders in the daw does. The vst ir loaders (kefir, reverberate etc) seems to get the closest to the matched tones. After spending 2 years with the live unit I have noticed that it probably is designed more for your software cabs (in terms of eq and how it recreates a cab) than third party ir. Some 3:d party ir sounds amazing but alot of the times it gets a bit fizzy and dry for my taste. So I also would like to have the option to use longer IR. Cuz there is definitely a difference in sound and feel.
 
Hi Guillaume!

what tool did you use to do your test? What 500ms file did you use and how did you truncate it?
I used Celestion ir's, 4x12 Greenback (57 balanced) + Room L. MixIR2 and WOS in ProTools. Same 500 ms for both. MixIR2 ads no latency while WOS ads a small amount according to ProTool.

you definitely get the early reflexions, not the reverb itself, again, on purpose.
I understand this was on purpose, but getting the actual sound of the room mics is preferable in my opinion.
 
Oh so you didn't use exclusively MIXiR2 by truncating the IR? You compare two different softwares so it's possible some difference happen that have nothing to do with the IR length.

MixIR2 ads no latency while WOS ads a small amount according to ProTool.

Worst case is 1 sample latency added, but run your session at 96KHz and it should be zero, it's the native frequency for WoS.

I'm uploading a blind test on Soundcloud for those who are interested, and it was done using one software for all cases (our factory IR loader, different from WoS) except 1 file is done with WoS. The source IR is a free one offered by Celestion, and I provide the audio sample if you want to try it on your side.

https://soundcloud.com/twonotesaudioengineering/sets/ir-length-blind-test

I'm very interested in what you hear, I can provide all the data we characterised with our normalized Audio Precision tools.
 
Bigshredder, would you be kind enough to share samples showing the "fizziness", and possibly one IR file you find to be obviously better in your software loader?

What does your setup look like in the case of using the Live or a software? Because obviously not only the DSP will be different, but also the whole hardware path, right?

It's important to make comparison all things being equal. For example, we test our hardware so when we compare the modelling against the real miking the difference is virtually impossible to make. We could also decide to "voice" it to sound warmer, but that's not really what we aim at here.
 
guillaume_pille":3mjgx10y said:
Bigshredder, would you be kind enough to share samples showing the "fizziness", and possibly one IR file you find to be obviously better in your software loader?

What does your setup look like in the case of using the Live or a software? Because obviously not only the DSP will be different, but also the whole hardware path, right?

It's important to make comparison all things being equal. For example, we test our hardware so when we compare the modelling against the real miking the difference is virtually impossible to make. We could also decide to "voice" it to sound warmer, but that's not really what we aim at here.


Yes I am planning to record some "this versus that" audio or video clip and upload but I never get around to. I must try to find some time to do it in a few days.
When I compare to the software ir-loader I just bypass the mic and eq in the live and use it just as an loadbox into my interface (rme babyface pro). So of course the ad/da conversation and preamp (thou very uncolored preamps in that interface) is gonna be before the ir as supposed to after. Same as for the logidy. One IR that I love to use is the old Brohymn mesa IR that can be found free on the web. It has an IR length of 2 seconds (logidy can't play those). The live unit truncates them and leaves them a bit harsh and fizzy but kefir or reverberate uses the whole 2 seconds and it gets fat and full with a darker smoother sound.
 
guillaume_pille":2cb7oeuz said:

I listen to the 100ms length compared to the 12ms length and the 100ms has that "swollen" effect that the 12ms lack. I play a mesa dual rec with huge amounts of headroom and that "swollen" effect (more omfh and bass response) is key for me. So I can definitely hear some difference. And the 12ms sounds a tab bit fizzier that the 100ms. That's my opinion:)
 
12 is definitely too short and it's the point where artefacts really start showing. that's why we didn't bother go under, you can buy an old Pod to hear that. ^^

But could you connect the A, B, C to the lengths? Care to share with me your list? ;)
 
Oh so you didn't use exclusively MIXiR2 by truncating the IR? You compare two different softwares so it's possible some difference happen that have nothing to do with the IR length.
I did a new test in 96k with MixIR2 set at 18 ms and 500 ms and WOS. MixIR2 at 18 ms was very close to WOS – I would not be able to pick one over the other in a blind test.
I also loaded the same ir in Avid Space convolution reverb/ir player and both this and MixIR2 at 500 ms do sounds a little fuller, more balanced.
There is also slightly more sustain/air to single notes with 500 ms. All which becomes more evident the further away the mic is placed from the speaker.

I will check out the blind test :)
 
People, you should use a room sim with the shorter IR's to add "room tone" (in addition to reverb and delay etc for maximum realism!).

This is typically the preferred method of working (see my post above for the reasons). The Torpedo H/W offers a room/sim for this purpose, and with WOS you can use one from your DAW. Essentially, any decent reverb set to room with a RT between 250ms - 750ms will get you started, mix to suit.

Try it, you'll find the flexibility and sonic advantages of using a separate room sim vs baking it into a long IR too numerous to count IMO!

Last, sure the Epsi can handle long IR's, however is doesn't have full MIDI control, tube power amp sims, 5 band EQ or better, separate room sim/reverb, a real file system with file naming, IR capture and mix software, software editor, or the WOS plugin like the LIVE/CAB do, and the Studio even provides more DSP as well. The difference between the Two Notes H/W and the Epsi's published RTL latency is only 0.2ms (that's 2/10 of 1/1000 of a second!).
 
People, you should use a room sim with the shorter IR's to add "room tone" (in addition to reverb and delay etc for maximum realism!).
In my mind, it is the other way around, to obtain "maximum realism" one should be able to play/hear the whole ir capture as they were recorded in the studio. At least having the option to do so is an improvement. Not only for 3rd party ir's but all the Two notes cabs too.
The further away you move the mic from the cabinet, less true to the actual sound/environment it becomes.

This has nothing to do with replacing reverb and delay, at least not for me. It's about getting the most balanced tone, and as close to the real thing as possible.
 
It's not about replacing reverb and delay, it's about "room tone".

If you really care that it has to be THAT room that the IR was created in, then use the long IR's, but to have any kind of flexibility you'll need a huge number of long IR's so that you have different room mixes, mic positions etc.

I find it much faster and way more flexible to use a room sim, either the Two Notes one when playing live (for DI work), or a real convolution reverb like TL Space or Altiverb in the studio. This way I can adjust room tone easily, and change rooms as need etc.

The short IR's do contain the room's influence on the mic position (early reflection's influence on the phase relationship and by extension the freq response), just not the tail etc. So, using a Two Notes cab with the mic pulled back along with a room sim essentially gives the you the same thing, except with a different tail.

Room tone is a subtle, but critical element for cab sims, and I've found that it's not exactly critical what you use for the tail, as long as you use something LOL!

Have fun!

SunKing":1syjamct said:
People, you should use a room sim with the shorter IR's to add "room tone" (in addition to reverb and delay etc for maximum realism!).
In my mind, it is the other way around, to obtain "maximum realism" one should be able to play/hear the whole ir capture as they were recorded in the studio. At least having the option to do so is an improvement. Not only for 3rd party ir's but all the Two notes cabs too.
The further away you move the mic from the cabinet, less true to the actual sound/environment it becomes.

This has nothing to do with replacing reverb and delay, at least not for me. It's about getting the most balanced tone, and as close to the real thing as possible.
 
SunKing I agree with you on having the possibility to do it. That is definitely something we can work on for future products.

Still, what about my blind test? ^^
 
Back
Top