Why Are FAS IR's Truncated To 170ms?

B:ASSMASTER

Experienced
Serious question. I've been thinking about this.

Say you have a Dual Rec and a Mesa 4x12 cab, close-mic'd with an SM57, going into an audio interface, recording into Pro Tools. If you strike an open A power chord and let it ring out for a 4-count, the SM57 is going to capture a lot more than 170ms of sound.

So, why doesn't FAS have longer length IR's in their cab packs? Why not 200, 300, 400, 500, ect.?

Maybe I'm missing something and someone can educate me.
 
You "typically" want nothing but the sound of the speaker/cabinet in your IR .... Anything longer than that and you start to get room reflections in the sound too ....
 
Serious question. I've been thinking about this.

Say you have a Dual Rec and a Mesa 4x12 cab, close-mic'd with an SM57, going into an audio interface, recording into Pro Tools. If you strike an open A power chord and let it ring out for a 4-count, the SM57 is going to capture a lot more than 170ms of sound.

So, why doesn't FAS have longer length IR's in their cab packs? Why not 200, 300, 400, 500, ect.?

Maybe I'm missing something and someone can educate me.

The length of time you hold a chord is irrelevant. The impulse response of a speaker cab is typically much less than 100 ms. Only when there is significant room reflections is the length greater. Then you get into the whole argument of whether the IR should contain any room information.
 
The impulse response of a speaker cab is typically much less than 100 ms. Only when there is significant room reflections is the length greater.
So, on a mic'd cab, any sound past 100 ms is just the microphone picking up room reflections?

Sorry if I come across as dense. I'm just trying to understand how IR's work, as opposed to mic'ing up a cab the old-fashioned way.
 
So, on a mic'd cab, any sound past 100 ms is just the microphone picking up room reflections?

Sorry if I come across as dense. I'm just trying to understand how IR's work, as opposed to mic'ing up a cab the old-fashioned way.
Exactly. But note that this is based on sending an impulse to the cab, which you can consider as a sound with an extremely short duration. So, send that very short duration sound to the cab, wait 100ms and what you'll still hear is only room reflections (if any).
 
Think of it this way: You put a very very short, but loud signal into a cabinet. Let's say it's 1ms long. The output of the cabinet won't be just 1ms long, because the room will reflect the signal back to your ears or the mic. But the speaker itself will be done much faster. It won't be completely still after 1ms, but it won't take more than 100ms.
Now think of your normal guitar signal as a line of almost infinite short signals.

And think of an IR like a very fine equalizer with thousands of frequency bands (8192 for UltraRes iirc) that can change over time. If you shout into a room and listen to what the room returns, you'll e.g. lose high and low frequencies and eventually all frequencies are down to -infinity, because there's silence again. That's how an IR "displays" a room. It's the same with a cabinet. You put in e.g. a noise that covers every frequency with the same volume. But the cabinet + mic won't return the exact same noise, it'll be something different. This is reproduced with this very fine equalizer. It can also be done with a frequency sweep (the preferred method).

So you give the room / cabinet an audio impulse and measure its response. The measured data can then be used to create a "snapshot" of the cabinet + mic combination (how the frequency bands behave over time). The math behind impulse responses is another thing, a beast actually. Not important to understand IRs ;)
 
Think of it this way: You put a very very short, but loud signal into a cabinet. Let's say it's 1ms long. The output of the cabinet won't be just 1ms long, because the room will reflect the signal back to your ears or the mic. But the speaker itself will be done much faster. It won't be completely still after 1ms, but it won't take more than 100ms.
Now think of your normal guitar signal as a line of almost infinite short signals.

And think of an IR like a very fine equalizer with thousands of frequency bands (8192 for UltraRes iirc) that can change over time. If you shout into a room and listen to what the room returns, you'll e.g. lose high and low frequencies and eventually all frequencies are down to -infinity, because there's silence again. That's how an IR "displays" a room. It's the same with a cabinet. You put in e.g. a noise that covers every frequency with the same volume. But the cabinet + mic won't return the exact same noise, it'll be something different. This is reproduced with this very fine equalizer. It can also be done with a frequency sweep (the preferred method).

So you give the room / cabinet an audio impulse and measure its response. The measured data can then be used to create a "snapshot" of the cabinet + mic combination (how the frequency bands behave over time). The math behind impulse responses is another thing, a beast actually. Not important to understand IRs ;)
Wow. Thanks for taking the time to explain it very well. I feel like I understand it more than I did before. Lots of heady stuff, but I think I'm catching on. :)
 
If you shout into a room and listen to what the room returns, you'll e.g. lose high and low frequencies and eventually all frequencies are down to -infinity, because there's silence again. That's how an IR "displays" a room. It's the same with a cabinet.
Okay. So, I've been thinking about this. If I use an IR with a length of 500 ms versus an IR of 170 ms, the 500 ms IR will have less high-end?
 
No ;) The echo is just longer!

Let's say the cab + room you captured is completely silent after 170ms. Idealized case! Then the IR which has a length of 170ms captures everything there is and the 500ms IR has 330ms of silence at the end. Therefore it would be a a waste of processing power to use such a long IR in this special case.

But rooms usually don't go silent that fast. Think of a church in where you clap your hands (= the impulse). The room will echo (= the response) to your clap much longer than 170ms, even longer than 500ms. So when you capture such a room and want to put it into a 170ms IR the echo will stop abruptly, which sounds unnatural. But the first 170ms of the 500ms IR sound exactly the same like the 170ms IR. Only the 500ms IR can still playback contents of the room's echo AFTER those 170ms, simply because there's more information stored in this longer IR.

But in the Axe we don't use IRs for rooms, we use it for cabs! Cabs are silent much faster than rooms. We can add the room with either the reverb block or the reverb tab in the cab block.


Now, why is UltraRes great and what problems can it cause:

+ With UltraRes we get quite long IRs, 170ms long. Cliff says that 100ms would be quite enough. What does this mean? It means that with UltraRes IRs you definitely have the whole response of the cabinet you made the IR of. Then in theory there's almost absolutely no difference between using the IR or using the miced cab, e.g. in a studio situation etc. With a shorter IR it's also possible to capture the whole response the cabinet gives. But if it's too short, it's possible that you'll hear a difference, especially in the lower frequencies, because there's more energy in low frequency tones (bigger amplitudes) which takes longer to fade away.

- The longer the IR, the more room you capture! Look at it this way: When you clap your hands in this church, it'll take a small amount of time until the room's echo will reach your ears, because the sound has to go from your hands to the walls, the ceiling or whatever and then back to you ears. When you capture a cab the cab's sound will also not be echoed immediately, it takes a little time until the echo reaches the mic. Because we know the speed of sound to be about 330m/s this can be measured. If e.g. the distance between the speaker to the closest wall (usually the floor) is 20cm and back to the mic it's 30cm (you Americans will have to translate it to your strange units ;)) there's a total distance of 50cm = 0,5m for the sound. With 330m/s this 0,5m distance takes the sound 0,5m/(330m/s) = 0.0015 seconds = 1,5ms. In this case there will be room content from the floor captured in the IR! But a room usually also has 4 walls and 1 ceiling. You can do the math for those if you know the distances.
The perfect case would be to have an IR that ONLY captures the cab, but no room content (or only content from the floor as just calculated). The problem is, if you capture a 170ms long IR in a bad room, the IR will sound shitty, because there's a lot of echo in it which stops abruptly. But if you record it in a good room which doesn't have much echo after 170ms and you fade out this last bit of echo that's still present at 170ms, your IR has the potential to sound awesome!

I've seen (heard) many free IRs with this problem! Even commercial ones. What needs to be done is to put a fade-out to the IR so the room doesn't stop abruptly! But if captured in a bad room with lots of echo it'll still sound really bad, because the fade-out just sounds weird. What else can be done? You can simply make the IR shorter, so there's less or even no room content in it, but then it's possible that the IR doesn't contain all of the response the cab gives.
 
Last edited:
Why are fas irs truncated to 170ms?
Mmmmhhh...let me put this the other way: fas irs are extended to 170 ms hehehe!

You can expand your info with some wiki readings:

https://wiki.fractalaudio.com/axefx...f_IRs_.28Ultra-Res.2C_Hi_Res.2C_Normal_Res.29

"Some cabinets displayed noticeable resonances at low frequencies. Others did not. The frequency of these resonances were not consistent and, not coincidentally, matched the measured resonance of the impedance sweep. It is a logical conclusion, therefore, that the resonance was NOT caused by the room but by the speaker/cabinet combination. Furthermore a plot of the group delay for the raw data showed that the delay of the resonance was too short to be a room mode."

When you're shooting a typical IR with a mic placed in front of the cone there are a chain of events as the impulse spreads, I'm not technical:
1. The coil of the speaker receives an impulse and moves air (it's a 2 way move, the coil moves forward and back), the air moved by the forward movement of the coil spreads from the front of the cone and is the first sound to reach the mic, I call this the laser beam
2. The air moved by the back return of the coil bounces in the cab enclosure and this fuller/darker sound reaches the mic a bit later, I call this the meat
3. As the sound spreads around the room there are complex reflections based on the material, dimensions and angles of the room, but remember that our mic is placed in front and near the cone, this is not the best position for room capturing, to each his own!

In the beggining of the IR development the files were fairly short and only could collect the first momentum.
Ultrares IRs allows to capture point 2: cab reflections rich in low frequencies. That's why we like ultrares: a speaker bolted to a cabinet sounds better than a speaker alone

IRs would have to have info of the third moment?
This is a subject of debate, but I think that rely the room content to an impulse is too much...there have been some interesting attempts but I think that capturing a room (mic far placement analog to the listener's position) is a monumentaly dificult task to acomplish and the increased processing resources required don't worth it, IMHO is more practical and flexible to use the IR as a speaker/cab/mic combo (yeah, a snapshot) and play with the room through the reverb block (the cab block have some nice parameters to play too), this is like taking your real cab to move it to another room and/or walk yourself around the room as listener
 
Last edited:
Good point. I think you can trim them in cab lab though. Would be an interesting experiment. For live use I would think the less room the better because at high volume wouldn't the real room provide sufficient character?
 
??? Doesn't that just decrease the resolution of the IR?

No, it shortens the length, which is sort of a same thing. The auto trim in the CabLab tries to cut the beginning of the impulse, not the end, so that you could adjust different IRs to start at exact same time, but it isn't perfect.

If you have a 5 second impulse miced from 1" away from the center of a speaker, you will have just as little audible room echoes as in a 100ms impulse, since the reflected sounds are so quiet compared to the 12" EV @ 1 inch.
 
Last edited:
Is there a difference between the convoluted file length and the de-convoluted length?

Even if you generate an ir file using a 500 ms sweep, in a dead room, after de-convolution there only seems to be info for 110 ms at an audible level, at least in my extensive testing.

The difference that I found is that the longer convolution sweep used to generate the file seems to produce more details in the ir file post de-convolution (170ms versus 500ms).

I can dropbox a set of files for @FractalAudio to analyze where I recently did an experiment with the same cab/unchanged mic & cab position, same sample rate/bit depth, only difference was the initial sweep lengths.

I imported both de-convoluted WAV files into Cab Lab and used with the same preset, the ir's audibly sounded different, and looked different on a frequency graph in Cab Lab, which truncated the longer de-convoluted sweep.
 
Back
Top Bottom