sampleaccurate
Inspired
Moderator's note: this exchange has been separated from the original "Redwirez tone recipe" thread
This topic is of intense interest to me. I've owned the Ultra almost 4 weeks and I'm now seriously getting into the user cab sims. I've used IR reverb extensively in my DAW, but never before have I used IRs for cab sims.
What's still missing?
Scott, you convinced me to purchase the BigBox IRs from Red Wire. I believe you when you say you're getting good results mixing IRs. I've been having my own fun creating cab IRs with GR3 and IRs I already have using a VST host in my computer to run parallel IRs. I disable the cab sim on my AXE before processing the output with a mix of cab sims from GR3 and "real" IRs with some interesting results. It's fun to experiment with the infinite possibilities, and I'm also having a blast exploring the capabilities.
I think mixing IRs will put the finishing touch on what I see as currently the weakest yet vitally important link in the signal chain. The main reason this will improve the realism, I believe, is that by using multiple IRs you are "filling in" the gaps or notches in the frequency response of a single IR by itself. Look at the response of most IRs and you'll see a pronounced comb filtering effect from the microphone placement. The fact that the sound from different parts of the speaker arrive at the microphone at different times is what causes the phase notching. Sound travels about 1 foot in 1 millisecond - that's an easy to remember approximation. So if you mike a 12" speaker near the edge, the sound from the far edge of the speaker reaches the mic about 1 ms later than the sound from the close edge (although it's much lower in amplitude) canceling frequencies near 500 HZ. The interaction is far more complex than that due to the fact the sound doesn't come from a point but from a large diaphragm that's 12 inches more or less in diameter, but the result is that some frequencies are notched out, that is, the EQ curve of the IR has deep, narrow depressions where a "slice" has been removed from the signals spectrum. I said in another post that most of the IRs to me sound like "caricatures" of cabs with exaggerated EQ and phase notching compared to a real guitar amp. Mixing the IRs should eliminate this problem since the average response obtained from doing so will be free of deep notches at any one frequency.
If you take a look at the EQ curves of the IRs you're creating by mixing them together I would suspect that the curves aren't as "jagged" and have fewer, much less pronounced notches in the response than single IRs. That's my experience with the GR3 cabs I've been playing with. Tonight I'll have the BigBox IRs to play with. Maybe I should have waited until I could experiment with the BigBox before writing this, but I'll have a report tomorrow.
I'd like to add another opinion to my post.
There are two serious problems as I see it with the IR method of cab simulation in the AXE. The first is being addressed by mixing IRs to get rid of those pesky notches in the frequency response. That one is or will be licked.
The second problem will have to wait for the AXE-FX II with gigabytes instead of megabytes of memory, and slots for hundreds of user cabs. The important thing is that it's designed such that the IRs are permitted to be much longer, perhaps as long as 1 second. This will require more processing power - a lot more. I believe the CPU load from an IR is proportional to its length, so 1 second is a pretty big leap (I could be wrong on that but I seem to remember reading it). But computers can process 1 second stereo IR convolutions with low CPU usage, so I see no reason the AXE-FX II can't be designed to do the same.
The reason this is necessary is that a 20ms IR isn't long enough to capture a room sound. 20ms of room reflection doesn't do much for the sound other than mold the frequency response through phase cancellation and reinforcement. The 20ms IR is effectively an EQ - a very complex one, but still just an EQ. IMO being able to capture the sound of a room and the full decay of the sound captured by the distant mikes would add the final touch of realism that is needed. As a substitute, the reverb sim module could be used to get a small room sound for a decent simulation, but it's not the same as having an IR recreate the complex pattern of delayed sound of distant mikes and real room reflections. It's OK, but a good IR would trump the reverb sim in the simulation of miking techniques that result in an IR that is longer than 20ms. I see this as the only "gap" in the effects and sims in the AXE-FX: not being able to capture the room reflections in the cabinet IR.
Are the Red Wirez IRs only 20ms long, and if so, how can they claim to capture the room reflections? If they are longer than 20ms, I guess the AXE just uses the first 20ms and truncates the decay. That's rhetorical since I'm buying the BigBox tonight and will find out for myself.
I quote the Red Wirez website: "voila, out comes the sound of your guitar played through a well-mice’s speaker cabinet, in an acoustically treated live room, run through a Neve 1073" (bold added by me for emphasis).
All sound that travels about 20 feet or more, including reflections, will be truncated by the AXE IR. That's not what I call the sound of an acoustically treated live room. Even the tiniest room will have reflections longer than 20ms. The reflection from a wall 11 feet away would be completely truncated since it would take 22 ms for the sound to make the round trip back to the mic.
I'm not here to rag on the AXE - it's the best deal I've ever gotten on a piece of gear - period. I'm just salivating at the next generation AXE that will accommodate (I hope) long stereo IRs for both cabinet simulation and room ambience and reflections and endorsing the concept of mixing IRs. Long IRs will add the only thing I can think of that my computer can do and my AXE can't. The alternative in the meantime is to run longer IRs in my DAW.
Stephen Cole
This topic is of intense interest to me. I've owned the Ultra almost 4 weeks and I'm now seriously getting into the user cab sims. I've used IR reverb extensively in my DAW, but never before have I used IRs for cab sims.
What's still missing?
Scott, you convinced me to purchase the BigBox IRs from Red Wire. I believe you when you say you're getting good results mixing IRs. I've been having my own fun creating cab IRs with GR3 and IRs I already have using a VST host in my computer to run parallel IRs. I disable the cab sim on my AXE before processing the output with a mix of cab sims from GR3 and "real" IRs with some interesting results. It's fun to experiment with the infinite possibilities, and I'm also having a blast exploring the capabilities.
I think mixing IRs will put the finishing touch on what I see as currently the weakest yet vitally important link in the signal chain. The main reason this will improve the realism, I believe, is that by using multiple IRs you are "filling in" the gaps or notches in the frequency response of a single IR by itself. Look at the response of most IRs and you'll see a pronounced comb filtering effect from the microphone placement. The fact that the sound from different parts of the speaker arrive at the microphone at different times is what causes the phase notching. Sound travels about 1 foot in 1 millisecond - that's an easy to remember approximation. So if you mike a 12" speaker near the edge, the sound from the far edge of the speaker reaches the mic about 1 ms later than the sound from the close edge (although it's much lower in amplitude) canceling frequencies near 500 HZ. The interaction is far more complex than that due to the fact the sound doesn't come from a point but from a large diaphragm that's 12 inches more or less in diameter, but the result is that some frequencies are notched out, that is, the EQ curve of the IR has deep, narrow depressions where a "slice" has been removed from the signals spectrum. I said in another post that most of the IRs to me sound like "caricatures" of cabs with exaggerated EQ and phase notching compared to a real guitar amp. Mixing the IRs should eliminate this problem since the average response obtained from doing so will be free of deep notches at any one frequency.
If you take a look at the EQ curves of the IRs you're creating by mixing them together I would suspect that the curves aren't as "jagged" and have fewer, much less pronounced notches in the response than single IRs. That's my experience with the GR3 cabs I've been playing with. Tonight I'll have the BigBox IRs to play with. Maybe I should have waited until I could experiment with the BigBox before writing this, but I'll have a report tomorrow.
I'd like to add another opinion to my post.
There are two serious problems as I see it with the IR method of cab simulation in the AXE. The first is being addressed by mixing IRs to get rid of those pesky notches in the frequency response. That one is or will be licked.
The second problem will have to wait for the AXE-FX II with gigabytes instead of megabytes of memory, and slots for hundreds of user cabs. The important thing is that it's designed such that the IRs are permitted to be much longer, perhaps as long as 1 second. This will require more processing power - a lot more. I believe the CPU load from an IR is proportional to its length, so 1 second is a pretty big leap (I could be wrong on that but I seem to remember reading it). But computers can process 1 second stereo IR convolutions with low CPU usage, so I see no reason the AXE-FX II can't be designed to do the same.
The reason this is necessary is that a 20ms IR isn't long enough to capture a room sound. 20ms of room reflection doesn't do much for the sound other than mold the frequency response through phase cancellation and reinforcement. The 20ms IR is effectively an EQ - a very complex one, but still just an EQ. IMO being able to capture the sound of a room and the full decay of the sound captured by the distant mikes would add the final touch of realism that is needed. As a substitute, the reverb sim module could be used to get a small room sound for a decent simulation, but it's not the same as having an IR recreate the complex pattern of delayed sound of distant mikes and real room reflections. It's OK, but a good IR would trump the reverb sim in the simulation of miking techniques that result in an IR that is longer than 20ms. I see this as the only "gap" in the effects and sims in the AXE-FX: not being able to capture the room reflections in the cabinet IR.
Are the Red Wirez IRs only 20ms long, and if so, how can they claim to capture the room reflections? If they are longer than 20ms, I guess the AXE just uses the first 20ms and truncates the decay. That's rhetorical since I'm buying the BigBox tonight and will find out for myself.
I quote the Red Wirez website: "voila, out comes the sound of your guitar played through a well-mice’s speaker cabinet, in an acoustically treated live room, run through a Neve 1073" (bold added by me for emphasis).
All sound that travels about 20 feet or more, including reflections, will be truncated by the AXE IR. That's not what I call the sound of an acoustically treated live room. Even the tiniest room will have reflections longer than 20ms. The reflection from a wall 11 feet away would be completely truncated since it would take 22 ms for the sound to make the round trip back to the mic.
I'm not here to rag on the AXE - it's the best deal I've ever gotten on a piece of gear - period. I'm just salivating at the next generation AXE that will accommodate (I hope) long stereo IRs for both cabinet simulation and room ambience and reflections and endorsing the concept of mixing IRs. Long IRs will add the only thing I can think of that my computer can do and my AXE can't. The alternative in the meantime is to run longer IRs in my DAW.
Stephen Cole