The Acoustic Guitar Forum

Go Back   The Acoustic Guitar Forum > General Acoustic Guitar and Amplification Discussion > Acoustic Amplification

Reply
 
Thread Tools
  #31  
Old 10-30-2016, 06:12 PM
Doug Young's Avatar
Doug Young Doug Young is offline
Charter Member
 
Join Date: Apr 2005
Location: Mountain View, CA
Posts: 9,879
Default

Thanks for the info and the offer! I'll PM you and send you some files
Reply With Quote
  #32  
Old 10-30-2016, 09:22 PM
Mischief Mischief is offline
Registered User
 
Join Date: Jun 2015
Posts: 582
Default

Cuki,
For sake of this extremely interesting thread.
I would like to ask just a few more questions here. You have covered some of this in our discussions. But if you don't mind I'll ask some here I think others would be interested to know the answers.

If you make files with your guitar
How significant are the little things like if you decide to radically change the pick you use?
Or if you changed strings gauge and type? How about if you re fretted the guitar with stainless frets. (This is something I was planning to do on my guitar in the near future.)

In your video you use a slide, do you need files of slide if you planned to do that in the future or can the picking strumming patterns work fine?

From our discussions I believe some minor changes won't effect things too much but I wonder where the line is that would require one to redo new files?

BTW the part your doing on your end is very clever.

One other question I had.
Since the EPSI can do stereo. Is it possible to have the acoustic guitar convolution take place on one channel and apply the loaded reverbs or custom room correction using the other channel? I may be way out of my depth here. But I'm thinking there maybe more possibilities with the second channel in addition to stereo application.

Oh and a question regarding multisources. Can the pedal once set up handle the blended signal as mono and output the different pickups separated and then in stereo. Ir does one need to provide a separate signal from each source?

Sorry if I'm asking so many questions. I do that when I get excited and want to learn more about something. Yeah I'm annoying.[emoji51]
Reply With Quote
  #33  
Old 10-31-2016, 06:59 AM
Cuki79 Cuki79 is offline
Registered User
 
Join Date: Feb 2012
Location: France
Posts: 3,005
Default

Hi Mischief

As usual, my answer is not going to be short… Sorry for that. To understand, one has to understand what is an Impulse Response (IR).

* First, what is an "impulse". An impulse, is a very short "sound". Something very percussive, like taping on your strings. From an electric point of view, imagine that the signal is constant all the time (for ex: 0V) and suddenly for a microsecond you get +1 Volt. Well such a signal is actually compose of a lot's of frequencies. To give another image, imagine that you have a microphone with phantom power plugged into a 30 bands graphic equalizer. If you suddenly unplug the microphone, the 48V on the input suddenly will experience a suddent change. You usually have a huge "pop" sound. If you look at the graphic equalizer, all bands should turn (orange or red) all together. It means that huge pop is composed of every 30 frequencies.

An Impulse response basically describes the response of a room (for reverbs) or any resonant box (ex: guitar) across all the frequency spectrum. There are two usual ways to measure that, either you apply a short pulse to the object (a sudden "pop" sound for example) and record what's echoed in the room, or you can do a sine-sweep and record the response sweeping over all frequencies.

* Second, what is a response of the resonant object?
Well for example, if one consider a HiFi loudspeaker as a resonant object, you will find that people wants to have its response to be flat all over the audio frequency spectrum. It means that the response of the speaker is fast enough and can accurately reproduce the sudden "pop" sound. To work correctly, a loudspeaker must work in "piston" mode, it means that it is working close to its fundamental resonant mode. However if you push it too hard, it starts to break-up and many higher order modes appears. It is also the reason why in the high frequency response of a loud speaker you have so many bumps. It is the higher order modes taking over the piston one.

* So what about guitars? Guitars are very "bad" speakers. The response of a guitar is not flat at all (fortunately for us)

- The most important part is the top. Because it has the biggest surface in contact with air, it is the part that actually couples the most energy to the air. The top has not a flat response. It features resonant modes. Those modes see picture below resonates at certain frequencies.


It means that the top enhance certain sounds and not some others. There resonant frequencies are function of the surface, the velocity of sound (wood nature), and there elasticity (wood and structure). Bracing a guitar top affects all these parameters and tunes the top. That's why there is a difference between forward shifted, backward shifted, scalloped, and straight braces.

- The second part is the back. The back also induces resonances (usually, it makes antiresonances (dip) in the frequency response of the guitar).

- Another resonance comes from the air in the box. The sound hole play a role similar to the bass reflex event you have in a loud speaker. The air mass inside the box of the guitar, is tuned to resonant at low frequencies (between 40 and 110 Hz). That's why Taylor guitars with their slightly smaller body have less "ompf" than all mighty Martin dreadnought… Of course all of these resonances are driven by the bridge and the neck movements.

Now you understand a guitar is a quite complex physical system. Usually to model those systems, physicist use numerical methods because it is too complex to be modeled and mathematically solved "by hand".

* Third, So what is IR measurement? Well if you want to make a perfect image of your guitar: You should apply a sudden pop, by tapping with a hammer on the bridge and record the impulse response (Please don't do that). Your recording would then show the interferences of the sound waves radiating from all those resonances and anti resonances making the complex tone of the guitar. Because of acoustic damping, this impulse response decays very fast and can be recorded in a short wave file.

Another measurement technique would be for example to take a single coil pickup, bring it in front of the string and perform a sine-sweep by applying every frequencies of the audio spectrum to the single coil. The strings would vibrate and you could in principle record the response for every frequency.

Well we can't do that, can't we? What we can do, is play the guitar and wish most of the frequencies will be played. Now, as I explained to you in a PM, my philosophy is that one does not need to know the response of a frequency (note) if that person is never going to play it anyway. So basically when you record yourself for 4 minutes, an average player would come through most of his knowledge as a guitar player. So my technique will never perfectly match your guitar but It can be a pretty good approximation of you playing the guitar.
Now I believe also that by chance, every guitar stroke is like a small impulse. And by playing 4-5 minutes, you have a good approximation of a single big hammer strike on the bridge.
__________________
Martin 00-18V Goldplus + internal mic (2003)
Martin OM-28V + HFN + internal mic (1999)
Eastman E6OM (2019) Trance Audio Amulet
Yamaha FGX-412 (1998)

Gibson Les Paul Standard 1958 Reissue (2013)
Fender Stratocaster American Vintage 1954 (2014)
http://acousticir.free.fr/

Last edited by Cuki79; 10-31-2016 at 01:56 PM.
Reply With Quote
  #34  
Old 10-31-2016, 07:11 AM
Cuki79 Cuki79 is offline
Registered User
 
Join Date: Feb 2012
Location: France
Posts: 3,005
Default

Hi Mischief,

Now here is my answer
Quote:
How significant are the little things like if you decide to radically change the pick you use?
My IRs were made only with fingerstyle play. I barely can pick. Although, the guitarist of my band said he really like the pick response of the IR. So I think it is OK.

Quote:
> Or if you changed strings gauge and type? How about if you re fretted the guitar with stainless frets. (This is something I was planning to do on my guitar in the near future.)
That's a tricky one. Changing gauge can change the guitar top loading. The point is that your guitar is going to sound different acoustically. Of course the IR can not anticipate that. Since the IR essentially models the top, back, sound hole… I would say it's going to be OK. The main drawback of my IR (and I did only few trials, so I don't claim to have any expertise as for example Tone Dexter or Fishman people), is the high end which is also due to sound emission of the strings…. So I'd say for now, it's going to be OK. If we can improve the process to be very sensitive to direct string sound… Then You'd be happy anyway because it would be a great achievement

Frets are very important sound wise. The two most important thing that people always neglect is the two places where the strings seat: The frets and the bridge. Everything comes from there. Try to change the bridge vibrato-mass of your stratocaster from pot metal to steel: you'd be surprise how the sound change. Same thing for the saddles. So yes your guitar is going to sound different and the IR can't anticipate for that.

Anyway it's not a big deal to make a new IR.

Quote:
In your video you use a slide, do you need files of slide if you planned to do that in the future or can the picking strumming patterns work fine?
I did not use any slide to make the IR. It still sounds good. Slides sound is not closer to a sudden "pop" than regular playing.


Note that if it was that difficult to make an IR, ToneDexter would have been a massive failure at NAMM If I remember well the pedal only count to 10 or 12 to make an IR.

Quote:
> From our discussions I believe some minor changes won't effect things too much but I wonder where the line is that would require one to redo new files?
I have no idea, and the question did not occur to me because I can make as many IR, I want… Sorry. People have used Fishman Aura for years. They changed frets, strings, bridge, nut, even pickups and never complained. It's a non existing problem really (or yet). The problem only starts if the IR I'll make for you is a game changer. And we don't know that yet. Maybe it only works for me? Who knows?

Quote:
>Since the EPSI can do stereo. Is it possible to have the acoustic guitar convolution take place on one channel and apply the loaded reverbs or custom room correction using the other channel? I may be way out of my depth here. But I'm thinking there maybe more possibilities with the second channel in addition to stereo application.
I had the same idea and ask the guy from logidy. Got no answer. I believe it would be complex and not very rewarding for him. I mean he runs a company. He has to make a living. It's like asking Leo Fender in 1953 if he could make a gold chocolate green custom color telecaster with no extra charge… Leo would probably say no. (Imagine what would be the price of such a horror tele nowadays )

More seriously, you would get a lot more latency (because latency sums) and you would lose quality because of an extra stage of A/D-D/A conversion. If you are really aiming for the real thing and get the tail of the resonance perfect, I suggest to buy a second-hand Mac Mini (300$) and use a small audio interface. Then the possibilities are unlimited because a computer is much more powerful calculation wise. I am not sure about the meaning of that. Because the public would probably not hear any difference.

Jason Isbell said about the Fishman Aura, that he uses it because it sounds really good and that he can not hear any difference with micing. If the PA in the venues he is playing can not make the difference then, my poor little acoustic amp will not either.

Quote:
>Oh and a question regarding multisources. Can the pedal once set up handle the blended signal as mono and output the different pickups separated and then in stereo. Ir does one need to provide a separate signal from each source?
No it can not. … Well may be it could if the two sources were "mathematicaly orthogonal" meaning that the IR signature of both signal being VERY VERY different. For example, a Cello blended with a ukulele. Maybe then you could do something… But it would sound bad, very bad.

The EPSI has two inputs. For stereo use, you should use both inputs. I know you are thinking about your AUX input. But for me if you put a mic in your Takamine, you won't need the EPSI IR. Don't' forget that before adding something in the signal chain, it MUST bring much more than it takes. Very few pedals do that (in my opinion). For example, I've been through about any parametric EQ on the market and now I start to feel that I should not have sold the Taylor K4… Rupert Neve Designs is a fine company.
__________________
Martin 00-18V Goldplus + internal mic (2003)
Martin OM-28V + HFN + internal mic (1999)
Eastman E6OM (2019) Trance Audio Amulet
Yamaha FGX-412 (1998)

Gibson Les Paul Standard 1958 Reissue (2013)
Fender Stratocaster American Vintage 1954 (2014)
http://acousticir.free.fr/
Reply With Quote
  #35  
Old 10-31-2016, 04:28 PM
Cuki79 Cuki79 is offline
Registered User
 
Join Date: Feb 2012
Location: France
Posts: 3,005
Default

Just for teasing,

I've made a user friendly interface of my program and automated the process (button "Clean the tail").



I will try it on Mischief or Doug Young's files.

Regards,
Kien
__________________
Martin 00-18V Goldplus + internal mic (2003)
Martin OM-28V + HFN + internal mic (1999)
Eastman E6OM (2019) Trance Audio Amulet
Yamaha FGX-412 (1998)

Gibson Les Paul Standard 1958 Reissue (2013)
Fender Stratocaster American Vintage 1954 (2014)
http://acousticir.free.fr/
Reply With Quote
  #36  
Old 10-31-2016, 05:26 PM
Mischief Mischief is offline
Registered User
 
Join Date: Jun 2015
Posts: 582
Default

Let me just say again thank you for your contribution and thorough explanations of all this.
Reply With Quote
  #37  
Old 11-01-2016, 06:37 PM
Cuki79 Cuki79 is offline
Registered User
 
Join Date: Feb 2012
Location: France
Posts: 3,005
Default

Hi everyone,

I borrowed a friend (Thank you Hoogy) a pair of behringer C2 microphones and Zoom R16. So I have tried the stereo IR.

Here is a video of the process (updated quality):

https://youtu.be/DpRP8olJR6s

Sorry for the distortion during the strumming part and the volume drop at the end. The video has poor image quality but it takes ages to upload it in youtube so I was not brave enough to make a good quality one. Anyway you get to see how it is done.

You can see at the end of the video thanks to a Stereoscope that the stereo IR does rebuild a stereo image from a mono signal.

Here are samples for the stereo imaging:
XY mic:


stereo IR 90% blend


stereo IR 100%


Lyric 100% (mono)


Please tell what you think.
__________________
Martin 00-18V Goldplus + internal mic (2003)
Martin OM-28V + HFN + internal mic (1999)
Eastman E6OM (2019) Trance Audio Amulet
Yamaha FGX-412 (1998)

Gibson Les Paul Standard 1958 Reissue (2013)
Fender Stratocaster American Vintage 1954 (2014)
http://acousticir.free.fr/

Last edited by Cuki79; 11-02-2016 at 12:47 AM.
Reply With Quote
  #38  
Old 05-01-2017, 03:43 PM
rschultz rschultz is offline
Registered User
 
Join Date: Jun 2007
Location: Knoxville, TN
Posts: 1,289
Default

Fascination thread


Sent from my iPhone using Tapatalk
__________________
'10 Wechter 5712c - Fishman Rare Earth
'13 Jaffrey #26 - Malaysian Blackwood!
'21 Gretsch 5622
'22 Furch Red Pure G-LR - Barbera Soloist
ST-300 Mini + DIY mic preamp
Reply With Quote
  #39  
Old 05-02-2017, 12:35 PM
rschultz rschultz is offline
Registered User
 
Join Date: Jun 2007
Location: Knoxville, TN
Posts: 1,289
Default

Quote:
Originally Posted by Cuki79 View Post

To see the Gilgaus Mod follow this link:
http://theunofficialmartinguitarforu...1#.WA5aBIVR3Rd
So the Gilgaus Mod... just moves the electronics outboard? Why is that so important? Couldn't you go directly into the IR Convolver from the stock Lyric?

You say the trick is to NOT use any EQ. Instead, to use a impulse response IR. I tried to follow along the explanation, but it's a bit over my head (I'm a mechanical engineer... not an EE!). Could you maybe explain it in more laymen's terms?

I'm pretty happy with my K&K + mag sound, but I have a Lyric that I was going to sell... now I'm curious...
__________________
'10 Wechter 5712c - Fishman Rare Earth
'13 Jaffrey #26 - Malaysian Blackwood!
'21 Gretsch 5622
'22 Furch Red Pure G-LR - Barbera Soloist
ST-300 Mini + DIY mic preamp
Reply With Quote
  #40  
Old 05-02-2017, 12:54 PM
Cuki79 Cuki79 is offline
Registered User
 
Join Date: Feb 2012
Location: France
Posts: 3,005
Default

Quote:
Originally Posted by rschultz View Post
So the Gilgaus Mod... just moves the electronics outboard? Why is that so important? Couldn't you go directly into the IR Convolver from the stock Lyric?

You say the trick is to NOT use any EQ. Instead, to use a impulse response IR. I tried to follow along the explanation, but it's a bit over my head (I'm a mechanical engineer... not an EE!). Could you maybe explain it in more laymen's terms?

I'm pretty happy with my K&K + mag sound, but I have a Lyric that I was going to sell... now I'm curious...
Gilgaus mod was just done in order to make no modification to the guitar and still install the Lyric with no battery inside the guitar. Probably I would not do it again.
1) I had tons of problem with faulty cables making huge loud pops
2) If I ever forget to turn off the pedal, the battery dies just amplifying whatever the Lyric mic take from ambient sound...

In very simple, an Impulse Response convolution can be seen as a graphic EQ with as many frequency bands as the IR length. Right now, my favourite IR is around 17000 long. Moreover on each frequency you can also adjust the phase: the time delay between each frequency.

When you can have a 17000 bands graphic equalizer, you don't need to use a 3-band EQ or any parametric EQ.

My process as well as Tonedexter's are about how to tune the 17000 bands of the graphic equalizer to mimic a microphone response.

Tell me if it is clear enough for you.

You can still sell your Lyric because the process works for any pickup except magnetic ones (soundhole pickups). It will work on the K&K as perfectly shown by Buzzardwhiskey in the Tonedexter thread. I purposely used the Lyric because
1) I knew the Lyric only missed the low end and plenty of midrange and high end
2) I Knew the Lyric would react as a mic.
I wanted that.

Cuki
__________________
Martin 00-18V Goldplus + internal mic (2003)
Martin OM-28V + HFN + internal mic (1999)
Eastman E6OM (2019) Trance Audio Amulet
Yamaha FGX-412 (1998)

Gibson Les Paul Standard 1958 Reissue (2013)
Fender Stratocaster American Vintage 1954 (2014)
http://acousticir.free.fr/
Reply With Quote
  #41  
Old 05-02-2017, 01:19 PM
rschultz rschultz is offline
Registered User
 
Join Date: Jun 2007
Location: Knoxville, TN
Posts: 1,289
Default

Quote:
Originally Posted by Cuki79 View Post
Gilgaus mod was just done in order to make no modification to the guitar and still install the Lyric with no battery inside the guitar. Probably I would not do it again.
1) I had tons of problem with faulty cables making huge loud pops
2) If I ever forget to turn off the pedal, the battery dies just amplifying whatever the Lyric mic take from ambient sound...

In very simple, an Impulse Response convolution can be seen as a graphic EQ with as many frequency bands as the IR length. Right now, my favourite IR is around 17000 long. Moreover on each frequency you can also adjust the phase: the time delay between each frequency.

When you can have a 17000 bands graphic equalizer, you don't need to use a 3-band EQ or any parametric EQ.

My process as well as Tonedexter's are about how to tune the 17000 bands of the graphic equalizer to mimic a microphone response.

Tell me if it is clear enough for you.
Yes, that makes sense... good explanation.

As an aside, I have never understood the hesitation to ream out the end pin to accept a standard 1/2" jack. New guitar, old guitar, I wouldn't hesitate as long as the end block is a standard thickness. If it's too old to ream out the jack, then it's too old to put a pick up in it! My 2 cents.

__________________
'10 Wechter 5712c - Fishman Rare Earth
'13 Jaffrey #26 - Malaysian Blackwood!
'21 Gretsch 5622
'22 Furch Red Pure G-LR - Barbera Soloist
ST-300 Mini + DIY mic preamp
Reply With Quote
  #42  
Old 07-29-2017, 03:47 AM
KLB KLB is offline
Registered User
 
Join Date: Jul 2017
Posts: 3
Default Computing the IR in Matlab

Hi,

This is very impressive work. I am interested in trying your method of creating and using my own IRs. I follow everything you did with the exception of this statement.

"Then I computed the IR Fourier Transform by dividing the Fourier Transform of both recordings using MATLAB."


I understand converting both wav files to their respective FFTs, but I don't understand why you divided them before doing the IFFT. And did you divide the FFT of the Lyric by the FFT of the microphone, or the other way around.

Thanks so much for the work, and thanks in advance for any help you can provide relative to my questions.

I am an amateur player, and proficient in the use of Matlab and Scilab.
Reply With Quote
  #43  
Old 07-29-2017, 09:56 AM
Vancebo Vancebo is offline
Registered User
 
Join Date: Feb 2003
Location: Salem, Or.
Posts: 2,027
Default

Thanks for resurrecting an old thread. I missed it before. Now that tonedexter is a reality, how does it compare to your work Cuki? (I know, you may have covered this in another thread)
__________________
Vancebo
Husband of One, Father of Two
Worship Leader, Music Teacher
Oregon Duck Fan
Guitars by: Collings, Bourgeois, Taylor
Pickups by: Dazzo
Preamps by: Sunnaudio
Amps by: Bose (S1)
Grateful
Reply With Quote
  #44  
Old 07-30-2017, 01:22 AM
Cuki79 Cuki79 is offline
Registered User
 
Join Date: Feb 2012
Location: France
Posts: 3,005
Default

Quote:
Originally Posted by KLB View Post
Hi,

This is very impressive work. I am interested in trying your method of creating and using my own IRs. I follow everything you did with the exception of this statement.

"Then I computed the IR Fourier Transform by dividing the Fourier Transform of both recordings using MATLAB."


I understand converting both wav files to their respective FFTs, but I don't understand why you divided them before doing the IFFT. And did you divide the FFT of the Lyric by the FFT of the microphone, or the other way around.

Thanks so much for the work, and thanks in advance for any help you can provide relative to my questions.

I am an amateur player, and proficient in the use of Matlab and Scilab.
Ok the idea is to compute the impulse response IR(t) that would bring the pickup signal p(t) to the mic signal m(t).

in time domain, m(t)=p(t) x IR(t). Where "x" denotes the convolution product that the pedal or convolution reverb plugin will perform. A convolution product in time domain, is equivalent to a multiplication in frequency domain.
So in frequency domain fft[m]=fft[p].fft[IR]
Here "." Denotes the multiplication and fft, the fast Fourier transform operation.

There is nothing patented here. Both Tonedexter and Fishman patents explain that based on a scientific publication published prior to their patent.

So fft[IR]=fft[m]/fft[pic]

Where " / " denotes the division.

So IR(t)=ifft[fft[IR]]

If you get there and blend, you have the Fishman aura.
__________________
Martin 00-18V Goldplus + internal mic (2003)
Martin OM-28V + HFN + internal mic (1999)
Eastman E6OM (2019) Trance Audio Amulet
Yamaha FGX-412 (1998)

Gibson Les Paul Standard 1958 Reissue (2013)
Fender Stratocaster American Vintage 1954 (2014)
http://acousticir.free.fr/
Reply With Quote
  #45  
Old 07-30-2017, 01:31 AM
Cuki79 Cuki79 is offline
Registered User
 
Join Date: Feb 2012
Location: France
Posts: 3,005
Default

Quote:
Originally Posted by Vancebo View Post
Thanks for resurrecting an old thread. I missed it before. Now that tonedexter is a reality, how does it compare to your work Cuki? (I know, you may have covered this in another thread)
Hi Vancebo,

It is impossible to answer since I don't have a tonedexter. The only one that was on sale in Europe was second hand and super expansive. I was not curious to invest and preferred to buy Try the Trance Audio Amulet.

From what I heard from Doug I think it should be totally comparable. The difference being that Doug can achieve the tone he wants in minutes be performing the process with different mic positions. When I was experimenting with Doug, he sent me each time the same mic position based on his studio favorite sound. Experience on the AGF seems to show that best studio mic placement may not be the best for IR making. I don't see why it would not apply to me.

I have taken a different approach, trying different recipe, and signal processing to optimize one mic placement for a simple reason: the only good mic I have is the clip on DPA 4099G. Because of the gooseneck I could not play much with mic placement...

I am pretty sure that if I owned a Tonedexter I could implement on my computer a similar algorithm. The patent does not hide much, it is well written.

Vancebo: if you want to try, just send me a pm. All you need is send 2 wav files. At home you need a computer, a program to record like GarageBand or cubase and an audio interface (presonus, tascam...)

Edited note: For now SpruceTop gets a better tone with tonedexter and his HD-28 + amulet than I do with my early stage IRs. What's great about Tonedexter is
1) That their algorithm is really well balanced (frequency wise)
2) The user can easily play with mic positions to get the end-tone he wants

Cuki
__________________
Martin 00-18V Goldplus + internal mic (2003)
Martin OM-28V + HFN + internal mic (1999)
Eastman E6OM (2019) Trance Audio Amulet
Yamaha FGX-412 (1998)

Gibson Les Paul Standard 1958 Reissue (2013)
Fender Stratocaster American Vintage 1954 (2014)
http://acousticir.free.fr/

Last edited by Cuki79; 07-30-2017 at 03:43 PM.
Reply With Quote
Reply

  The Acoustic Guitar Forum > General Acoustic Guitar and Amplification Discussion > Acoustic Amplification

Thread Tools





All times are GMT -6. The time now is 02:14 PM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.
Copyright ©2000 - 2022, The Acoustic Guitar Forum
vB Ad Management by =RedTyger=