Improving the LR Baggs Lyric sound

[Cuki79]

Hi everyone,

I am close to get the sound I want from the LR Baggs Lyric. There are more and more people struggling with that pickup so I want to share what I did to improve its sound.

First, here is my rig:
* 2012 Martin D-18
* LR Baggs Lyric (in position 3 in Baggs recommandation, on the side close to low E string) with homemade Gilgaus Mod (= Tapastring Vintage jack + Baggs preamp in a pedal) with an added mute switch.
* Homemade slightly modified CAE Boost/Line driver pedal
* Logidy EPSI IR convolver

To see the Gilgaus Mod follow this link:
http://theunofficialmartinguitarforu...1#.WA5aBIVR3Rd

To see the LR Baggs lyric positions follow this link:
http://medias.audiofanzine.com/image...al/1573014.jpg



So the trick is not to use any EQ or parametric EQ. I have a D-tar Equinox 3 band full parametric EQ, and I could never dial a fully satisfying sound.

Why? Because the Lyric has an intrinsic mid range/boxy voice due to the fact that the mic is in the guitar. So my strategy was to make an impulse response IR (like the Fishman Aura, D-Tar mama bear, Zoom A3, or ToneDexter). Only the Tone dexter enables to make “easily” your own IR but its price is way over of my budget. So I used a guitar cab/reverb convolver and made my own IR on a computer.

For the cab IR convolver, I chose the Logidy EPSI because it can do 1.5s IR with zero latency. 1.5s is longer than the AMT Pangea or Line 6 Helix. It is also cheaper than the torpedo C.A.B for example. But it should work with any IR convolver.

How did I made the IR?
I recorded myself with both the DPA 4099G and the lyric plugged into a Focusrite 2i2. Then I computed the IR Fourier Transform by dividing the Fourier Transform of both recordings using MATLAB. Back in time-domain, I faded the tail of my IR to 150 ms with audacity. I also computed different blend settings (wet/dry - similar to Fishman aura blend knob). The resulting IR wav files were then loaded on the SD card of the logidy convolver pedal, "et voilà" the job was done.

You can hear some samples here (sorry for the bad guitar skills and volume mismatch). The recordings are:
1) 100 % IR compared to DPA microphone
2) 67% IR + 33% Baggs Lyric compared to DPA microphone
3) 67% IR + 33% Baggs Lyric compared to 100% Baggs Lyric

Everything is recorded direct into a Focusrite 2i2 and Garageband, no EQ, no Effects. Of course, these recordings are differents from the ones I've used to make the IR (it would be cheating).

To listen click the links:




For the comparison between Lyric EPSI IR and straight Lyric, please listen until the very end: When strumming the guitar with a pick you can really hear the lack of bass and low midrange on the lyric alone.


Same samples without the ping-pong (it's a playlist: there are 5 samples: EPSI 100%, DPA 100%, EPSI 67% (best), DPA 100%, and Lyric (100%)).


Note that the latency of the pedal is very good. In my opinion, the feeling is much better than with the Fishman Aura. Morever, you keep the dynamic response of a microphone.

Next step: Make another IR with a proper microphone (DPA 4099G is quite boomy due to proximity effect) and try the stereo mode... But I need another sound card with at least 3 inputs (impossible with my Focrusrite 2i2). Maybe after Christmas

Tell me what you think.

[dannyg1]

That's an exciting improvement you've made. It does sound very much like the Tonedexter demo's too.

[Cuki79]

Thanks,

last week I let the two guitarists of my band make a comparison. They were both playing. One with the DPA 4099G on a martin HD28 and the other with my Martin D-18 equipped with Lyric + EPSI IR plugged in a SR-technology JAM 150+.

I was very happy, in live condition, you could not tell the quality difference. (But we are not professionals, probably a trained ear would tell)

Of course I lose something with the pedal. Using a software convolver, putting the quality to maximum and not limiting the IR length on the MacBook you can really reach an outstanding quality. But then it's not very user friendly...

I think I can improve the tail of the guitar "sustain". I can hear the difference between the one I get from the pedal and the one I get on the computer (with quality set to max). Here on the samples you hear it's a bit dry and cold because the fading is strong. I am sure I can make it better EQing for example the dry path and the tail of the wet path... However the guy from logidy told me I was wrong and that the guitar sustain is not that long.

Anyway in the future I will try to better capture that warm modulation on the tail of the D-18 sustain (even if it comes more from the room than from my D-18.

[Petty1818]

Sounds good! However, that's a lot of work to make a pickup sound good. I guess it does go to show that experimentation is often necessary though.

[Charmed Life Picks]

Thanks for this post. I'm just getting into this as well.

Scott

[Cuki79]

Hi Petty1818,

It is a bit of work but note that:
* The Gilgaus Mod
* The homemade Boost
* And position 3 are not compulsory as soon as you use an IR convolver.

I did the Gilgaus Mod because I did not want any battery in the Martin. Position 3 was meant to have a more balance sound. Position 1 is balanced but VERY midrangy. Position 2 has more bass but had a nasty resonance somewhere on the 1st or 2nd string. Position 3 was better. However, with the IR convolver I could go back to position 1 and make the corresponding IR.

The two mods combined resulted in a less-loud pickup. Probably because I have 3m of cable before the preamp and that the midrange mostly contributes to loudness (Human ear/brain efficiency).

This is the reason why I made the boost/line driver preamp.

So if you have a Lyric equipped guitar and want to make a try. It costs nearly zero.

1) Record both pickup and microphone on your computer.
2) Make the IR (that costs a bit of knowledge, but you don't have to use MATLAB: Octave or Scilab are free GNU equivalent).
3) Try a free convolver on your PC like NadIR Ignite amps.

[dannyg1]

I'd really like to hear the dynamic range you've described in action. Any chance you might record the system heavily strummed with a pick? That seems to be the toughest sound for a pick-up to reproduce and makes a mess of most systems (Undersaddles especially).

[martingitdave]

Brilliant! You created your own Aura system (Tonedexter) calibrated for the Lyric! I'm impressed! What I also find interesting is the experiment refutes the argument that many Lyric users make that the lack of bass is an accurate representation of the guitar. When I tried the lyric in a 12 string a few years ago, it sounded great. Then, I attempted to use it on my new Martin. It was too midrangey and had insufficient bass. I ended up with a multi-source system. The fact that the IR can add back the missing frequencies that are recorded with a high quality mic suggests that the internal preamp should be revoiced, or offer multiple voicings. It bodes well for the future of Tondexter! If I were a touring artist, I would most certainly invest in a Tondexter and equip my guitars with lightweight passive piezo pickups.

Viva le mathématiques! (My French stinks)

[Cuki79]

Here is a video of heavy strumming with a pick.

I strummed the hardest I could. I am sorry about the lousy playing, I am more a geek than a guitar player.



If the inserted video does not work (it does not on the preview)
Folllow the link
https://youtu.be/uf2rYHEOg84

Cuki

[dannyg1]

Thanks for doing that, and so quickly! It was very informative as there was an obvious difference in compression style artifacts and an also obvious difference in dynamic range. Still, it sounded very impressive.

I'm looking forward to your next IR and, given what I've heard so far, I'd like to buy it and buy one of those EPSi pedals to put it in. I'll give you 10 bucks for it.

[Cuki79]

To answer Martingitdave,

I am not convinced that making your own IR would make a piezo "feel" better. For heavy strumming for example, it is well known that piezo compress and quack. My feeling is that there are not linear (contrary to Tonedexter claims).

The weird thing is that the most linear pickup (physicaly speaking) would be the magnetic pickup. But since it is positionned somewhere close to the soundhole you get a comb filtering effect that depends on the string and note you play... No convolution can correct that. So when Tonedexter says that magnetic pickups are not linear: It is wrong and true. It depends on the definition of "linearity".

To come back to the piezo, of course the EPSI may sound better (Since Larry Fishman said in an interview that the Aura is a 2000 taps filter... much less than the EPSI)... But it will never correct nonlinearity.

That is the reason why I chose to correct the Lyric response...

However I doubt LR Baggs could much improve the preamp. It something that I find impossible to do with standard analog electronic. A good try is to record your Lyric and apply as many filters and multiband compression you can with VST or AU plugin. It is very difficult to make the Lyric sound perfect. For me it is related to the box effect on the microphone. In my scientific calculation program, I can actually look at what the IR do in the frequency domain, and there is a lot of fine tuning in the midrange to get a natural response. I know because I tried cutting the bass and keeping only that midrange fine tuning and found the sound to be right. I lost the boom but still made the sound more natural. There are too many resonance modes from the "box" to correct for standard parametric EQs or multiband compressor.

Finally, I have to add that I think the Tonedexter is way too expansive for what it does. It is probably because they raised 14 435$ on kickstarter but did not reach 100k$.

If I had time I could probably program a JAVA version of my little program with user friendly interface and anyone could make his own IR for free. I could put a "buy me a beer" button as seen on some free VST plugin with a 5-7$ paypal transfer link

PS: In the Quest for ultimate pickup, my 2 cents is that people do not look in the right direction. I believe the magnetic pickup is the best pickup. It can be noiseless, does not feedback and is linear. The only problem is to get the computer calculation power and a fine algorithm to guess what strings is played every time and correct the comb filter effect.

[Cuki79]

the compression and artifact are also due to the fact that I strummed the hardest I could. So the Focusrite was blinking orange-red... Moreover iMovie did not do well compressing the audio for youtube...

If I ever have time to make the Java version of the software, I'll put a button so you can paypal me those 10 bucks

Also note that I made a 16-bits, impulse response but the pedal can stand 24bits that makes a huge difference in dynamic range.

Cuki

[martingitdave]

Originally Posted by Cuki79

To answer Martingitdave,

I am not convinced that making your own IR would make a piezo "feel" better. For heavy strumming for example, it is well known that piezo compress and quack. My feeling is that there are not linear (contrary to Tonedexter claims).

The weird thing is that the most linear pickup (physicaly speaking) would be the magnetic pickup. But since it is positionned somewhere close to the soundhole you get a comb filtering effect that depends on the string and note you play... No convolution can correct that. So when Tonedexter says that magnetic pickups are not linear: It is wrong and true. It depends on the definition of "linearity".

To come back to the piezo, of course the EPSI may sound better (Since Larry Fishman said in an interview that the Aura is a 2000 taps filter... much less than the EPSI)... But it will never correct nonlinearity.

That is the reason why I chose to correct the Lyric response...

However I doubt LR Baggs could much improve the preamp. It something that I find impossible to do with standard analog electronic. A good try is to record your Lyric and apply as many filters and multiband compression you can with VST or AU plugin. It is very difficult to make the Lyric sound perfect. For me it is related to the box effect on the microphone. In my scientific calculation program, I can actually look at what the IR do in the frequency domain, and there is a lot of fine tuning in the midrange to get a natural response. I know because I tried cutting the bass and keeping only that midrange fine tuning and found the sound to be right. I lost the boom but still made the sound more natural. There are too many resonance modes from the "box" to correct for standard parametric EQs or multiband compressor.

Finally, I have to add that I think the Tonedexter is way too expansive for what it does. It is probably because they raised 14 435$ on kickstarter but did not reach 100k$.

If I had time I could probably program a JAVA version of my little program with user friendly interface and anyone could make his own IR for free. I could put a "buy me a beer" button as seen on some free VST plugin with a 5-7$ paypal transfer link

PS: In the Quest for ultimate pickup, my 2 cents is that people do not look in the right direction. I believe the magnetic pickup is the best pickup. It can be noiseless, does not feedback and is linear. The only problem is to get the computer calculation power and a fine algorithm to guess what strings is played every time and correct the comb filter effect.

Great analysis. You points about linearity and pickups are correct. It is interesting that playing style dictates how much, if any, difference it makes for a player. I suppose some of the other SBT systems like Trance or Dazzo might perform better at high input, and be better candidates for your system. What you've done with the internal mic is clever. You point about the mag pickup makes perfect sense to me.

Great job and thanks for sharing your research and experimentation with us.

[Cuki79]

Hi

I think I once saw an explanation/patent of the FRAPS/Amulet system and It looked really nice. I think it was something like 3 cantilevers in X,Y, and Z position, each carrying a piezo. It is then able to sens the 3-D vibration of the top. Something a MEMS accelerometer would do owadays but with probably worse noise performance.

The Dazzo patent is more mysterious for me. I kind of get the Bessel mode thing but for me it does not make sens with regards of the acoustic wavelengths.

Anyway, as Doug Young said in a previous post, the Lyric is more a soundboard tranducer than a real mic. So the IR strategy would probably work well with any of these soundboard transducers.

Cuki

[Cuki79]

Hi

To avoid too much of video compression here is nearly the same recording... With the same lousy playing... I've put the IR to 100% so you can compare.

DPA:


100% IR:


I think the rig behave's well under heavy strumming.

As far as compression and dynamic are concerned, what you hear comes also from the Lyric preamp. The Lyric's preamp have two stages of compression in the lower register and a third one in the midrange.

(see LR Baggs patent https://patents.google.com/patent/US20140205106A1/en)

The patent says the multiband compressor works as explained in the following:

* STAGE1: one soft peak limiter (20-250Hz):
"The frequency selective soft-knee peak limiter has
two effects on an audio signal. First, it reduces any low
frequency thumps caused by hitting the top of the guitar with
too much force. Secondly, it allows the low end to instanta
neously compress and release as the guitar is strummed
harder and then softer. This is tuned to mimic the low fre
quency clamping that is done mechanically by the body and
mechanics of the instrument and keeps the tone tight in the
sub low frequencies when played lightly."

* STAGE2: A RMS compressor (20-200Hz)
"An RMS compressor can be used to reduce the
potential for “bloom” feedback at the natural resonance fre
quencies of a guitar or other instrument. In several embodi
ments of the invention, an RMS compressor includes an RMS
detector and a compressor. [...] the compressor is designed
to be fast acting (i.e., having quick attenuation above the
threshold) and slow releasing with a soft knee. This is desir
able to track the dynamics of an acoustic guitar or other
acoustic instrument. The compressor is used to change the
complexity of content at low frequencies by adding a small
amount of non-linearity to the low end that results in added
harmonics in the midrange and better resistance to feedback.
The compressor is run by an RMS detector as opposed to a
peak limiter so that it acts much more quickly on strummed
chords than on individual notes. This allows the compressor
to ride with the dynamics of the notes being played without
clamping the energy and musicality of the note."

STAGE3: Mid-frequency compressor near the "quack-shrill" frequency:

"Human ears are typically most efficient at perceiv
ing sounds around 3 kHz and are sensitive to high volumes in
a band adjacent 3 kHz. Loud sounds in that range can be
perceived as extremely harsh. In various embodiments of the
invention, a mid frequency compressor is utilized that is very
band selective at around 3 kHz and/ or generally affects fre
quencies in the band 3-6 kHz, but primarily 5-6 kHz."

I think LR Baggs did a great job there. I think the Lyric sucess is mostly due to those compression stages. They kind of recreate the "feel" of the guitar.

Note that in my case this is were the choice of position 3 and the Gilgaus mod have more impact.

1) Changing the position of the Lyric affects the top vibration modes you are sensing the most and thus their frequencies (see picture)

So of course you get more of that and less of this... and at the end, it changes what frequencies interact the most with the compressor.

2) The gilgaus mod (I am not sure of that, because I did both mods at the same time) may attenuate the signal that the lyric preamp expect to get. If it is true (but only Gilgaus can answer) then my guitar would be less compressed than a standard Lyric wiring. But I have no proof of that.

Note that the complex electronic is probably the reason why the Lyric is "picky" with regards to guitars. All those compression and EQ stages are probably tuned for a Martin D28 or HD28...

Cuki