Welcome, Guest
Username: Password: Remember me

TOPIC: Gibson Les Paul soundfont

Gibson Les Paul soundfont 1 month 4 days ago #1

  • Bernhard Trummer
  • Bernhard Trummer's Avatar
  • Offline
  • New Member
  • Posts: 3
I recently sampled my Gibson Les Paul guitar for a private music project. Since the final result will be released under CC BY-SA 4.0, I decided to also release this soundfont. So here we go...

Download link: (Google Drive)
Version: 1.0, released February 2020
License: CC BY-SA 4.0
Size: 917MB (zip download), 1.87GB (unpacked)

Description:
The guitar is a Gibson Les Paul Standard, produced in the year 2001 (serial no: 02611484).
The strings are D'Addario EXL115 Nickel Wound (medium gauge)
All strings are tuned one half tone down (e.g. the E string is on D#).

A capo was used to hold down the strings on the respective frets.
All non-picked strings were dampened with some tissue, making the recordings as dry as possible.
The bass was recorded directly using a Behringer UPhoria UMC202 HD device.
Also a wiring hack was applied to the guitar in order to record both pickups simultaneously.

The original samples were recorded at: 44.1kHz, 24bit
I didn't use the loud samples for the soundfont.
So I digitally amplified the selected samples by the factor 2
and reduced them to 16bit depth so the total soundfont size stays within the 2GB bounds.

The sound font contains two banks:
  • 000:000-012:
    The preset name indicates the fret range.
    For example: "07-11" will contain the sounds of fret 0-11 of the low E string,
    7-11 of the A, D, G, and B string, and 7-13 of the high E string.
  • 001:000-003:
    The preset name indicates the used string.
    For example 2_G#1 will contain the sounds of all frets (0-13) from the A string only.

The "sweet spot" for the MIDI velocity lies at 110 +/- 3.

Used software:
  • Ardour: for recording (one wav file per tone, each containing some samples with a different picking strength)
  • A custom written sample cutting tool[1], sox, and some bash script magic to:
    • cut out the samples
    • apply a 2ms fade-in using a half sine wave
    • apply a linear fade-out for the last quarter of the sample
    • detect the peak level and rename the sample accordingly
  • Polyphone: to create this soundfont

How to use properly:
  1. The stereo wave output of your sf2 synth will have the following channel mapping:
    • left: signal from neck pickup
    • right: signal from bridge pickup
    This way you can choose an arbitrary signal processing for your mix:
    • Just take the signal from one pickup and ignore the other.
    • Mix both signals together (in phase or phase inverted) and apply your processing afterwards.
    • Apply different signal processing to each individual channel before mixing them together.
  2. Try to remain near the MIDI velocity value of 110, because in this area the sample density is high.
  3. Use a little velocity randomization on your MIDI track. This way different samples will be used for the same tone which makes the track sound more realistic.
  4. If you want, you can also add some timing randomization.
  5. Use a separate FluidSynth instance (e.g.) where you can boost up the gain/volume without boosting it also for your other MIDI tracks. The loudness of this soundfont is a bit low compared to standard General MIDI soundfonts.
  6. Apply plugins (e.g. Calf Equalizer, Calf Compressor, Guitarix, etc.) to shape the sound like you want. Remember that the output of the soundfont is just the dry notes directly recorded from the pickups.
The administrator has disabled public write access.
Time to create page: 0.103 seconds