How to make a sample (music) for a soundfont


There are a lot of documents telling you how to build a soundfont, but few describe the process to build the samples you put into the soundfont. This could be because the sample is the heart of an artist’s creation, for which every person is very jealous.

But because I am a poor artist (: (musical creation is my hobby and not my work), I can talk sincerely, without worries to damage any of my professional chances. 🙂

First of all you must download the Viena soundfont editor, it’s free and useful for whom has rudiments of electronic musical creation.

Now I want to describe one of my experiences in creating samples. It’s only an example of what can be reached with a little bit of imagination and, if anyone knows a richest experience and wants to share it, he may write to me using the form at the HOME page of this blog. I’ll be happy to add his experience close to the mine.

breaking line

The sample from field recordings

I took a pair of woods from the musical shop, those used to follow the latin-american rhythms. With them I went through the home, percussing and registering all the things they seemed to have a musical tone. The choice stopped near two natural instruments: the radiator and the cage of the ventilator. On both of them I registered many scrapings with the woods mentioned above.

Then I cleaned the sounds with a sound editor, I choosed the one it sounded better. I stretched the sound queue with a tool, until it could maintain the same frequency without to change the tone. I saved it. I doubled the sound and reversed the second one. With the sound editor I attached the reverse and forward sounds putting the composition in loop. You must correct your sample until the loop is missing of clicks. It happens when the sound reaches the end and it starts the sequence again. It’s not easy to obtain it, but if your sound will be an atmosphere, you can insert a micro fade-in at the beginning and a micro fade-out at the end, just a 10 milliseconds, without worries to listen a sound hole during the play. Choosing to start the loop with the reverse piece and ending with the forward one, having the most intensive sound at the centre of your sample, you could find an easy way to a successful sample. You will adjust the centre of the loop, in which are the sounds adjacent, with a micro cross-fading. Now you have the loop.

Playing your loop and using an instrument, try to find what is the note played by your sample. Listen to many times how it plays and, if you think the sound is too much slim or fat, put the sound in a track of your sequencer (I use Cubase, but there are many free sequencers), add a track and play one note all along your sample, giving to it the right consistency. The added note must be one of the harmonical chord played by your sound, for example, the fifth higher if the sound seems dark, or the fourth lower if it seems shrill.

Now I want to tell you the story about the “palo” (pole) sample, one of the instruments in my Falling_Stars.sf2 soundfont.

  • I percussed many times the pole of a country street, in a day without wind and birds noise
  • At home I took the recording and chose the best sound: clear, crystal
  • With a fade out I cutted the hiss at the queue of the sound
  • I stretched the sound, I faded out the queue, I stretched the sound. All repeated several times
  • I cutted a piece of the starting sound. Being a percussion, the starting sound is often distorted
  • I multiplied the sound in three, four tracks. Each sound starting after a while and with a different fade in. I obtained a long sound with a slow cascading velocity and no hiss.
  • I put a fade in at the beginning, just to smooth the start

Once I reached a good performance, I saved it as a sample and I loaded it to the .SF2 using Viena. But when I went to play the soundfont in a VST instrument, it was dull, empty. So I added many layers of the same sound in the SF2 instrument with Viena editor, as you can see to the image below.Instrument configuration with VienaEach layer is using the same sample but with a different root key (104 92 80 112 109 97). The last three layers at the right, introduce a harmonic concept, because they start with a root key (112 109 97) at the third and fifth tone of the original sample.

Furthermore, you can see that the last three layers have a different volume envelope attack (1,00 1,25 1,50). The effect is a harmonic sound it fills gradually the ears. The high root key of those last three layers have the effect of a low pitch harmony always following the singing high pitch part of the same sound. So, using a unique ugly sound, but multiplying the layers, I obtained a harmonic fat sound, good enough to be listen to the human ears.

The image below represents the graphic evolution of the sound: above the original stereo field recording and below, the final sound loaded into the SF2 bank.the evolution of "palo" sound


The sample from human voices

It is very difficult to play human voices, at least you would record one sample for each note of your keyboard, otherwise keep in mind that you can spread only one octave lower and no one note higher. You need to find a singer who keeps a long single note, without surrounding music. You need also to avoid commercial singers, because their voice is copyrighted.
I chose a singer from an unknown chorus of an unknown village, better if you have a friend who masters a good voice and he wants to give a gift to you with a 7 + 7 scale notes.
Sample the voice, eventually put in sequence two or three vowels: aaaaa eeeee ooooo; the result will be more interesting.
Put in loop the sample and clean all the clicks, then save it.


The sample from vsti synthesizers

Choose a vsti it plays an interesting sound. Avoid commercial synths because they are copyrighted. I choosed ZynAddSubFx by Octavian Paul Nasca, very good synthesizer, multi timbral and open source.


I must spend a few words about it. It is a synthesizer oriented to Linux operating systems, but there are some old versions running very well with Windows and so stable that continue to be stable and well performing also on my 64 bit Windows7 operating system. Actually I’m using ZynAddSubFx standalone Windows version 2.2.0, good to create new instruments, quickly play and register music and sounds, and when inspiration is running.

downloadZynaddsubfx Windows standalone executable v.2.2.0
installation package for Windows7 and previous

This version has a virtual keyboard with which I play almost randomly. Otherwise I play using the keyboard attached to my computer and well recognized by ZynAddSubFx. When I find an interesting combination of instruments, I save it and close the standalone program.

[How to install ZynAddSubFx 2.2.0 on Windows8 and Windows10]

Unfortunately, when you try to install ZynAddSubFx on Windows8 and Windows10, you’ll receive an error because some DLL libraries are no more on the new operating systems. Don’t worry. You must simply manually register the DLL. Please follow these instructions:

  • copy Setup_ZynAddSubFX-2.2.0.exe into an empty folder
  • using an UnZip program (I’m using the free 7-Zip) extract the content:
    • right click the setup file + choose 7-zip + extract to here
  • delete the folder $PLUGINSDIR
  • into the folder $SYSDIR you’ll find the DLL: pthreadGC1.dll
  • Copy pthreadGC1.dll to C:\Windows\System folder (not System32 folder!)
  • Open elevated Command Prompt
    • move mouse pointer to lower-left corner,
    • right-click on the Windows 8 or 10 start thumbnail
    • and select “Command Prompt (Admin)”
  • Change directory to C:\Windows\System path (type cd C:\Windows\System);
  • then register the DLL typing the command: regsvr32 pthreadGC1.dll
  • close the command prompt typing: exit
  • now you can launch your ZynAddSubFx
    • do not double click on zynaddsubfx.exe
    • always launch the procedure: launch_zynaddsubfx.bat

downloadmy ZynAddSubFx Instruments (Last updated 7th March 2018).

I open my host player: Cubase by Steinberg, well known program. It doesn’t need words of explanation or, in any case, not in this document. I’m using with Cubase the ZynAddSubFx vsti version 2.4.0. You can play it only within a host player. It is not the last version, it’s almost old, but very stable. It runs better than most of other vstis created with SynthEdit, the software used to develop Cubase synthesizers. The stability is the reason because I don’t want to change version if not necessary to my interests. I create an instrument track and I load ZynAddSubFx in it.

downloadZynaddsubfx vsti DLL v.2.4.0 installation package.

Once configured the track and the instrument, play one note for the entire length of your sample. I choosed to build samples each one of four bars at 70 bpm, it is almost wide to unchain the inner musical fantasy. Add something like a bell sound, or some simple notes just to give an effect of animated sound when it will be played. But pay attention, do not fill the sound of a lot of variations because when played it will result in noisy and confused sound.

So you can play a chord in one single track, but you can play also a chord in one track and a similar chord in a second track.

Just for example and only for a simple example:

  • – C+ in the first track (E,G,C)
  • – a- in the second track (E,A,C)
  • – F+ in the first track (F,A,C)
  • – a- in the second track (E,A,C)

Than do a fade-out to the entire first track and a fade-in to the entire second track. The result will be a moving athmosphere, to play in one or two notes simultaneously, but not more. Put all in loop, clean the clicks using micro fade-in and micro fade-out, and the sample is ready.

One good effect is to add a reversed cymbal to the sample, it will give a shake to your attention each time it plays.


The sample technique

The Viena soundfont editor is well described in its documentation and also through a document “SF2-Tips&Tools” that you can find in the HGF forum of KVR Audio.
When you load a new sample with Viena to the soundfont, follow these instructions in order:

  • add the new samples
  • to the sample parameter, set the “loop S-E” to the length of the entire sample (the “size” parameter will tell you the length
  • to the sample parameter, set the root key. Usually to what was your primary note in the sample, but you can also assign a higher root key
  • add the new instrument, give it the name and assign the samples to the layers
  • to the instrument, assign each sample to the right key range
  • to the instrument, assign the “sample mode loop”: 0=do not loop, 1=normal loop, 3=loop and play to end on key release
  • add the new preset, with the same name of the sample
  • to the preset, flag the instrument to use in it
  • to the preset, assign 0-127 to the key ramge
  • save the soundfont, that’s all

At least you would have two samples to assign to the keyboard of your new instrument: one playing near the note C4# and the second playing two octaves lower, near the C2#. Better is to save the sample in mono format. Usually I do a .wav audio file, 44100 khz, 16 bit, mono. For what concerns the assigning to the virtual keyboard in Viena, keep in mind that is better rendering a sound assigned to a higher octave and then played with a lower octave, and not the contrary: a sound assigned to a lower octave and then played with a higher. So, the C4# sample will be assigned from maximum (127) to D2, and the C2# sample will be assigned from C2# to minimum (0).

breaking line

The sample player vsti instrument

It is with big sorrow I learned about the desease of Guenter Fortune. After a brief but severe illness, Guenter passed away November 2014. His site disappeared October 2015, the synthsizers are no more available from there, but they are still sparsely available through sites collecting the most interesting synths to download for free. Here’s one of them:

HGF synthesizers had inspired a part of my recent musical production. Through his forum is still available a brief document explaining how to use Viena, the SF2 SoundFont editor to build banks for HGF Synts.

Here’s below a list of the HGF synths better performing.

Alionoctis Pro loads SF2, not very stable. Useful for space, horror, mysterious sounds
Anvilia Pro loads SF2, almost stable, rather simple
Arracis Gold Pro loads WAV files, good arpeggiator, useful for catastrophic athmos: desertification, overpopulation….
ASW Pro loads SF2, stable, clean sound. The best performing synth
Atonoise Pro loads SF2, not stable, but very good for noisy, horror, mysterious sounds
DrumBurst Pro loads SF2, very good beat machine
Percumat 2 Pro loads WAV files, very good beat machine
Protoplasm Pro loads WAV files, good for strange and alien athmos
ScapesWizard loads SF2, not stable, sometimes the sound banks are muted, but good machine for shaking alien athmos
Shuniji Pro loads SF2, rather stable, excellent mystery sound machine, the best after ASW
STS21 Pro loads SF2, rather stable, good performig
STS24 Pro loads SF2, rather stable, good performig
STS26 Pro loads SF2, rather stable, good performig
STS33 Pro loads SF2, rather stable, good performig
Swamp XT Pro loads SF2, not stable, but very noisy machine, for alien and catastrophic sounds
XWOF-4 Pro loads SF2, very very good rythmic sound machine, complete, well performing
Altair 4 no SF2 loading, but very interesting for sci-fi sounds
Alphatron no SF2 loading, but with an interesting arpeggiator
The Tiger Free no SF2 loading, but with an interesting arpeggiator

Just a few recommendations about the use of HGF VSTIs and Cubase.

  • First of all, save the project the most frequently as possible.
  • If you want to use more than one vsti tracks in the same project, you must freeze the actual track before to add a new one.
  • Sometimes the HGF VSTIs go wrong and the VSTBridge connection with Cubase will be interrupted. Don’t worry, just close Cubase and reopen it. It’s done because the HGF VSTIs are rather old, not recompiled for the new instances of Cubase and therefore unstable.
  • save the VST preset of all the HGF instruments before to close the Cubase project. I experienced that, once reloaded the project, the instrument configuration was lost.

With a bit of patience and some bad words when the application seems to drive on an aerostatic ballon 🙂 , the work can go succesfully to the end and you will be at least satisfied.

Now it’s time to make my experience available for free, hoping to do something that will fire your creativity and inspiration too.


Here’s my soundfonts, created with Viena and played with HGF synths through Cubase host sequencer and player.

breaking line

The Atonoise soundfont

Atonoise.sf2, initially designed for HGF’s Atonoise synthesizer, was gradually extended to all the HGF’s synths with stunning results.
It includes the following instruments:

  • - arabia amusing arabian like looping sound
  • - aramaic2 arabian like sound
  • - bells russian bells
  • - carillon carillon Christmas sound looped
  • - carillon_c4 carillon atmosphere
  • - chitarra guitar sound and atmosphere
  • - cicale cicadas stretched with added harmonics
  • - coyote tropical forest, bark of a coyote with twittering
  • - darkness dark and obscure atmosphere
  • - japan amusing japanese like looping sound
  • - mare sea waves
  • - metodio small strokes on a metal door with distorsions
  • - moscasino tropical forest, donkey and flies with twittering
  • - mosche flies and bees
  • - piano piano instrument atmosphere
  • - rane tropical forest, frogs and cicadas
  • - sitar indian sitar atmosphere
  • - sitar tender atmosphere

downloadAtonoise.sf2 (Last updated 21th January 2016)


The Falling_Stars soundfont

Falling_Stars.sf2 had its original name from the first sample created with Viena, but than, just keeping the same name, I added more samples to the soundfont. It includes the following instruments:

  • - arpeggio cascade arpeggio with an echo queue
  • - bombola shuffling of an old gas-cylinder
  • - caraffa crystal glass and decanter percussion and granulation
  • - chiavi home keys shaked with added harmonic tones
  • - chitarra guitar arpeggio
  • - cinese chinese like sound
  • - crashing object crashing granulation
  • - falling_stars ZynAddSubFx instruments transposed to soundfont
  • - flute expressive south american flute sound
  • - ghost ZynAddSubFx instruments transposed to soundfont
  • - gusli russian gusli (harp) played randomly
  • - hammers ZynAddSubFx instruments transposed to soundfont
  • - harp amusing irish harp looping sound
  • - lighting_stars ZynAddSubFx instruments transposed to soundfont
  • - metallo metal shuffling heavy stretched
  • - palo percussion of a metal pole along a country road
  • - piano piano notes
  • - planets ZynAddSubFx instruments transposed to soundfont
  • - presenze scary metal sound with added harmonics
  • - revells reversed bell sound with added tones
  • - robot ZynAddSubFx robotic sound
  • - secchio dustbin percussion and granulation
  • - spettro ZynAddSubFx instruments transposed to soundfont
  • - tenor russian bass and tenor voices
  • - termo shuffling one's radiator
  • - ventilatore shuffling one's ventilator
  • - vetro glass imitation with distorsions
  • - zufolo ZynAddSubFx imaginary flageolet or God's flute

downloadFalling_Stars.sf2 (Last updated 21tst January 2016)

The two soundfonts are work in progress. Periodically I’ll update them. So don’t forget to visit this page from time to time.


dancing notes

The music

Please note:  If you have troubles to listen the sounds when clicking the link below, then right click the link and choose “open”.

The sample had its origin by a percussion of my home dustbin. I stretched the sound cutting the first few seconds and I kept only the queue. There was a problem: the sound had a trend to low and high of tone, so I made the fundamental tone stronger by adding a perfect harmonic sound. I like it, but sometimes it seems too much distorted. It is played with HGF’ STS-21 Pro vsti, using mare, caraffa and secchio instruments.

The sound, produced with STS24 Pro by HGF and Atonoise soundfont, plays flies. The athmo is oniric, neither amusing nor rythmic, but just an athmosphere. Let you add your track sound, texture or fx, and make it your music. Good work 🙂

Thoughtful sound, produced with Shunijii Pro by HGF and Falling_Stars soundfont, instruments: metallo and arpeggio. The athmo is an athmosphere. It seems telling of a person lost in an alien dark planet, alone and cautious. It seems also a dream, something unusual, an unknown forest with an obscure drum in the dark.

This is a melting athmosphere. Were are we now? What land is it? The sound is formed by two tracks: the first is the rythmic base in which I used Anvilia Pro vsti by H.G.Fortune, whith samples: “chiavi” and “harp”. The second track is the singing trembling voice for which I used “c” instrument and ZynAdSubFx vsti. Then I added a lot of dark reverb and stereo delay.


New compositions, if relevant for a comment, will be continued in our next issue.


Effettua il login con uno di questi metodi per inviare il tuo commento:

Logo di

Stai commentando usando il tuo account Chiudi sessione /  Modifica )

Google photo

Stai commentando usando il tuo account Google. Chiudi sessione /  Modifica )

Foto Twitter

Stai commentando usando il tuo account Twitter. Chiudi sessione /  Modifica )

Foto di Facebook

Stai commentando usando il tuo account Facebook. Chiudi sessione /  Modifica )

Connessione a %s...