Another methode how to import soundfonts

[siebenhirter]

Hello,
lately I've received some inquiries about inadequacies in importing sf2 samples and how to work around them - I'd rather answer that here than PM.

Some basic information:

Sf2-files are structured as sf2-bank, sf2-preset, sf2-preset-zone, sf2-instrument, sf2-instrument-zone, sf2-sample, all memorized in RIFF-format.
Audio data are (WAV, PCM) are located in the stda chunk of the RIFF, the playback parameters described for the instrument and presets are in its pdta chunks.

Soundfont bank samples in PCM format are mapped to sections on a musical keyboard. Synthesis parameters are described in instruments and presets (loops, vibrato effect, velocity sensitive volume changing etc) and are capabilities to render audio content to get the quality of a professional digital sampler.
*
sf2 files therefore are divided into two essential parts:
- part one of sf2-files contains audio samples in PCM format
- part two of sf2-files contains describtions of playback synthesis parameters
*
Software that will open / convert or fix SF2 files mostly only use the first part - the DEFINED GROUP OF PCM AUDIO SAMPLES.
Simple sf2-files have an empty second part and do not contain PLAYBACK PARAMETERS for instruments and presets. Such sf2-files are ease to convert, because only a simple separation of audio files and non existing descriptions of the playback parameters is necessary.
*
To completely import of more demanding sf2-files only make sense if all PLAYBACK PARAMETERS of sf2-soundbanks are considered. Most sf2-converters - like import function of Korgs Pa-keyboards - ignore data in describtions without converting them, because it is difficult to automatically convert / implement playback parameters for sound systems / generators.

With more demanding sf2 files, however, playback parameters are mostly responsible for high quality properties, so it is necessary to MAKE ADJUSTMENTS OF PLAYBACK PARAMETERS for Korgs oscillators with built-in soundeditor..

Also with YEM sf2 can be imported as content of Yamahas packs, but it is the same - it may not be possible to accurately convert all of the data from .sf2 files.

If imported data does not produce the expected sounds, editing functions are necessary to make adjustments to take into account play parameters (overlapping sample areas, envelopes, filters, lfo, modulators etc) as described in sf2.

I do not believe there are sf2-converter which satisfactorily implement all play parameters as expected for the produced sounds, but maybe YEMs software’s editing functions are somewhat more comfortable to make the necessary adjustments, than is the case with Korg's built-in sample software of Pa-keyboards.

For me a better way to handle KORG for using Sf2 files for Korgs Pa-Keyboard is to convert sf2 to sfz format - therefore I prefer the use of "Polyphones" export function:

Load sf2-file and export it as sfz-format and you get all samples in a separate folder in wav-formats. Wav-format seems to be better suited to be imported with Pa-Keyboards (also possible with previous Pa800/3x).

Exporting sf2 as sfz also generate a sfz-file, that can be read like a txt-file, in case you open it with texteditor.

After you get some experience how to interpret the parameters of this sfz-textfile, with values of that parameters it is much more easier to make all necessary adjustments in soundeditor of Pa-keyboards.

[NativeAngels]

Thanks for that most helpful. Just a shame we don’t have most slots for sounds,

[Asena]

Thank you, As always you try to help, And Thats kind, But If you convert to SFZ , how will the PA models read all sounds in once, and All velo layers, Thats the reason why we need sf2, Since there is no more option to pick a sound and load with all Layers and velos.

[josf.kumar@gmail.com]

Hello,

I found the following website allowing SF2 soundsfonts to download for free.
https://www.producersbuzz.com/category/ ... fonts-sf2/

Hope someone finds it interesting to download them and experiment with installing those sounds in their Korg pa 1000.

With Regards
D. Joseph
India

[siebenhirter]

.. how will the PA models read all sounds in once, and All velo layers, Thats the reason why we need sf2, Since there is no more option to pick a sound and load with all Layers and velos.

Hello Asena,
your remark is misleading, because by importing sf2 instruments you also cannot read sounds in once for Pa-keyboards - and especially not including all layers and velos.
*
Since it's not easy to explain, allow me to go a little further - what is meant by a SOUND in this context?!

A playable user sound is a PCG file in the user bank of a Pa keyboard. With the sf2 format, a playable sound is an sf2 PRESET. The import function does not import sounds, i.e. no SF2 presets, but rather sf2 INSTRUMENTS for saving multisamples from these instruments.
*
No sound is picked with sf2-import. These import-tool pick sf2-Instruments to create and prepare multisamples - nothing else. Multisamples are what Pa keyboard sampler requires in order to be able to create PCG sounds with its sound editor.

Neither all sounds are imported in one go nor Velo Layers are read, because Velo-Layers are not taken into account at all during import - and parameters of sf2-presets generelly remain unconsidered.
*
The on-board import tool is suitable for generating multisamples from sf2 instruments programmed without overlap in one go. However, no VeloLayers are transferred, but rather the PCM samples are arranged according to their key ranges in the key zones of the multisamples.

The zone parameters of the samples are transferred: original note, start/loop/end, level, pitch, from-to (key range) - but nothing else.

By the way, Level has nothing to do with velo layer and actually should be called attenuation, because these parameter adjust the relative level of the selected zone - its value can only be a negative offset to the overall volume of the multisample.
*
No further sf2 parameters are transferred automatically; these can only be entered manually when creating the sound and taken into account in the oscillators. However, individual parameters for envelopes, LFOs, filters, amplifiers, reverb, chorus can only be assigned to a single sample in sf2 format / for sf2 environments.

The same applies to the parameters of Sf2 modulators that allow programmable real-time control of the sound. The parameters entered for modulators of sf2 instruments and sf2 presets for remote-controlled chorus, reverb, filter cutoff etc only can be taken into account with the sound editor.
*
CONCLUSION: After an "successful" LOAD of an sf2 instrument the import unfortunately gives the user the questionable assurance that all parameters have been transferred correctly to produce a Pa-sound.

But that is not the case, because the significant parameters of the sf2 presets are not imported at all!
If you want to create a pcg sound corresponding to the sf2 preset, the parameters of the sf2 presets must be set manually using the sound editor in addition to the imported multisamples.
**

.. If you convert to SFZ , how will the PA models read all sounds in once, and All velo layers...

As mentioned above also with the sf2 import tool that cannot be done!

But you are right to need sf2-import without using sfz-conversion.
Simple Sfz-conversion with software "Polyphone" I prefer for control not to loose that sf2 parameters mainly programmed in sf2-presets and to get represented real number of instruments necessary for the existing layers.

You do not get these infos with monolithically defined sf2-structures, where PCM-samples and definitions reside in a common file, whereas with non-monolithically SFZ-structure definitions of the parameters (sfz) and samples are separated.

With some experience with SFZs parameter list it is clearer how to prepare sf2-data before and for an satisfactory import and how to create sounds with its imported multisamples.

To prepare sf2-files before importing often is necessary due to deep differences with KORG’s own format with:
- Instruments with overlapping zones
- Instruments with L/R samples
- Instruments with velocity layers
- Instruments with few parameters but mainly programmed with parameters in presets etc.

.. I found the following website allowing SF2 soundsfonts ... Hope someone finds it interesting to download them and experiment with installing those sounds in their Korg pa 1000. ...

Hello D. Joseph,
most of the soundfonts you mentioned for download have been available on the www for about 25 years and are all very simply programmed (such as in feb_Top_14_Free_Flute_Soundfonts).
They have no synthesis parameters, but consist exclusively of the PCM samples, so at least with that soundfonts there can be no problems with the sf2 import.
Due to the simple structure of such sound fonts, you can easily experiment with the synthesis parameters using the Pa-Keyboards sound editor.

[MusicLover400]

in the SOUND mode , in order to create new keyboards by adding sound samples using 12 oscillators , you can add sound samples by typing their numbers on the SOUND mode screen. Where can I find these sound sample numbers. couldn't find them in the users manual . Thanks for the any information you can provide.

Al

[siebenhirter]

in the SOUND mode , in order to create new keyboards by adding sound sample ..

Soundmode use multisamples - you can find multisamples in "Muscial Resources" page 1011 in pa1000 manual (table contains factory multisamples)