The MuseScore SoundFont Group Sample Specification

• May 19, 2012 - 18:26

MuseScore SoundFont Project

Sample Format specification

Samples should normally be presented in 16 bit mono 44.1 kHz format.

This will ensure compatibility with all systems.

For SoundFonts stereo samples are not advisable as the presets may be made up of several copies of instruments placed in different parts of the stereo soundstage, so for this to happen effectively mono samples are required.

16 bit samples are preferred as the native storage format of the SoundFont stores samples in 16 bit words. 24 bit samples can be used, but this involves storing each unit of the sample in 2 words rather than 1, thus both increasing space and processor overhead.

The SoundFont format can process samples between 400Hz and 50kHz, however, the 44.1kHz CD quality sample-rate ensures playback compatibility on all systems, and there is very little to be gained from increasing this to 48kHz in terms of sample overhead.

There may be exceptions to this for some instruments eg pianos and organs do occupy width in terms of the stereo soundstage, so some soundfonts for these instruments could benefit from stereo samples, although this could be managed artficially from within the soundfont.

Church Organist May 2012


Comments

Seems solid to me. I assume this is just about the sample format itself, and other issues - like how many sample points, how many velocity layers, etc - are separate, and more instrument-specific anyhow.

In reply to by Marc Sabatella

so I'm reading this from a perspective of very little knowledge about how soundfonts are created, so my questions are at a rudimentary level. By "samples should be presented", do you mean sample files that are submitted by contributors? Is it implicit that contributed sample files should already be in soundfont format? (as opposed to a collect of .wav files, for example?)

In reply to by mtherieau

Yes, "samples should be presented" means sample files submitted by contributors.

It would be counter-productive to receive samples in soundfont format.

They would only have to be fished out of the soundfont in order to be added to another, and thus possibly suffere from degradation in the process.

Any PCM format should be OK, bearing in mind we are all using different platforms.

My own preference is for WAVs as I'm working on Windows and Viena expects them in WAV fornat, but Audacity will read AIFF and FLAC amongst other formats.

In reply to by ChurchOrganist

I see -- thanks. I promise you I'm not being intentionally dense -- I'm thinking questions like: "If I were to contribute samples, how would I go about it?".

btw: I remember a blog entry about how to create a soundfont and one of the recommendations was to start with a good naming convention for the individual sample files and make sure you stick with it (darn -- just tried to find that link but cannot).

I also saw some interesting links on tools/utilities that aid in processing .wav recordings, for example: Learjeff's Nerd Soundfont Tools . I'd certainly be interested in helping out with these types of tools, if desired.

regards,
markt

Do you still have an unanswered question? Please log in first to post your question.