Crescendo/Diminuendo One Note

A list of Advantages for the use of Linuxsampler with Musescore.

1. More sample choices. Linuxsampler can make use of sf2, sfz, and giga files. You can even mix/match the three formats. For instance, you can have and SF2 trumpet, a SFZ trombone, and a GIGA piano all working at the same time.

2. Layering. You can set multiple instruments on one channel. For instance, you can load up four individual french horns and set them all to the same channel so you can create your own horn section, if you only have solo instruments.

3. Better rendering. Some soundfonts don't work particularly well with Fluidsynth, yet all the soundfonts that I have had sound issues with in Fluid have worked well with Linuxsampler.

4. Better performance. If configured properly Linuxsamper is one of the most efficient samplers out there. I have Gigastudio 3 and GIGA files load faster and work better with much less CPU in Linuxsampler than in Gigastudio. The same is true of SF2.

A list of Disadvantages for the use of Linuxsampler with Musescore.

1. Not native. Has to be patched to Musescore via the Jack Audio Connection Kit.

2. GM soundfonts don't seem to work. You must have soundfonts of individual instuments to utilize SF2.

A list of Advantages for the use of SFZ vs. SF2 that require some work from the Musescore user.

1. Key switching. This is of course accomplished with the now infamous invisible note. However, the user can already make use of some articulations in certain instuments in Musescore. A midi volume controller does not create a true crescendo or dimuendo. Using a midi volume controller, especially since most SF2 are single velocity samples, only boosts or diminishes the volume of the given sample. If you were to play a recording of a violin on your home stereo, and you turned up the volume on the stereo when the violinist was playing pianissimo, it would not make you think the violin was being played fortissimo. All you hear is a soft tone being played at a loud volume, as the tone, articulation, vibrato, and bowing or the violist changes at different volumes. The same can be said of the trill. If Musescore were to add playback of a trill, it would be through the manipulation of the samples (either modulating the sample up to the next note of the scale or alternating between two samples). Thus the true articulations of the violinist would be lost, as the trill was "faked" just as the volume was "faked" in the above example. Here is an example from my library of a violin playing a trill and then a decrescendo on a sustained note. These are real samples of a real violinist performing these articulations. You will hear changes in vibrato, bow noise, et. al. This sample was accomplished using nothing outside of the channel changes (Pizzicato and Tremolo) supplied by Musescore. I don't have keyswitching set up yet, so I thought you would like to hear what is possible outside of any "hacks."

http://www.fileswap.com/dl/Ihoh4Gh5h/

Of course with keyswitching you are not limited to just picking three articulations, however, an extra note will have to be entered to trigger the key change and yes, it will have to be invisible not to appear in the score and it will have to appear above every note for which an articulation change is required.

Most professional notation software comes with a default library, such as Finale with Garritan Personal Orchestra, so they are able to add keyswitching into the notation editor itself. When you click on a trill in the notation editor it sends the keyswitch back to the sampler engine so the user can switch from one articulation to another with the simple click of a mouse. That is of course the ideal way for keyswitching to work with notation software. My invisible note was the only work-around that I could think of, and I truly don't find it as difficult or time consuming as using Musescore's current channel switching by way of staff text, as this certainly takes more time and involves more steps than the invisible note.

A list of Advantages for the use of SFZ vs. SF2 that require no work from the Musescore user.

1. One-shot loop mode. This has already been demonstrated in the harp clips above; however, here is another example this time with a Marimba playing 16th notes. The original samples were recorded in a concert hall so no reverb has been added by me. It will sound as though more reverb has been added to the SFZ because the notes will be allowed to sustain and the reverb tails will not be artificially removed.

SF2 --- http://www.fileswap.com/dl/zgP5j4zTrt/

SFZ --- http://www.fileswap.com/dl/g40DJFpfhW/

2. Round Robin. This is where more than one sample is used for a given note. Although it is useful for percussion it is, by all means, not limited to it. For example, you may have four samples of a violin playing C5, so you can set up a sequence in the SFZ file so that the sampler alternates through the four samples. A violinist will never truly articulate or vibrate a note exactly the same in a real performance, so when you play samples that only have one recording per note, it is obvious that you are using samples, because the note sounds exactly the same every time it is played. Say you have four quarter notes on C5 all in a row. The SFZ will cycle through the four samples so that no two notes are exactly alike. Round Robin is useful for percussion to help avoid the dreaded machine-gun effect. Below are two examples of the snare drum from my library which has true left hand and right hand samples. SF2 can only make use of one sample so it repeats one sample over and over. I have artificially panned the two samples in the SFZ to slightly left and right, so that you will hear that indeed two samples are being used and alternated. In the below clip, both round robin and one-shot loop mode are used in the SFZ.

SF2 --- http://www.fileswap.com/dl/EE4HN6Nfq/

SFZ --- http://www.fileswap.com/dl/tJAn7tsBcg/

3. Release Samples. You can add release samples in SFZ so that you can hear the instrument actually release the note. This is useful for such things as a jazz piano piece where as much realism as possible is desired. This is a subtle thing so you may have to turn up the volume or use headphones to hear this. I have scored out the classic jazz 6/9 chord with major 7th on top, and as the chord is released you will actually hear the release of the keys/hammers.

http://www.fileswap.com/dl/nkKoSXXn9/

4. Modulation, EQ and other advances. This is where it gets impossible to explain all the advantages of SFZ, as they are complex and too numerous to go into here. With sfz you get a filter unit, an amplifier, an EQ and a pair of effects sends, so the possibilities are incredibly numerous. For instance, you can set up velocity filtering where you can blend the transitions between velocity layers. My library will switch from the MF samples to FF samples at a velocity of 93. So at 92 the sampler is using the MF sample and at 93 it will use the FF sample. This can cause a "jolt" in the smoothness of a crescendoing or decrescendoing line. The filter can smooth out the transition so that it is barely perceptible that a change in samples has occured. EQ can be added to the samples if needed, and in this case the SFZ when translated to SF2 would sound completely different even for the basic sustain patches as all the equalization would be lost. As I have said, it is too difficult, time consuming, and technical to go into detail here, but the possibilities are astounding.

5. Probably the biggest advantage to SFZ is that it is a living, actively developed framework upon which new features are constantly added. There is no such hope for SF2. It is has been nearly a decade since it has been developed which, in computers, is a long time. Think about the computers and software of 10 years ago and compare those to today's. Then look at my number 4 above. There is nearly 10 years of advancement there and that is why it is too difficult to explain, especially to the layman.

After all this has been said, I will add a few key points. I have tried to avoid using some of the things available in SFZ that could not be translated, as I have expected that people may want my library in different formats. For example, I remastered the original samples in my library so that the samples themselves have been equalized rather than using the equalization available in SFZ, so that if my library were translated to SF2 the basic sound would remain relatively the same; however, there are a lot of things that have to be done within the SFZ as there is no other way to accomplish them and many things that SF2 simply cannot do not matter what provisions are made by me. The velocity filtering that I mentioned above will not be translatable, so the samples will still sound slightly different especially around the pivot points of velocity and the "jolt" mentioned above will still exist. Basically something will be lost in just about every instrument when translated to SF2. Some things will be highly noticeable such as in the harp, marimba, violin, and snare samples above, and some will be less noticeable.

A list of Advantages of SF2 format versus SFZ for the Musescore user.

1. Native.

* I will add that the Fluidsynth team is adding SFZ support to Fluidsynth, so at that point I would assume that SFZ will become native and all the above advantages of SFZ will be available to the Musescore user natively.