Computers & Music:
by Kai Pihl
In the previous section we've handled percussive sounds, eg. sounds that happen in a quite short time. These sounds can not be elongated in time because of their nature. To this group of sounds belong piano, harp, vibrafone, marimba, drum sunds, etc.
There's also sounds we can make longer just letting the sound continue. They are called sustained sounds. Typical instruments that give sustained sounds are organs, strings and horns. The length of sustained string and horn sounds are restricted by human characteristics.
A sustained instrument sound has in general all the same parts of volume envelope than the percussice instruments. However, there is a new phase, the sustain phase. In that phase the sound of a sustained instrument sound stays allmost the same. In violin sound this corresponds to the bowing, in organ sound it means keeping the key down and in horn sound the player continues blowing with the same power.
In the phase of a such sustain, the sound of an instrument consist of successive sound waves which are almost similar to each other. Examples of an individual violin sound wave at the sustain phase can be seen in the picture below. The upper picture is a capture from the Vienna 1's loop window; the lower picture shows the same sound recorded and viewed in a sound editor.
In the upper picture the blue steps in the sound wave corresponds to the individual samples of the SF-bank sound. In the lower picture EMU8000 has made it's interpolation and put addional samples between the original ones. As can be seen, the shapes of waves are similar except that the interpolated sound wave is smoother. In this axample the beginning of the wave has the same amplitude and slope than the end. When this kind of wave is played in loop it sounds like a continuing sound. If the amplitude or slope differs in the joint point (eg. the end and the beginnig of the wave) there's going to be heard a glitch.
This single wave has been selected from the end of the recorded violin sound (exactly said, it's the part between the green vertical lines). The endpart of the sound can be cut of, because the release phase uses automatically the sustained part to continue the sound. Let's check out how we can set the loop points of a sound sample with the Vienna 2 SoundFont editor.
In Vienna 2.0 there's a special 'Loop settings' -window for adjusting the loop points of a sustained sound. Doubble click on a sample name in the Instruments folder and you get into the 'Loop settings' dialog. It looks like this.
The upper window shows the entire record. This window is zoomable and in the picture it is used to show the area of loop part of the record, the same we had as an example above. In the lower window we can see the loop ends combined together like in model below.
As you can see, the wave continues in the loop from it's end to the beginning as smoothly as in the original wave form. In fact, there is at the end of the sound - outside the sustain loop - four samples that are equal to those in the beginning of the loop (see the upper window). These samples can be of big value when EMU8000 performs it's interpolation pitch shifting the sound. If your recording does not contain this type of additional samples, you can copy them from beginning of the loop. Remember that the interpolation method used few neighbour samples to fit the curve exactly between them. If there's no further neighbours (as in the case of loop end) the curve can get a wrong shape. That's why it is important to have the same shape of the curve right after the end of the loop than in the beginning of the loop. Four or five samples should be enough.
To be continued soon...