Multi-Layer Sampling

The simplest configuration, of course, is the Single-Layer patch, in which one sample is used for all velocity levels. This can work for sounds that are used infrequently in a piece, or where the sampled source is a synthetic sound. Both the NN-19 and NN-XT are capable of single-layer sample playback and each can play separate samples in different ranges of the keyboard (keyzones). In Reason, only the NN-XT is capable of having more than one sample assigned to an individual key. (Logic's EXS-24 is also capable of this.)

In a Simple Layer configuration, a second sample is assigned to the each keyzone so that pressing a key makes both samples sound together. This is done primarily to create a more interesting texture. To create an "L.A. Piano" sound, layer an Acoustic Piano with a bright DX-7 style Electric Piano. One big advantage of the simple layer over the single-layer patch is the ability to detune the two samples and/or pan them to separate locations in the stereo field. This can create a much broader and more interesting sound.

Single Sample Simple Layer Stack
Single-Layer, Simple Layer, and Stack Configurations.

A variation on the simple layer is a Stack, in which three or more sounds are layered per key to create a thicker timbre. For example, you could layer Trombone, Tenor Sax and Trumpet samples on each key to create a Brass Section stack.

The problem with a single-layer (or simple layer) configuration is that it is not very expressive. Most sounds change their intensity in multiple ways with a stronger note. For example, if you were to compare a loud note on the trumpet to a soft note, you would find that the louder note was:

  • Quite a bit Louder (higher in Amplitude)
  • Noticeably brighter in Timbre (a greater amount of the higher Harmonics)
  • More punchy in its Articulation (quicker Attack Time)
  • Ever so slightly sharper in Pitch (higher in Frequency)

This is where multi-layer sampling comes in. In a sampler capable of multi-layer patches, a key's MIDI Velocity is used to vary the output of two (or more) samples. There are four main configurations used:

  • Velocity Switching
  • Velocity Layering
  • Velocity Fading
  • Velocity Crossfading
Velocity Switching

In Velocity Switching, one sample is assigned so that its velocity level ranges from 1 to a moderate value, such as 85. The second sample is assigned so that its velocity level ranges from the next velocity level (86) to 127. Incoming velocities of below the split point (85) will trigger the first sample; velocities above it will trigger the second sample. If the second sample is a little bit brighter and louder than the first, you can mimic the change of tone that occurs on stronger notes.

This method is very efficient and relatively simple to execute. Note that, even though there are two samples assigned to each key, only one sample is output per note played. This could be helpful when there are a limited number of voices of polyphony are available.

Multi-Layer Switching

Additional expressiveness can be gained with three or more velocity zones per key. For example, a violin could be sampled at dynamic levels of p, mp mf and f. The piano (p) dynamic could be assigned to velocities from 1 to 40; mezzopiano (mp), from 41 to 63; mezzoforte (mf), from 64 to 95; and forte (f) from 96 to 127.

Of course, you are not limited to dynamic changes with velocity switching. You could create an Acoustic Guitar patch in which velocities over 120 trigger a guitar harmonic instead of a plucked note. A Vocal Pad patch might switch from "Ooh" to "Aah" above some velocity level.

Velocity layering

In Velocity Layering, one sample is always heard and the second sample is only added when the velocity is above some specified level. The fullness of the first sample will still be there, but bite will be added to the notes with higher velocities.

For example, to create an effective electric piano patch, find two samples: one should be a general-purpose sound that's a little bit on the mellow side; the second sample should be brighter (you may want to use a high-pass filter on it to thin out the bottom end.) Set the mellower sample to play at all velocity levels; Set the brighter, thinner one to play only when the velocity goes above the split point (e.g. v=100).

Note that this configuration requires only one voice when velocity is low, but uses two voices at high velocities.

Velocity Fade

Samplers that have a Velocity Sensitivity setting for each sample can support Velocity Fading. This is similar to Velocity Layering in its basic concept, but accomplishes the trick with greater finesse. The second sample is set to have a higher sensitivity to the incoming velocity such that the level of sample two is proportional to the velocity level. When a low velocity is played very little of the second sample is heard; when a higher velocity is played, it gets louder. Instead of two possibilities (the mellow and the bright samples vs. the mellow sample alone), you have 127 possible mix combinations of the two samples.

Note that this configuration requires two voices at all velocities.

Velocity Crossfade

It would also be possible on these samplers to set up a Velocity Crossfade. In this configuration, as velocity increases, sample one gets softer and sample two gets louder. This could be used to gradually morph one sound into another similar (or completely different) timbre. To create this configuration, simply set sample one to have a negative velocity sensitivity value and sample two to have a positive sensitivity value.

Note that this configuration also requires two voices at all velocities.