Page 4 is used to create voices through additive synthesis. The voices tend to have an "electronic" sound to them.

The preceding image shows a Mode 1 voice with several harmonic profiles. In effect, this is a "side view" which shows the harmonic amplitudes varying over time. Up to 32 harmonics can be controlled.

A Mode 1 voice consists of 32 waveform segments, represented by the vertical lines in the graph area. That gives 32 points per harmonic profile. The length of time that each point sounds for is specified by a duration profile (DUR), while the overall amplitude envelope of the sound is indicated by the energy profile (ENG).

The numbers across the bottom of the image show the status of the 32 harmonics. A cross (+) below the number indicates that there is data for that harmonic (the profile is non-sero). A triangle indicates that the harmonic is currently displayed on the graph. So the image shows that harmonics 1, 2, and 6 are displayed along with the energy and duration profiles, while harmonics 3 and 12 have data but are not visible.

The JOIN/PLOT selector controls how the profiles are drawn with the lightpen. JOIN connects each point to the previously drawn one, creating a smooth curve. PLOT allows individual points to be specified without affecting the surrounding ones. The CLEAR command hides all profiles from the display, while the DELETE command hides just one. The RESET command actually sets all harmonic values to zero, while the ZERO command does that for just one profile.

Once harmonic amplitude profiles are drawn, the COMPUTE command mathematically processes them to create the waveforms for each of the 32 segments. Also, the energy profile is automatically calculated from the combined amplitudes of the harmonic profiles. If the energy profile is modified and the SCALE command is used, the harmonic profiles will all be modified accordingly.

The INTERP option controls the COMPUTE process. If INTERP is ON, each waveform segment is calculated from a mix of the harmonic profile values of the current segment and the next one. This provides a smoother sound.

The LOOP control is a very handy visual indicator of which segments are repeated when the voice is played. The start and end points can be set with the lightpen, as well as mapped to controls (on Page 7) and dynamically adjusted.

Mode 1 sounds are unique on the Fairlight CMI because they play for a fairly similar length of time, no matter the pitch. When a Mode 1 voice is played each segment is repeated a number of times, which is determined by a combination of the duration profile and the pitch. This is in contrast to sampled sounds and Mode 4 sounds, which become shorter as the pitch increases and the playback rate increases.

The following image shows a Mode 4 voice. In fact, it is the previous Mode 1 voice converted to Mode 4, and then edited slightly.

The harmonic profiles from the Mode 1 voice are compressed to the left-hand side in this image because a Mode 4 voice consists of 128 segments, in contrast to the 32 segments of a Mode 1 voice. Notice that even the loop has been visually compressed, but it will need to be re-established anyway as editing progresses. In a Mode 4 voice, the vertical divisions in the graph area represent every fourth segment.

The 1st harmonic profile has been selected for editing. This is indicated with a double line on the graph and a box around the profile number.

The ENG and DUR controls aren't available to a Mode 4 voice because the overall amplitude characteristics of the voice are controlled on Page 7 and each segment is sounded only once. Similarly, the SCALE command is not available.

Mode 4 voices use the entire 16,384 bytes of waveform memory, which is divided into 128 segments of 128 sample points each. Each segment contains one cycle (or period) of the waveform.

Mode 1 voices only use the first 32 segments, or 4096 sample points total. If a Mode 1 voice were played at CD-quality speeds it would last less than 1/20th of a second!

Although Page 4 is primarily used to create sounds with additive synthesis, it can also be used to alter sampled sounds. Sampling is usually performed into a Mode 4 voice, but if the voice is then converted into Mode 1, interesting results can be obtained.

The energy and duration profiles have an effect on a converted sample, but the individual harmonic profiles should be left empty during this process. The energy and duration profiles take effect immediately, without the need for executing the COMPUTE command. (In fact, because all of the harmonics are empty, COMPUTE would blank out the waveform!)

Some sampled voices sound rather robotic when converted into Mode 1. They have a kind of shimmering, quantized effect that is the result of stepping through the segments at a rate (determined by the duration profile) that is slower than normal.

Sounds which are slower moving and less varying (such as choirs or string pads) sometimes respond well in Mode 1, because it extends the sample and makes the "throbbing" of the loop less prominent (or at least the same duration at every pitch).

The following table summarizes the features of Mode 1 and Mode 4.

				Mode 1				Mode 4
Segments used				1 to 32				1 to 128 (all)
Each segment is played				depends on DUR profile				once
Amplitude				depends on ENG profile				depends on COMPUTE results
Overall duration				fairly constant				inversely related to pitch
Primary purpose				harmonic synthesis				drawing and sampling

Previous page - Page 3: Keyboard Control