Exercise: Music Sequencer

Objective

Generate tone patterns over time from a string representation of a melody.

Many physical systems require generating patterns of control outputs over time. For example, a dishwasher or washing machine controller is primarily a clock which times the phases of spraying or agitation. There is some sensor feedback for process transitions based on physical state (e.g. the water level reaching a threshold), but much of the process is simply timed.

Performance machines, especially musical ones, also frequently utilize timed sequences. It is also convenient for these machines to have a representation of the pattern which can easily be used for authoring and editing the performances.

This exercise introduces a very simple pattern language which represents a rhythmic tone pattern as a fixed string of text. The pattern string is interpreted by the program. This data-driven approach allows more compact representation than the equivalent code sequence. It also integrates much more easily alongside simultaneous processes such as sensor data acquisition.

Steps and observations

1. Load and run the MusicSequenceDemo sketch. The example program generates audio-frequency square wave tones with different rhythms and pitches on pins 4 and 5. This demonstrates an event-loop and rhythm pattern control structure allowing parallel execution of multiple timed tasks.
2. Observe the initial outputs on an oscilloscope.
3. Design and construct an Arduino circuit incorporating two audio speakers for digital output. The speakers can be driven by digital outputs using a ULN2803 attached to the output pins defined in the sample sketch. The speaker drive circuit can use the same structure as the previous ULN2803 exercise, using the speakers on the output side. A small speaker is typically about 8 ohms and can handle a fraction of a Watt; assuming 1/8 Watt, the power supply voltage would be typically be about a volt above the ULN2803 forward drop, so start with a power supply voltage of 1.6V and experiment.
4. Modify the rhythm pattern string to generate a different beat pattern or melody.

Comments

This exercise uses a C++ class named MelodyGenerator to implement the tone and rhythm pattern generator. This was written as one class for simplicity, but logically might be broken out into separate ToneGenerator and PatternGenerator classes. This class was also incorporated directly into the sketch but would more conventionally be provided in separate files as a generic re-usable component.

Challenges

1. Add a sensor acquisition process to read an external input and modulate the performance.
2. Extend the pattern language to include more tones.
3. Extend the timing control to include articulation (short silences) to define boundaries of repeated notes.
4. Extend the timing control to vary the pulse width to control tone volume.
5. Rewrite the MelodyGenerator class into generic ToneGenerator and PatternGenerator classes.
6. Add a phase-lock algorithm to beat-match to an external sensor input.

Arduino Code

1. Documentation: MusicSequenceDemo Arduino Sketch
2. Sketch Folder: MusicSequenceDemo