A newer version of Max is available. Click here to access the latest version of this document.

MIDI Tutorial 2: MIDI Synthesizer

In this tutorial we show how to create and work with a simple 4-voice MIDI-controllable synthesizer. Along the way, we discuss simple methods of polyphonic routing as well as using a variety of common MIDI messages to control synthesis parameters.

• To use the tutorial patchers in this section of the tutorial, make sure you have a correctly configured MIDI controller connected to your computer. The tutorials in this section use a variety of MIDI messages as example input; if your controller lacks any of these features, you can simulate their input with user interface objects in the tutorial.

Playing our synthesizer

• Take a look at the tutorial patcher. Notice that it contains a fair bit of patcher logic involving four different categories of MIDI input objects sending messages and signals to four copies of an abstraction called synthvoice~. Turn on audio by clicking the ezdac~, turn up the gain~ slider, and play some notes on your MIDI keyboard. The kslider object should animate in response to your actions. While holding a note, adjust the pitch bend wheel on your keyboard. Adjust the modulation wheel or another controller that can send CC#1. Listen to the different changes to the sound.

Our tutorial patcher deals with MIDI through a number of methods. Let's look through these one at a time.

• Play notes on your keyboard one at a time and look at the area of the tutorial patcher labeled 1. Notice what the poly object does to the values. Try playing a note and holding it, then adding another while the first key is still down.

Our tutorial patcher is capable of playing four sounds at the same time, due to there being four different copies of the synthvoice~ abstraction in our patch. In order to take advantage of the polyphony however, we need to figure out how to route our MIDI values to the different voices so that the appropriate copy of the synthvoice~ abstraction receives each message. The poly object takes MIDI pitch and velocity and performs voice assignment on the values based on the arguments to the object. The first note the object receives will be given voice number 1, the second note voice 2, and so on. Once we exceed the polyphony the object is programmed for (in our case, 4 voices), the object will roll around and start again at voice 1. Note-off events will map to the same voice as their corresponding note-on events, guaranteeing that the message to stop a MIDI event goes to the same destination as the one that started it.

• Using your MIDI keyboard, play a four-note chord. Now, without releasing any of the keys, press a fifth note. Notice what happens.

The second argument to poly (1) tells it to steal voices if the polyphony is exceeded. If we attempt to sound more notes than the poly object allows for, the oldest note will be dropped and its voice will be recycled.

The output for our poly object is sent into a pack object and then through a route object to send the remaining lists of pitches and velocities to the appropriate copy of our synthvoice~ abstraction.

• Take a look at tutorial area 2. Move the pitch bend wheel on your keyboard controller. Notice that, unlike most controllers, its resting point is at the middle of the range at 63. See what the patcher logic below does to scale the values between -2. and 2.

To create our pitch bend value, we take the MIDI pitch bend wheel and split its range to two different scale objects. The left-hand scale object takes the lower half of the MIDI range and scales it from -2. to 0.; the right-hand object scales the upper range between 0. and 2. This guarantees that the center value in the range (63) maps to a value of 0. in all cases.

• Take a look at the tutorial area labeled 3. Move the continuous controller and look how it's scaled. Look at the tutorial area labeled 4. If you have a MIDI controller with real-time transport capabilities (i.e. it can send MIDI beat clock), set it up to transmit at any tempo you like and start the transport on the controller. If you like, you could also use an inter-application MIDI routing utility and a software sequencer on your computer to do this. If not, click in the number box labeled tick speed and enter the value 20. Change the continuous controller and play some notes. Listen to the result. Change the tick speed to something faster (like 10.). Notice what happens.

The MIDI CC# and the real-time messages are sent in by the ctlin and rtin objects, respectively, to work together controlling a low-frequency oscillator (LFO). The real-time messages that define the beat clock control the rate of the LFO; the controller messages change the depth. MIDI beat clock typically runs at 96 PPQ; the timer object measures the intervals between ticks, which are then scaled to derive the rate of the LFO so that it lasts one measure. MIDI real-time message 250 (the 'start' message on a sequencer) resets the phase of the cycle~ object so that the LFO will re-synch with a sequence if it starts on a barline.

Note how the output of the LFO is scaled so that even with the depth sending a signal of 0., the signal sent into the synthvoice~ abstraction is guaranteed to be centered around 1. Let's take a look at what's in that patch.

The synthvoice~ abstraction

• Double-click any of the abstractions named synthvoice~ and look at the patcher logic inside.

Our synthvoice~ abstraction has three inLets. The first inlet takes lists of pitch and velocity values from our MIDI keyboard input. The pitch value is sent to drive a constant signal (sig~) which has the signal from inlet #2 added to it (+~). This value is then interpreted as a MIDI number and converted to a frequency in the signal domain by an object called mtof~, which behaves just like the mtof object but operates on continuous MSP signals instead of Max numbers. This frequency value then feeds two band-limited oscillators: a square wave (rect~) and a sawtooth wave (saw) which are mixed together. The frequency of the rect~ object is multiplied by the LFO signal coming in inlet #3 so that a rich, chorused tone can be acheived when the depth of the LFO is increased.

Standard synthesizer envelopes: adsr~

Meanwhile, the velocity output of the MIDI notes is scaled between 0. and 1. and sent to an object called adsr~. The object stands for Attack, Decay, Sustain, Release, and generates signal ramps in a standard configuration borrowed from analog synthesizer design:

A standard ADSR curve.

The arguments to adsr~ are interpreted as an attack time, a decay time, a sustain level, and a release time. The sustain level is a multiplier of the overall amplitude which the object outputs during a sustaining note.

The adsr~ object takes a value and interprets it as an envelope trigger of a certain amplitude. Any number higher than 0. triggers the attack, decay, and sustain phases, scaled to match the amplitude of the trigger (e.g. an input value of 0.5 will trigger a softer envelope than a trigger of 0.8). The object then stays at the sustain phase, putting out a constant value until it receives a 0. It then continues to the release phase of the envelope and ramps to 0.

• Now that we understand how the synthesizer elements work, return to the main patcher and see if the controllable parameters make sense. Try exploring the full range of the MIDI control to see how expressive it is.


Within a patcher, MIDI note events can be routed polyphonically to different copies of the same abstraction using a poly object. Objects such as bendin and ctlin can be scaled to match different synthesizer parameter ranges, and real-time MIDI commands can be used to derive tempo data for LFOs and sequencers within Max. The adsr~ object generates envelope ramps based on triggers for a note-on and a note-off, making it ideal for use with MIDI note-based systems.

See Also

Name Description
poly Allocate notes to different voices
mtof~ Convert a MIDI note number to frequency at signal rate
adsr~ ADSR envelope generator