getting started with alv

alv (“alive”) is a language for creating and changing realtime programs while they are running continuously. It can be used to create music, visuals or installations, but by itself creates neither sound nor video. Rather, alv is used together with other tools and synthesizers (for example SuperCollider or Pilot). In such an ensemble of tools, alive takes the role of a ‘conductor’, telling the other tools what to play when by sending commands to them using a variety of protocols, such as OSC and MIDI.

Before we get to making sound though, we should learn a bit about the alv programming language, and how to install and use it.


alv is written in the Lua programming language, and is compatible with both Lua 5.3 and luajit.

unix/linux and mac os

Your distribution should provide you with packages for Lua and Luarocks. On Mac OS X, both are provided through homebrew. After installing both of these, you should be able to start the Lua interpreter from the shell:

$ lua
Lua 5.3.5  Copyright (C) 1994-2018, PUC-Rio

You can exit using CTRL+C. If the version you see is not 5.3, double check your distribution packages or see if it was installed as lua5.3 or lua53. Similarily, you should be able to run luarocks, luarocks53 or luarocks5.3:

$ luarocks list

Rocks installed for Lua 5.3

Again, double check your installation or try adding --lua-version 5.3 if the displayed version is not 5.3.

With everything reacdy to go, you can now install the dependencies for alv:

$ luarocks install moonscript
$ luarocks install luasystem
$ luarocks install osc
$ luarocks install luasocket
$ luarocks install

While moonscript and luasystem are required by the core of alv, the other packages (osc, luasocket and lua-rtmidi) are specific to some modules of the alv language, and as long as you don’t need to use these modules their installation is optional.

In a later part of this guide, we will be using modules that require osc and luasocket, so it is recommended to install at least these two. However it is possible to follow a large portion of the guide without any of them. There will be a note marking the parts of the guide where specific dependencies are required.

After installing the dependencies, you can download the alv source code from the releases page, or clone the git repository:

$ git clone

To run the copilot, open a shell and navigate into the repository. You can now run the hello.alv example script using the following command:

$ moon init.moon hello.alv
hello.alv changed at 1585138092

You can stop it by pressing ^C (control-C).


For Windows, a binary package is available from the latest github release. It includes not only the alv source code, but also a compiled version of Lua 5.3 as well as Luarocks and all of alv’s dependencies.

To use the binary package, simply extract the archive and move the folder wherever you want. You can now start the hello.alv example script by dragging it onto the copilot.bat file in the folder, or by running the following command from the main directory in cmd.exe:

C:\…\alive>copilot.bat hello.alv
hello.alv changed at 1585138092

You can stop it by pressing ^C (control-C).

evaluating code

To get started writing your own code, create an empty file in the text editor you want to use, and save it as test.alv in the same folder as hello.alv. Now restart the copilot as described above, but substituting the new file.

You should see a note indicating that alv processed the file. From now on, whenever you change test.alv and save the file, alv will reload it and execute your new code. When you are done, you can stop the copilot at any time using ^C (control-C).

alv’s syntax is very similar to Lisp. Expressions take the form of parenthesized lists like (head a b c...), where the first element of the list (head) is the name of an operator or function, which defines what the expression as a whole will do, while the other elements are parameters whose meaning depends on the head. Let’s start with a simple operator, print: print is used simply to print messages to the copilot console. Enter the following in your file and save it:

(print "hello world!")

As soon as you save the file, you should notice two things happening:

  1. The copilot should print two new lines to the terminal:

    hello.alv changed at 1583424169
    hello world!

    In the first line, it notifies us that the file has changed. In the second line, you can see the output from print.

  2. The copilot will make a small modification to your file. Depending on the editor you are using, this may either result in you seeing the modification immediately, or a notice appearing that offers the option to reload the file. If it is the latter, confirm the notification to accept the changes. If there is an option to do so, you may want to configure your editor to always reload the file automatically.

The code should now look like this:

([1]print "hello world")

The [1] that the copilot added to our expression is that expression’s tag. In alv, every expression has a tag that helps the copilot to identify the individual expressions as you make changes to your code. The copilot will make sure that all expressions are tagged by adding missing tags when you save the file, but you have to watch out not to duplicate a tag when copying and pasting code. When you duplicate code that already has tags, you can either manually change the tags to be unique, or simply delete the whole tag (including the square brackets) and let the copilot generate a new one for you the next time you save the file.


As we just saw, expressions in alv take the form of parenthesized lists. Elements of an expression have to be separated by whitespace, but any type of and amount of whitespace is valid: feel free to use spaces, tabs, and newlines to format code to your liking. The following are all equal and valid examples:

(print "hello world")

(+ 1

    "hello world")

  ( print "hello world" )

It is however recommended to follow the clojure style guide as much as it does apply to alv. All further examples in this guide will respect this guideline.

basic types

Strings can be written in two ways: using double quotes ("), as we did above, or using single quotes ('). In both types of strings, you can escape a quote that otherwise would signify the end of the string by adding a single backslash before it. Consequently, backslashes also have to be escaped in the same way. The following are all valid strings:

"hello world"
'hello world'
"it's a beautiful day"
'it\'s a beautiful day'
"this is a backslash: \\"
"this is a double quote: \""

Aside from strings, there are two more types of values that you can use when writing alv programs: numbers and booleans. Numbers use the digits 0-9 and can be integers, contain a decimal point, or start or end with a decimal point. Numbers can start with a negetive sign. The following are all valid numbers:


There are only two boolean values, true and false:


The operator print, that we have been using above, only works on strings, but there is a similar operator called trace that can be used to inspect any kind of value. It prints the value itself alongside more information, such as the values type. Give it a try:

(trace "hello")
(trace 2)
(trace true)

This will print the following:

toast.alv changed at 1585138575
trace <ValueStream str: hello>: <ValueStream str: hello>
trace <ValueStream num: 2>: <ValueStream num: 2>
trace <ValueStream sym: true>: <ValueStream bool: true>


To annotate your code, you can use comments. In alv, comments begin with #( and end on a matching ). This way you can comment out a complete expression simply by adding a # character in front.

#(this is a comment)

#(this is a long,
  multi-line comment,
  (and it also has nested parentheses).
  It ends after this sentence.)

You can put comments anywhere in your program where whitespace is allowed and it will simply be ignored by alv.

importing modules

Apart from trace, there are only very little builtin operators in alv - you can see all of them in the builtins section of the reference. All of the ‘real’ functionality of alv is grouped into modules, that have to be loaded individually. Modules help organize all of the operators so that it is less overwhelming to look for a concrete feature. It is also possible to create your own plugins as new modules, which will be covered in another guide soon.

Let’s try using the + operator from the math module. To use operators from a module, we need to tell alv to load it first: We can load all the operators from the math module into the current scope using the import* builtin:

(import* math)
(trace (+ 1 2))


trace (+ 1 2): <Value num: 3>

Because it can get a bit confusing when all imported operators are mixed in the global scope, it is also possible to load the module into its own scope and use it with a prefix. This is what the import builtin is for:

(import math)
(trace (math/+ 1 2))

defining symbols

Another element of code in alv that we haven’t discussed in detail yet are symbols. Symbols (like trace, import* or math/+) are names that serve as placeholders for previously defined values. When code is evaluated, symbols are looked up in the current scope and replaced with the corresponding value found there.

When an alv file starts running, a number of symbols are defined in the default scope: These are the builtins mentioned above, and of which we have already been using trace, import, and import*.

To define a symbol yourself, the def builtin is used. It takes the symbol as its first, and the value to associate as its second parameter. After a symbol is defined, the name becomes an alias that behaves like the value itself. For example, we can use def to associate the result of our calculation with the symbol result, and then refer to it by that symbol in the trace operator:

(import* math)

(def result (+ 1 2))
(trace result)

Symbols need to start with a letter or one of the characters -_+*/.!?=%. After the first character, numbers are also allowed. There are two types of symbols that are treated specially: symbols containing a slash (math/+), and symbols starting and ending with asterisks (*clock*):

Both import and import* are actually shorthands and what they accomplish can be done using the lower-level builtins def, use and require. Here is how you could replace import:

#(with import:)
(import math)
(trace (math/+ 1 2))

#(with def and require:)
(def math (require "math"))
(trace (math/+ 1 2))

require returns a scope, which is defined as the symbol math. Then math/+ is resolved by looking for + in this nested scope. Note that the symbol that the scope is defined as and the name of the module that is loaded do not have to be the same, you could call the alias whatever you want:

#(this not possible with import!)
(def fancy-math (require "math"))
(trace (fancy-math/+ 1 2))

Most of the time the name of the module makes a handy prefix already, so import can be used to save a bit of typing and make the code look a bit cleaner. import*, on the other hand, defines every symbol from the imported module individually. It could be implemented with use like this:

(use (require "math"))
(trace (+ 1 2))

use copies all symbol definitions from the scope it is passed to the current scope.

Note that import, import*, def, and use all can take multiple arguments:

#(using the shorthands:)
(import* math logic)
(import midi osc)

#(using require, use and def:)
(use (require "math") (require "logic"))
(def midi (require "midi")
     osc  (require "osc"))

It is common to have an import and import* expression at the top of an alv program to load all of the modules that will be used later, but the modules don’t necessarily have to be loaded at the very beginning, as long as all symbols are defined before they are being used.

nested scopes

Once a symbol is defined, it cannot be changed or removed:

(def a 3)
(def a 4) #(error!)

It is, however, possible to ‘shadow’ a symbol by re-defining it in a nested scope: So far, all symbols we have defined - using def, import and import* - have been defined in the global scope, the scope that is active in the whole alv program. The do builtin can be used to create a new scope and evaluate some expressions in it:

(import string)

(def a 1
     b 2)

(trace (.. "first: " a " " b))
  (def a 3)
  (trace (.. "second: " a " " b))
(trace (.. "third: " a " " b))

This example prints the following:

trace (.. "first: " a " " b): <Value str: first: 1 2>
trace (.. "second: " a " " b): <Value str: second: 3 2>
trace (.. "third: " a " " b): <Value str: third: 1 2>

As you can see, within a nested scope it is possible to overwrite a definition from the parent scope. Symbols that are not explicitly redefined in a nested scope keep their values, and changes in the nested scope do not impact the parent scope.

defining functions

Another builtin that creates a nested scope is fn, which is used to create a user-defined function, which can be used to simplify repetitive code, amongst other things:

(import* math)

(def add-and-trace
    (a b)
    (trace (+ a b))))

(add-and-trace 1 2)
(add-and-trace 3 4)

Here a function add-and-trace is defined. When defining a function, first the names of the parameters have to be given. The function defined here takes two parameters, a and b. The last part of the function definition is called the function body.

A function created using fn can be called just like an operator. When a function is called, the parameters to the function are defined with the names given in the definition, and then the function body is executed. The previous example is equivalent to the following:

(import* math)

(def add-and-trace
    (a b)
    (trace (+ a b)))

  (let a 1
       b 2)
  (trace (+ a b)))

  (let a 3
       b 4)
  (trace (+ a b)))

and the output of both is:

trace (+ a b): <Value num: 3>
trace (+ a b): <Value num: 7>

In alv, functions are first-class values and can be passed around just like numbers, strings, etc. However it is very common to define a function with a name, so there is the defn shorthand, which combines the def and fn builtins into a single expression. Compare this equivalent definition of the add-and-trace function:

(defn add-and-trace (a b)
  (trace (+ a b)))

evaltime and runtime

So far, alv may seem a lot like any other programming language - you write some code, save the file, and it runs, printing some output. “What about the ‘continuously running’ aspect from the introduction?”, you may ask yourself.

So far, we have only seen evaltime execution in alv - but there is also runtime behavior. At evaltime, that is whenever there is change to the source code, alv behaves similar to a Lisp. This is the part we have seen so far. But once one such eval cycle has executed, runtime starts, and alv behaves like a dataflow system like PureData, Max/MSP or vvvv.

What looked so far like static constants are actually streams of values. Whenever an input to an operator changes, the operator (may) update and respond with a change to its output as well. To see this in action, we need to start with a changing value. Number literals like 1 and 2, which we used so far, are evaltime constant, which means simply that they will never update. Since all inputs to our math/+ operator are evaltime constant, the result is constant as well. To get some runtime activity, we have to introduce a side-effect input from somewhere outside the system.

The time module contains a number of operators whose outputs update over time. Lets take a look at time/tick:

(import* time)
(trace (tick 1))

This will print a series of numbers, incrementing by 1 every second. The parameter to time/tick controls how quickly it counts - try changing it to 0.5 or 2. As you can see, we can change time/tick while it is running, but it doesn’t lose track of where it was!

All of the other things we learned above apply to streams of values as well - we can use def to store them in the scope, transform them using the ops from the math module and so on:

(import* time math)
(def tik (tick 0.25))
(trace (/ tik 4))

Note that if you leave the time/tick’s tag in place when you move it into the def expression, it will keep on running steadily even then.

making sound

As mentioned earlier, alv doesn’t produce sound by itself. Instead, it is paired with other tools, and takes the role of a ‘Conductor’, sending commands and sequencing other tools.

For the sake of this guide, we will be controlling Pilot, a simple UDP-controlled synthesizer. You can go ahead and download and open it now. You should see a small window with a bunch of cryptic symbols and a little command line at the bottom. To verify that everything is working so far, try typing in 84c and hitting enter. This should play a short sound (the note 4C, played by the 8th default synthesizer voice in Pilot).

To talk to Pilot from alv, we will use the pilot module. Note that for this module to work, you have to have the osc and luasocket dependencies installed. To play the same sound we played by entering 84c above every 0.5 seconds, we can use time/every to send a bang to pilot/play:

(import* time)
(import pilot)
(pilot/play (every 0.5) 8 4 'c')

You can play with the voice, octave and note values a bit. To add a simple melody, we can use util/switch, which will cycle through a list of parameters when used together with time/tick:

(import* time util)
(import pilot)
(pilot/play (every 0.5) 8 4
  (switch (tick 0.5) 'c' 'd' 'a' 'f'))

Now we can have the voice change every other loop as well:

(import* time util)
(import pilot)
(pilot/play (every 0.5)
  (switch (tick 4) 8 9)
  4 (switch (tick 0.5) 'c' 'd' 'a' 'f'))

To round off the sound a bit, we can turn on Pilot’s reverb using pilot/effect. Add the following somewhere in your file:

(pilot/effect "REV" 2 8)

Now it’s time to add some rhythm. The kick drum is voice 12 by default, and we can also add something like a snare on channel 3:

(pilot/play (every 0.75)
  12 2 'd' 3)
(pilot/play (every 2)
  13 4 'a' 4)

Note that since we are using multiple individual time/every instances, the timing of our voices relative to each other is not aligned - each voice started playing when the file was first saved with it added, and kept the rhythmn since. By deleting all their tags and re-saving the file, we can force alv to re-instantiate them all at the same time, thereby synchronising them.