Microcsound is a command-line tool that serves as a general purpose Csound score-generation front-end, as well as a real-time interactive performance shell. It started life as an early ASCII-notation to MIDI parser/compiler called et_compose, which later became micro_composer. I finally decided to take MIDI out of the picture and focus on Csound-centric functionality; hence the latest evolutionary branch being called Microcsound.

Microcsound was developed to ease the pain of writing contrapuntal and harmonic music by hand in the standard Csound score format, where each event follows the next in a non-musician friendly vertical page fashion. The standard Csound score format makes the confluence of different musical layers difficult to parse, since most most musicians read layers of music within a score with time laid out horizontally across the page rather than vertically. The solution is a front-end which would translate these more intuitively seen, horizontal text layers, called a Microcsound score, into the SCO format that Csound works with, bridging a communication gap between human composer and computer friendly formats.

Microcsound is used in four modes from the command line:

A special feature of Microcsound that is particularly useful for microtonalists (which explains the name of the script) is that one can represent just intonation in ratio notation, as well as having two modes of symbolic notation to express any arbitrary equal division of the octave (or even non-octave). Possible uses of this feature include generating fully realized microtonal or justly-intoned musical works, creating test tones for guide tracks of acoustic instruments, demonstrating principles of musical tuning in a straightforward way using a friendly syntax, experimenting with flexibly tuned sonorities and auditioning microtonal intervals and chords, and tuning a real-world acoustic instrument to an alternate tuning. (The author has done precisely this with his kalimbas).

Since Microcsound is simply a Python script designed to function in a command-line environment and avoids a GUI, it might appeal to those who like lean shell environments to work from without the overhead of a GUI, or even to those who would just want a clean and simple way to do their Csound work. Also, using Microcsound's interactive shell mode is a quick, convenient and easy to understand a way of auditioning any arbitrary microtonal sonority, one which transcends the limitations of more traditional ways such as setting up and retuning standard MIDI software synthesizers. This is true for those who do not have or are not interested in generalized keyboard setups for microtonal experiments and composition, and for whom dragging and dropping operations slow down their flow. Nevertheless, for those who want the features of a full GUI front-end environment, Microcsound can be adapted (as hinted to me by Csound guru Steven Yi) in the future to integrate with Yi's blue composition environment, giving users both GUI and quick non-GUI advantages simultaneously.

For composers who want to script algorithmic compositions, Microcsound's syntax should make it easy to do so. With modest programming experience, it is fairly trivial to implement any number of designs that output valid Microcsound score. The author has experimented along such lines many times using the Python language.


I. Basics of use

Downloading and Installing

Microcsound is available at There is a choice between a gzipped-tarball (.tgz) for use with Linux or OSX, and a ZIP (.zip) file for use with Windows. The Linux version is the primary version developed by the author; the ZIP version has small changes to the source code to conform to the Windows system. Mac users should use the Linux version, since they will share a very similar Unix-type base code. Anyone with questions, suggestions, comments, or needs help with installation can contact the author for further assistance.

After uncompressing the chosen archive file, Linux and OSX users must edit the script to set some system variables. You will need to edit the values of orc_dir and possibly some other variables towards the top of the script between where it says "EDIT DEFAULTS HERE" and "END USER EDITING". Microcsound, among other things, needs to know where to find Csound orchestras and whether or not to use the supplied microcsound.orc. In addition, you may set command-line settings when calling Csound from Microcsound (i.e what buffer settings you would typically use, etc.).

On the Linux version, after editing the values in the Microcsound script itself, you can try the script by typing the following at the command-line prompt:


This script will ask some questions interactively. (You may find that the script's defaults do not work well for your setup and that it needs editing to reflect variables and assumptions true for your operating system and file-system.) Once answered, these questions serve to fill in certain install variables. The script will then install Microcsound and some optional files based on the data provided.

Windows users should use the script. Windows version users will still have to edit the defaults at the top of the file in the same fashion as mentioned for the Linux version above, but there are slight differences that make it easier for you: there are no references to the readline library, which does not exist in Windows, and the temporary directory is named "/temp" instead of the Linux version's "/tmp".

Supplied with the program is a starter Csound orchestra microcsound.orc that allows you to get up and running with some decent-sounding instruments right away. The script also serves to show some stepping off points for exploring your own instrument designs. It is beyond the scope of this article to fully explain each instrument in the supplied orchestra. The orchestra code is reasonably well commented, and anyone familiar with the basics of Csound orchestra files and instrument design should be ready to grasp what is immediately available for use with Microcsound.

Invoking the program

Using the program involves invoking the Microcsound script from a command line shell—typically, a BASH terminal in Linux or Mac, or a CMD shell in Windows. As mentioned there are four modes of operation: score-only render, WAV render, real-time render, and interactive. You can get some helpful usage information by typing microcsound -h at the prompt:

akj@myhost:~$ microcsound -hUsage: microcsound [options] []This is microcsound version 20110407Options:-h, --help            show this help message and exit-o OUTPUT_WAVE_FILE, --output=OUTPUT_WAVE_FILEoptional wave file output name-s, --score-only      only generate a score to stdout, do not post-processit with csound-r, --real-time        render audio in real-time-i, --interactive     use an interactive prompt, render audio in real-time aswell, does not work when -s is also specified-t, --stdin           read text from stdin-v, --debug           turn on debug mode--orc=ORC_FILE        specify an orchestra file for csound to use, which is		                      not the default (microcsound.orc)

The options are somewhat self-explanatory, but I will clarify: the normal mode of operation is to invoke Microcsound followed by the name of a given .mc file, a text file that contains Microcsound score. Microcsound will take the .mc file, generates a temporary Csound .SCO file, and then run Csound with the .SCO file with the default or designated Csound .ORC file. The output will be saved as a WAV file in the the current directory, using the same pre-extension basename of the .mc file, but replacing .mc with .wav. For example, running "microcsound" will produce a WAV file called mypiece.wav in the same directory as where the Microcsound script is called. To do the same thing, but hearing the audio from your soundcard rather than saved to a wave file, use the -r switch (real-time), e.g. "microcsound -r". If the variables at the top of the Microcsound script are correctly configured for the setup of your system, the result will be a real-time performance of your .mc file.

If you would rather see the Csound score yourself and not have Microcsound do the final compilation to audio, use the -s switch. This will print the generated Csound score to the screen. On Linux/Unix, you can then capture this output to a file using a pipe redirect, e.g. "microcsound -s > mypiece.sco". You can then compile the captured .SCO file yourself with Csound, using an orchestra of your choice, e.g. "csound myorc.orc mypiece.sco" along with whatever Csound options you want. This is useful if you need to further process a score in ways other than Microcsound provides, or to debug the output of Microcsound itself if you are running into difficulties.

The use of the interactive shell by invoking "microcsound -i" deserves special mention. The only difference between this and the standard syntax for Microcsound scores is that the prompt needs to know when to stop expecting input and to actually process your commands into real-time audio. This is done by typing "done" and hitting ENTER or RETURN at the prompt as soon as all the musical information has already been entered. You will see then the Csound command being called from within the shell, and start realizing the performance. Some examples that follow the explanation of Microcsound syntax will suffice to get you up to speed in understanding how this works.

Understanding the syntax of Microcsound

Microcsound scores are comprised of voices. These are best thought of as layers of a texture, similar to the idea of a channel in one of the 16 channels of MIDI. Everything that happens in Microcsound happens in the context of a voice, including setting up variables (such as tempo) and mixer values. (Note: All variables in Microcsound are somewhat global in their scope. If a voice does not explicitly set a variable, it will use the value that was previously set the previous voice, i.e. voice 2 will use the values of voice 1 unless set explicitly.) A good practice is to use voice 1 to set up variables that all other voices will share. One exception, however, is the case of setting tempo. If you need to implement a gradual tempo change between note events, it might be best to use whatever voice of the texture that has the fastest-moving rhythmic values to put your t= tempo settings.

The beginning of every line in a Microcsound score has a voice indicator. This is simply the integer number of the voice (1 or greater) followed by a colon then a space, for example:

1: div=31 i=1 c d e f g

This line begins with a voice indicator for voice 1 (indicated by "1: "), then indicates to use a 31-edo (equal division of the octave) scale and perform with Csound instrument 1 a single 5-note scale.

Multiple simultaneous voices are indicated by using a new higher number. The Microcsound rendering engine scans the file, looking for the first "1: ", merges all events that fall under this voice, interprets them and generates the Csound score. After reaching the end of the last event in that voice, the engine starts over with the next highest number, "2: ". You can skip voice numbers and for example next use "3: ", as Microcsound will always count in order and will simply skip voice numbers it does not find. Once it finds the next voice, the clock is reset to time "0", and it will interpret all of the events for the next voice and generate the Csound score. After the final voice is interpreted, all the voices are sorted and written to a temporary Csound score, which is then compiled with the appropriate orchestra of instruments to either produce a WAV file output, real-time audio output, or a Csound score for further processing elsewhere.

There are 4 types of commands or syntactical elements in the Microcsound language that can happen in any voice: comments, variable setting commands, events, and time-stamp manipulations.

1. Comments are simply lines which begin with '#'. Comments are ignored by Microcsound and can contain any information that helps the composer. Comments end when a carriage return or newline of text is found. For example:

#This is an EXAMPLE of a Microcsound comment

In addition to comments, a bar line (indicated by "|", the pipe character) is also ignored:

1: div=31 i=6 "0.02%0.02%4%0.5" 1/8 c2 de f2 1/12 gab | 1/8 c'2 g2 e2 g2 | c4 z4 ||

Composers may find it useful to use barlines to visually group parts of their music into measures for ease of human-readability.

2. Variables are important in setting up parameters for your performance. There are two types of variables: set variables that are defined with x=y syntax and those that use a special syntax. Variables that use the x=y syntax include include:

Variable Description
divdivision of the octave to use for microtones
iCsound instrument number
panstereo panning
mixinstrument volume in the mix
gvGaussian (random distribution) variance of volume
grGaussian (random distribution) variance of rhythm
gsstoccato articulation

Setting these variables is simple: in the Microcsound score, surround the left and right of the statement with white space, and write variable=value, for instance:

1: i=1 t=110 div=31 pan=0.4 gv=0.01

declares that voice 1 will use instrument 1, set the tempo (globally for all voices, since that is the nature of tempo) to 110 BPM, set the octave to be divided into 31 equal pitches, sets the instrument just left of center (pan is 0-1, hard left to hard right), and declares that we want a little randomness in the attack volumes, the effect of which varies according to the design of the instrument.

Variables that use special syntax are: default running rhythmic values (8th notes, quarter notes, etc.), attack volume defaults, and extra p-field variables.

Default rhythmic values are indicated by typing a fraction surrounded by white space. For example, typing 1/8 will indicate that the next events are running 8th notes. Typing any numerator and denominator works, so rather complex rhythmic structures are at your disposal. However, Microcsound will not keep track for you how things line up if you get crazy with this--it is your job to do the math and not get lost! Typing 7/15 is perfectly legal for your Nancarrow-esque experiments in tied groups of 7 15th notes, but take care that you know where you will end up with regard to other voices, if this is a concern.

Attack values are indicated by @ immediately followed by a floating-point number between zero and one, e.g. @0.6 or just @.6. There are ways to crescendo and decrescendo using ramps, but this will be discussed later in this article.

Finally, it should be explained that an instrument's extra p-field variables (p8 and above) can also be accessed from within Microcsound using a special syntax. P-fields 1-7 are reserved by Microcsound for instrument, onset, duration, attack, pitch, pan, and mix values. The rest of an instrument's p-fields, if they exist at all, must be set using a double quote statement with the values separated by a "%" character, e.g. I might have a classic ADSR envelope values being set with pfields 8-11, and would use "0.02%3.7%0.01%0.4". Once an extra p-field variable is set, it remains in use until set by another value or a ramp is found("<"). Currently, all of the extra p-fields must be set at once, and individual extra p-fields may not be accessed individually for manipulation. Fortunately, none of my work has ever needed such a feature, but it may be crucial for others, so it is a planned TODO in a future version.

Expanding on our previous example, we can add some new set variables:

1: i=1 t=110 div=31 pan=0.4 gv=0.01 1/8 @0.67 "0.02%3.6%0%0.3"

The above statement, in addition to what was already explained, sets the default running rhythmic value to an 8th note, the default attack to 0.67, and sets a hypothetical set of p-fields 8-11 (such as values for an ADSR envelope) in instrument 1 (remember, the 'i=1' statement earlier) to 0.02, 3.6, 0 and 0.3.

3. Events are the heart of Microcsound and consist of notes, chords, rests, and triggers. When working with equal divisions of the octave (or non-octave, for that matter), notes can be indicated freely by an syntax, or a standard diatonic-letter syntax closely akin but not freely interchangeable to the ABC standard. These two standards, numeric or diatonic-letter, can be mixed and matched, but it is probably best to stick with one or the other to avoid confusion.

When working with just intonation, one simply uses a N:D notation scheme, where N and D are integers, regardless of the value of the div variable, which is ignored for all N:D events.

Notes are explained below in the subsections according to the 3 styles one may indicate them. After every note event, the clock is advanced by the currently set rhythmic length (indicated by a fraction, 1/8 being the default) multiplied by the optional length or tie indicators attached to any given note event.

Chords are really best understood as a way of stopping the default advancement of the clock by the use of brackets. This allows an accumulation of sonorities that would otherwise follow one another in time to happen simultaneously, which we commonly understand as a chord. A opening bracket, "[" stops the clock, while a closing bracket "]" starts it again. Thus, a C-major chord is indicated: [ceg], or with optional spaces: [c e g]

Rests are simply indicated by r or z followed immediately (without white space, that is) by an optional integer multiplier.

Triggers are special cases where one sets div=0. When triggers are used, Microcsound will generate Csound notes without pitch information, instead passing on the designated values as p5 in the generated notes. This is useful when you want to simply turn on a mixer with a dummy p5 variable (as is the case in the provided microcsound.orc mixer instrument), use a percussion-based instrument as a drum set, trigger a MIDI note number in a special instrument, etc. A typical way to use drum tracks from Microcsound would be to set div=0 and have a helper percussion instrument function as a virtual drum set, where the p5 values are sent to trigger a given percussive instrument. See the code for instrument 120 in the microcsound.orc file for an example of how this is done.

4. Time-stamp manipulations involve being able to set, or reset, the beat number that Microcsound is currently on. More is explained on this in the subsection below "Using the time pointer for one-off counterpoint".

Note-event syntax within Microcsound and choosing the right one for the job

Explaining the various note syntaxes is important, as this is the heart of using Microcsound for musical work. Here we will expound the various ways of symbolically representing pitch available to you when using Microcsound.

i. Traditional diatonic letters

Perhaps simplest from the traditional point-of-view is the use of standard diatonic-letter notation. The format for such an event is:


where anything in brackets is optional (only the note base letter, a-g is absolutely required). The optional parts of a note's parameters occur in the above order.

In addition, there are microtonal accidentals that can additionally modify the pitch:

The above modifiers are an alternative way to compose in (quasi-) just intonation besides the use of rational notation, which will be explained later, provided that one uses an appropriate div with high enough frequency resolution, or a medium sized EDO (equal division of the octave) number that well-represents the just intonation ratios one is interested in. For instance, the author has composed little works in virtual 7-limit just intonation using Microcsound by setting div=171, because 171-edo contains excellent approximations to all just intervals with prime factors of 7 and under.

Further examples are given within the next section, with comparison to Numerical syntax; in addition, you can refer to the larger score example at that end of this article.

ii. Numerical syntax

Numerical syntax is simpler from a certain point of view: all pitches are numerals. Middle-C is given the number 0, and the behavior of all other numbers is dependent on how the div variable is set and whether the octave has been set by previous "oct.deg" events. Numbers lower than middle-C are negative, e.g. -7, and optionally, one may use an "[oct.]deg" syntax that is also fairly simple: one indicates a pitch with a number (positive or negative), preceded by an optional octave plus decimal point (middle C is octave number 5).

In more formal terms, this syntax is:

[articulation][octave]deg[optional legato end][ties, using 't', spaces allowed]

Putting it all together with examples, here is a simple C-major scale in different syntax standards (notice the occasional use of staccato and legato articulation:

Standard diatonic-letter syntax:

1: i=1 t=110 div=31 pan=0.4 gv=0.01 1/8 @0.67 "0.02%3.6%0%0.3"1: c d .e .f .g a b (c' d' e' f' g' a' b' c'')t t t

or numerical syntax:

1: i=1 t=110 div=31 pan=0.4 gv=0.01 1/8 @0.67 "0.02%3.6%0%0.3"1: 0 5 .10 .13 .18 23 28 (31 36 41 44 49 54 59 62)t t t

or numerical syntax w/use of oct.deg helpers:

1: i=1 t=110 div=31 pan=0.4 gv=0.01 1/8 @0.67 "0.02%3.6%0%0.3"1: div=31 5.0 5 10 13 18 23 28 (6.0 5 10 13 18 23 28 7.0)t t t

Taking out the now-understood first line, which sets up variables, and just illustrating now the note-events, coming down from middle C, we could do:

1: c b, a, g, f, e, d, c, b,, a,, g,, f,, e,, d,, c,,,


1: 0 -3 -8 -13 -18 -21 -26 -31 -34 -39 -44 -49 -52 -57 -62


1: 5.0 4.28 23 18 13 10 5 0 3.28 23 18 13 10 5 0

Notice that the octave before the decimal point, once set, becomes optional, until we want to switch to a different octave.

iii. Rational notation

Rational notation allows the user to compose and experiment with just intonation and rational intonation quite easily. The syntax is similar to numerical notation in its ordering, except that one doesn't have octave decimals: to get above or below the middle-C octave, you multiply the first number (the numerator) of a given ratio by a power of 2 (2, 4, 8, 16....) to get a higher octave, or do the same to the 2nd number (the denominator) to get a pitch from a lower octave. For instance, middle-C is 1:1, so c' above middle-C is 2:1 and the octave below middle-C is 1:2.

The syntax, more formally, is:

[articulation]N:D[legato end][ties]    

As before, we can illustrate rational notation using a simple example again, this time a scale in the virtual key of 'D', which we arrive at by using the "key=" variable setting:

# below we see the selection of a harpsichord soundfont preset# eighth notes are used1: i=12 "7%.01%.03" 1/8 key=9:8 1:1 9:8 5:4 4:3 3:2 5:3 15:8 2:1 t t t # now we rest for four bests and then hit a full D-major chord with 6 notes.# it's still D-major b/c we haven't reset the 'key=' statement to equal "1:1"# we use a whole note value to make the chord a bit longer:1: r4 1/1 [1:2 3:4 1:1 5:4 3:2 2:1]

A note on ramping

One of the nice features of the Csound score is the use of the "<" character to set a p-field value to to the value interopolated between a starting and ending value found in other notes. This allows behavior such as smooth interpolations of parameters for crescendi/diminuendi, signal panning, etc. Ramping is available for all of the p-fields except for p5, which by the architecture of Microcsound, actually triggers the instrument and causes the time-line itself to advance. To do ramping of any p-field parameter, you must ensure that you supply a defined starting and ending value, as the rest is taken care of by Csound. There are a couple of things to remember as well: although p4 (attack) can be ramped by using "@<", pan and mix must use the less-than sign surrounded by double quotes: e.g. pan="<"'. A simple C-major scale below demonstrates simultaneously a crescendo as well as ramped panning:

#below is the setup for a chip-tune instrument from 'microcsound.orc'1: div=12 i=6 ".02%.02%2%.6" mix=0.65 t=90#now we pan left to right and crescendo simultaneously#csound performs all the interpolation for us.1: @0.3 pan=0.1 c, @< pan="<" d, e, f, g, a, b, @0.8 pan=0.9 c4

Using the time pointer for one-off counterpoint

Time in Microcsound by default marches along by beats, advancing whenever an event (note, chord, rest or trigger) is found in the score. However, there are certain instances where one may wish to modify the onset time so that one may go back and add additional layers of counterpoint, or set the clock to zero, etc.

For example, suppose we have a piece where a monophonic voice sometimes need to split off into dyads. We could set the time pointer to return N beats back by using "&-N", for example, to go three beats back and add an addiontal line of notes, we would type "&-3" (without quotes) in our .mc score.

Here is an example of the beginning of the melody "For the Beauty of the Earth", set to contain a bit of 2-voice counterpoint towards the end, but "embedded" in one voice:

#example showing resetting the time pointer1: div=19 i=10.1 "0.02" (g2 ^fg a2g2 c'2 c'2 b4) | e2 ^f2 g2 e2 d2 .d2 d4 1: &-16 i=10.2 "0.02" (c4 b,4 a,b,c2 b,2 g,2) ||

Notice the use of fractional instruments so that Csound understands how to handle multiple instances of legato ties using the same instrument within the orchestra.

The time pointer can be set to the beginning of the piece by using "&0", and if you need to advance the clock for any reason (typically, there are not many reasons for doing so) you can use "&+N" where N is the number of beats of the current rhythmic time value (set with the fraction, as explained earlier). In addition, one can go to any arbitrary beat number based on the default rhythmic length times the value of N. For instance, if the current rhythmic value is "1/8", to go to the 17th eighth note in the piece, you could write "&17".


II. Odds and Ends

More on Interactive mode (calling the script with "microcsound -i")

The use of interactive mode by calling Microcsound from the command line with the "-i" switch is helpful for various situations, including making quick sketches, performing acoustic demonstrations or tuning demonstrations, creating droning sonorities, tuning an acoustic instrument with a suite of test tones, etc. The difference as far as the script is concerned is that all the input data flows in a classic read/eval/output loop familiar to anyone who has used an interpreted programming language from LISP onwards, such as Python, Scheme, PERL, Basic, Lua, etc. To end the read cycle and have Microcsound evaluate and output sound based on the text entered from the live interpretive loop, you must write "done" followed by a return all alone on a single line. The example below illustrates a C-major chord being played by one voice while a 2nd voice descends a scale at the same time. You will also get a feel for the kind of output you are likely to see from Microcsound, and from the background engine messages from Csound, which is ultimately being called to process the sound:

akj@myhost:~$ microcsound -imicrocsound--> 1: div=1 i=12 "7%.01%.02" t=90 1/1 [1:2 3:4 1:1 5:4 3:2 2:1]microcsound--> 2: i=12 "7%.01%.02"  1/16 1:1 t 15:16 5:6 3:4 2:3 5:8 9:16 1/2 1:2 microcsound--> donemicrocsound--> doneok...processing sound nowaha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!aha, a ratio!executing csound command...PortMIDI real time MIDI plugin for Csoundvirtual_keyboard real time MIDI plugin for CsounddBFS level = 32768.0Csound version 5.13 (float samples) Jan 19 2011libsndfile-1.0.21orchname:  /home/akj/audio_and_midi/csound_files/microcsound.orcscorename: /tmp/Microcsound.scortmidi: PortMIDI module enabledrtaudio: ALSA module enabledorch compiler:	opcode	tieStatus	i			opcode	AllPass	a	a		instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr			instr		sorting score ...	... doneCsound version 5.13 (float samples) Jan 19 2011displays suppresseddBFS level = 1.0del: -11959.000000del: -11959.000000del: -11959.000000del: -11959.000000del: -11959.000000del: -11959.000000del: -4870.000000del: -2346.000000del: -4308.000000del: 1902.000000del: 1902.000000del: 1902.000000del: 1902.000000del: 1902.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -12000.000000del: -619.000000del: -8154.000000del: -8154.000000del: -10057.000000del: -10057.000000del: -10057.000000del: -10057.000000del: -10057.000000del: -10057.000000del: -10057.000000del: -10057.000000del: -10057.000000orch now loadedaudio buffered in 2048 sample-frame blocksALSA: -B 4096 not allowed on this device; using 3763 insteadALSA: -b 1881 not allowed on this device; using 940 insteadwriting 8192-byte blks of shorts to dacSECTION 1:Score finished in csoundPerform().inactive allocs returned to freespaceend of score.		overall amps:  0.73753  0.73753	overall samples out of range:        0        0errors in performance8192-byte soundblks of shorts written to dacmicrocsound--> 

The example above shows an interactive session in Microcsound, displaying a mixture of Microcsound text output followed by text output from Csound itself. As you can see, anything that can be done in Microcsound's normal compilation mode can be accomplished in real-time interactive mode as well, provided that the computer processor and memory can handle the real-time audio processing power required.

Tips for creating instruments that work with Microcsound

When designing Csound instruments that will work with Microcsound, all of the p-fields from 1-7 must be reserved for the following values:

You should always start with these in mind when designing new instruments or modifying old ones to integrate into a Microcsound orchestra. As mentioned in the section regarding extra p-field variables, you can optionally use p8 and up for any number of additional instrument parameters. Here is a list of tips to remember when designing instruments:

  1. Make your instrument expect any pitch from p5 to be in hertz (HZ), or cycles-per-second. Microcsound converts all note pitch values to HZ itself, and no conversion will need to be done in your Csound code, such as using cps2pch.
  2. The instruments should accept (expect) an attack value between 0.0 and 1.0. This is typical for 0dbfs type design, anyway.
  3. When designing a legato instrument, remember that Microcsound will send a negative duration to p3. Please refer to Steven Yi's excellent tutorial regarding legato instruments in Csound ("Exploring Tied Notes in Csound"--listed in the references) to know what kinds of code (including user-defined opcodes) to use to handle legato without glitches. Another point to remember is to use fractional instruments from the Microcsound score when you have multiple parts using the same instrument, and at least one of them is legato. Csound needs to know that each instrument is a unique voice, and it must know what legato ties belong to what voices. An example of using a fractional instrument in Microcsound would be setting the instrument variable like so: "i=3.1" .... note the decimal point setting this instance of instrument 3 apart from all others.
  4. Pan and Mix are also designed to use a range of 0.0-1.0, so please design all instruments accordingly for that range of values.
  5. One tip for using percussion instruments with Microcsound: use helper instruments. For example, if you look at the code of microcsound.orc, you will find instrument 120, a virtual drum kit. This instrument reads p5 from Microcsound and triggers another instrument in the orchestra which corresponds with the value of p5 plus some constant (remember, this is possible due to using div=0). This allows actions analogous to MIDI key numbers to trigger certain drums. Anyone familiar with MIDI GM drum kits will find such a paradigm comfortable.

Using a mixer or effect instrument from within Microcsound

One of the things we often use in our music work is a mixer. In Csound, mixers are often implemented using the ZAK patching system or one of the newer channel opcodes. The author is more familiar with the former means and as a result the default orchestra file, microcsound.orc, uses the ZAK patching system to set up a default mixer and reverb effect. When you look at the microcsound.orc file, you will find that instrument 200 is a mixer, instrument 201 is a helper instrument that triggers the mixer instrument (instrument 200) to turn on, and instrument 202 provides the user the ability to modify the 4 variables that are used in the mixer instrument: dry volume, wet (reverb) volume, reverb amount and reverb cutoff frequency. These variables are accessed via p8-p11 for instrument 202, and are modifiable from Microcsound by using the percent sign syntax, as discussed above. For example, setting the dry volume to 1, wet volume to 0.5, reverb amount to 0.75 and reverb cutoff frequency to 7430 would mean writing the following extra p-field string before any trigger event: "1%0.5%0.75%7430" (double quotes here are literal--they must be typed!)

Note: the use of instrument 201 to turn on the ZAK mixer in microcsound.orc is no longer necessary and deprecated as of Microcsound version 20100803--there is a default score trigger for this instrument within Microcsound.

The following example assumes the use of microcsound.orc and illustrates the changing of the default settings for the mixer/reverb global ZAK instrument from within Microcsound:

1: div=0 i=202 "1%0.5%0.8%5500" 11: &0 i=11 1:1 2:1 t t 7:4 t 1/16 3:2 7:4 5:3 7:4 t t 1/8 3:2 5:4 9:8 1/2 1:1

An example of Microcsound at work

Users interested in digging deeper into how Microcsound works in a larger piece than the examples shown here can look at the source file for the author's own work, Puhlops and Laugua's Big Adventure, written in 17-edo back in summer of 2009. You can find the .mc source file at

The nice thing about this example is that it not only illustrates using microtonality with typical synth instruments one might design in Csound; it also illustrates using drum techniques from within Microcsound.



I hope you will try Microcsound and find, as I have, that it is a useful and flexible front-end that speeds up and eases the process of using and interacting with Csound. While Microcsound is designed for composers interested in microtonality and Csound, I do hope that there is sufficient evidence that this program is general enough to serve the needs of many others, including those interested in didactic demonstrations, algorithmic music, creating musical sketches, and using reference pitches easily from the command line.



I would like to thank Steven Yi for inviting me to write about my software for the Csound Journal. Also, a big thank you to the once, current and future developers of Csound itself, for your magnificent and generous code, code that creates one of the most powerful pieces of audio software the planet has yet seen.



  1. Richard Dobson, "Designing Legao Instruments," The Csound Book: Perspectives in Software Synthesis, Sound Design, Signal Processing, and Programming, 2nd Edition, pp. 171-186, 2001.
  2. Steven Yi, "Exploring Tied Notes in Csound", Csound Journal, Issue 1, 2005.
  3. Microcsound Homepage: