10. For BP1 old-timers...

The first rules of "-gr.dhin--", a grammar producing a North Indian qa'ida (a set of variations used by tabla drum players) look like:

The expression "4+4+4+4/6" indicates how musical items derived from "S192" should be displayed. In earlier versions of BP2 (before BP2.5.2) it was notated "4/4/4/4/6"; although this notation is still valid it should be abandoned for the sake of clarity.

First consider "/6". This is an explicit tempo marker, indicating that each beat of the metronome contains six sound-objects. (This is referred as chegun by Indian musicians.)

The sequence "4+4+4+4" indicates how beats should be grouped in a measure ( tala) containing 16 beats (referred to as tintal). Here the measure is divided into four sections of four beats. Consequently, a musical item produced by this grammar may for instance be displayed:

Beats are separated by periods (that could be replaced with tabulations in a word processor), and sections of the measure are displayed on different lines. Here the item covers two measures, i.e. eight lines containing 32 beats and 32x6 = 192 sound-objects (as expected).

If periods are inserted in data or grammar arguments, they are used by the compiler to change the tempo. For example, if a,b,c are terminal symbols,

The polymetric expression "{1,abba}" (meaning: "abba" is performed during 1 beat) is more flexible than "/4 abba". The former is a relative, and the latter an absolute, tempo specification.

Conversely, if a string of data is selected and " Show periods" is invoked (see the Control panel, type cmd-=), beat and section markers are automatically inserted. For instance,

10.2 Parsing items

Parsing ("Analyze selection" in the "Action" menu) is the attempt to check whether or not the arbitrary item you selected might have been produced by the grammar. Parsing is also called a membership test. There are severe restrictions in the grammar format that make this test valid. When compiling a grammar BP2 checks whether it is a " true" BP grammar (see reference manual §7) and then only it authorises parsing.

Grammar "-gr.dhin--" is an example of a "true" BP grammar. First you should compile the grammar (cmd-k) and generate its templates (see Control panel or "Action" menu). Then produce a few items in the "Data" window. You will notice that the produced items are displayed without any structural marker (brackets indicating repetitions, etc.). This display is a comprehensive "score" for Indian drum players. Structural markers will be automatically inserted when matching items against the templates.

Select one or several items and select "Analyse selection" in the "Action" menu (or click the same on the Control panel). BP2 will prompt you to use templates. Answer "yes". You may then decide whether you want to find only the first template matching each item and yielding a positive membership test, or check all templates to find which ones are acceptable. While the first method yields a quick assessment of the item in reference to the language generated by the grammar, the second will also point out structural ambiguity, e.g. the same item may be assessed "correct" by the grammar while matching several templates:

>>> Analyzing item: /6 dhin--dhagenadha--dhagenadhatigegenakadheenedheenagenatagetirakitadhin--dhagenadhatigegenakateeneteenakenadhagenadhin--dhagenadha--dhagenadha--dhagenadha--tagetirakitadhin--dhagenadhatigegenakateeneteenakenatin--takenata--takenatatikekenakateeneteenakenataketirakitatin--takenatatikekenakateeneteenakenadhagenadhin--dhagenadha--dhagenadha--dhagenadha--tagetirakitadhin--dhagenadhatigegenakadheenedheenagena

10.3 Learning weights

The process of weight inference has been introduced in (Kippen & Bel 1992). Given a set of items accepted by the current grammar it is possible to infer rule weights from the set of examples. The weight of each rule will be the number of times it has been used in parsing the sample set. Weight inference is used to improve the quality of randomly generated musical items by increasing the probability that these items resemble a set of items given as an input by an expert musician. See for instance the weights of rules in "-gr.dhin--".

Try this procedure as follows. Load "-gr.dhin--" and "-da.dhin--". Compile the grammar and produce its templates. (It's a good idea to save the current grammar with its templates, but remember that templates should be computed again each time structural rules have been modified in the grammar.) Now click "Save weights" on the Control panel to save current grammar weights in a "-wg" file. Then click "Set weights" and reset all weights to 0. Select all items in the "Data" window, click "Learn weights" and answer default options in all dialogs.

10.4 Keyboard encoding

It is possible to reprogram the keyboard in order to type certain strings ("tokens") with a single key stroke. This is generally used for the names of terminal sound-objects. Current keyboard encoding is visible in window "Keyboard" and may be saved with prefix "-kb". The name of the corresponding file (e.g., "-kb.kathak.azerty") is automatically inserted on top of the alphabet file. Conversely, inserting a name will prompt BP2 to load the declared "-kb." file as soon as the alphabet file is loaded.

Using window "Keyboard" is self-explanatory. To activate/deactivate encoding you may switch button "Use these tokens" on and off, or check "Use tokens" / "Type text" in the "Misc" menu, or type cmd-option t.

Only 52 characters may be mapped to tokens. Any string shorter than 255 characters may be mapped to a key as a token. Fig.30 shows the encoding used for Kathak material on an English keyboard (QWERTY).

Keyboard encoding in versions e 2.5 takes different keyboard layouts into account. Two layouts are currently supported: US (QWERTY) and French (AZERTY). The default layout is QWERTY, but if you change it to AZERTY when tokens have been defined, then the mapping is modified automatically so that tokens appear at the same locations on the keyboard -- except for the ones mapped to the 'M' key, which cannot be remapped. (When changing from AZERTY to QWERTY and vice-versa, the 'M' key is swapped with another key that does not contain alphabetic characters.)

10.5 Documentation of Indian music

An example of keyboard encoding is the material contained "-da.kathak tihais". When this data file is opened, it automatically calls the "-ho.kathak" alphabet which in turn opens the "-kb.kathak.qwerty" encoding. (You may change it to AZERTY if your keyboard is in French.) All tokens are bols used by (Lucknow style) Kathak dancers in North India, in a comprehensive transliteration designed by Andréine Bel. Numbers used as terminal symbols (in the so-called "number tihais " are notated #1, #2, #3... (pronounce ek, do, tin ... in Hindi, otherwise one, two, three ...) because ordinary integers are reserved for indicating durations.

Indian rhythmic material lends itself to descriptions by formal grammars, as exemplified by grammars "-gr.dhati", "-gr.dhin--", "-gr.dhadhatite" and their variants able to produce qa'idas (theme and variations pieces) in the Lucknow style of tabla drumming. Most grammars supplied as examples have been commented in our publications.

A grammar producing a famous Kathak number tihai composed by Pandit Birju Maharaj will be found in "-gr.12345678". More tihais have been implemented in "-gr.ShapesInRhythm". If you need to play lahras (cyclic melodic phrases) to practise Indian percussions, use "-da.lahras".

Indian raga music is also a good candidate for composition/improvisation models based on grammars and automata. The subtleties of alankara and gamaka may be rendered by performance parameters. Examples are still lacking since performance controls have been introduced recently. When Kumar S. Subramanian spent a few days in Delhi, he started composing items related to South Indian tunes. I have taken the liberty to insert some of his work in the example folder ("-gr.trial.mohanam" and "-gr.kss2").

We recently came to know that a team located in Mumbai (= Bombay) had devised an excellent method for the synthesis of tabla sounds. We hope to make it accessible to BP2 users shortly.