Vidya's thoughts on RGP (Fwd: RGP)
Suraj Kumar
on RGP, you can simply do a reply if you have something to tell her.
My comments follow in a separate reply.
thanks,
-Suraj
---------- Forwarded message ----------
From: Vidya Jayaraman <vidy...@gmail.com>
Date: Jul 13, 2006 11:20 PM
Subject: RGP
To: sun...@gmail.com
Hi Suraj,
Just some thoughts based on what I understood after reading you RGP
write up..Some of which you might have already thought up.I am
assuming you are looking at - An online database of raga phrasal
patterns (or genes as you call it) with illustrative clips plus some
representational format to encode the information.
-----------------------------------------------------------------------
Some generic constructors for a raga will be :
Notes:
- The obvious arohana and avarohana constructors
- The key note (graha ,amsa nyasa svaras,alpa swaras - ie notes which
allow rest, notes which predominate in occurrence which allow being
used as a starting point for phrases etc,notes which are of minimal
occurrence and usage)
Gamakas:
- Notes are oscillated, which are NOT to be oscillated.
- Size and Range of oscillations
- Slides and other gamakas
Phrases:
- The Main and Repetitive signatures,phrasal patterns
- Jhanta swara patterns and other peculiarities (eg. gG rs in Darbar)
- Special and Supplemental phrasal patterns ,Accidental notes
- Range / Sthayi limitations , Madhyama sruti requirements (navroj, kuranji)
- Things like non-occurrences , varja phrases
Time measure issues:
Kalapramanas sung in- speeds (Ahiri and pace)
Relationship with other ragas:
- Janya Janaka Relationships and inherited properties acquired out of
that Eg,. Gandhara in Suddhaseemantini,asaveri etc gets inherited from
the property of Todi and other melas in 6-12
- Metamorphosis of ragas - Changes that happened
-Variants of the same raga.Variants in schools.
-Allied ragas
-Arriving at other ragas by shift , leaving out patterns, zigzag patterns etc
I suggest you leave out time of singing (morning,evening etc) and this
rasa theory.They are subjective and have no rhyme or reason.
Todo:
So one needs to come up with some way to encapsulate this info whether
it is based on MusicXML or a simple database table structure or
anything else that you have in mind.
-----------------------------------------
2.One could choose a set of ragas and try to implement this..
We should choose 5 ragas which give a wide scope to getting as much
information as possible.
Ranging from Kanada(repetitive key phrases), Begada (accidental
notes), Darbar-Nayaki or Madhyamavati-Brindavana Saranga (same scale
diff type of gamakas and note differences)
Todo:
Come up with a small subset of 5 ragas and get information on them
------------------------------------------------------
3.Other sources
1.I will supply some audio clips which I will be recording in August -
September.I am actually working on recording some encyclopedic raga
material sung and demonstrated by my father in law.We were planning to
record it raga-wise but based on a historical evolution.For example we
planned to record Nattakuranji - history, raga, key phrases, snippets
plus pretrinity , trinity and postrinity compositions.
I can share some of those clips with you (ie those that are relevant
to the raga part alone)
2.Once some initial work is done on this - I can also write to these
people (www.musicresearch.in) for their inputs.This will facilitate
information on other and additional ragas, plus any ideas others may
have.
------------------------------------------------------
This is all that I can that of at the moment.Hope this helps..If you
want to add me to the mailing list please use vidy...@gmail.com
Vidya
--
The absolute reality about absolute reality is that absolute reality
doesn't exist.
Suraj Kumar
| Some generic constructors for a raga will be :
| Notes:
| - The obvious arohana and avarohana constructors
| - The key note (graha ,amsa nyasa svaras,alpa swaras - ie notes which
| allow rest, notes which predominate in occurrence which allow being
| used as a starting point for phrases etc,notes which are of minimal
| occurrence and usage)
This is roughly the idea of the UI:
http://swing.rutgers.edu/raga-gene-definition-ui-mock.jpg
1. One can mark a note
2. Add a second note and the UI will automatically connect the two
notes.
3. Mark a point and call it 'the note' that is present in the gene.
So in effect, this UI simply records the 'movements' and not the
syntax itself. Thus making it possible for the user to define any
kind of ghamakkam / movement / etc.,. without making the system
'understand' what kind of ghamakkam exists. The power of 'tagging'
will enable the contributor to add info (if a raga has kampita
ghamakkam, the contributor simply tags it as 'kampita').
I think its important we don't bring in very carnatic-specific rules
here because if someone wants to use our work in building an
equivalent service for Hindustani ragas or for 'Tamil Panns', they
can use it without having to 'work around'.
Now, Let us try and define the arohanam of shankarabharanam (s r2 g3
m1 p d2 n3 S). Here is how the user will do it:
* Define a gene for 'sa-ri':
- Mark 'sa'
- Mark 'ga'
- Mark 'ri'
- Select 'sa' and call it a note.
- Select 'ri' (the terminating point of the gene) and call it a
'note'.
- Select the 'ri-ga-ri' section and call it 'variable'. (so
'sa-ga-ri' and 'sa-ga-ri-ga-ri' will be treated as the 'same').
* Define 'ri-ga': (pretty straight forward)
- Mark 'ri'
- Mark 'ga'
- the UI connects them. That's it!
* Define 'ma':
- Mark 'ga'
- Mark 'pa'
- Call the section 'variable' (so 'ga-pa' == 'ga-pa-ga-pa').
(Those who know the 'ma' ghamakkam in shankarabharanam's ascend
would probably find the UI much simpler here. Note how the system is
agnostic about the underlying rules of ghamakkams and is still able
to represent the ghamakkam well).
The UI will generate the underlying 'syntax' (yet to be thought
about) and the user can also enter the 'syntax' directly.
I suppose you have a pretty good idea by now to figure out how the
rest of the notes can be represented.
Now, this is just a 'gene'. Such a gene could be present in other
ragas as well. the Gene is independant of the raga. However, when a
gene is associated with a raga, it must be given a 'weightage' for
how important (frequently occuring) the gene is. So when a gene is
associated with a raga, one can therefore define vadi, vivadi,
samvadi, etc.,. simply by assigning weightage to the
swaras. Therefore, instead of 'bucketing' a gene into vadi, vivadi,
etc.,., one can put it on a more smoother, non-discrete scale and
still be able to find out vivadi, samvadi, etc.,.
Back in our shankarabharanam example, once the 7 genes ('sa-ri',
'ri-ga', 'ga-ma', 'pa', 'pa-da', 'da-ni', 'ni-sa') for the arohanam
are defined, you have a complete arohanam here. The arohanam must
also have individual notes (just a 'ga', 'da', etc.,.). Similarly,
the avarohanam will have to be defined. Then weightage can be
assigned for this raga ("ma-ga" gene gets a high priority, for
example).
While the process might seem cumbersome, what you should keep in
mind is that such a defined gene will be available for use in other
ragas. For example, Kalyani's 'sa-ri' will simply use the same gene
and will not need re-definition. Same applies to multiple ways of
playing a note (ex: flat 'ni' vs 'ni-sa-ni-sa' shake), there will
have to be two 'ni' genes and depending on the raga, one can assign
weightage.
I suppose I've explained the 'gene' idea in a very verbose way and
bored you enough already ;)
For argument's sake, if we write a 'melody generator' on top of this
data, we should be able to have some simple melody conforming to the
rules of shankarabharanam :) As and when we add more complex 'genes'
into the system, this 'melody generator' can surprise us as it picks
up all the rich definitions to produce interesting, rules-conforming
shankarabharanam.
Now, I've heard from two people thus far that this UI seems
complicated. Do let me know your suggestions on how we can simplify
it! (or look at the entire thing in a new way).
,----
| - Notes are oscillated, which are NOT to be oscillated.
`----
Doesn't matter. For representing a note that shouldn't be
oscillated -- define a 'gene' for the note that shouldn't be
oscillated (don't give it a 'movement'. Just define the note), give
it a very high weightage.
,----
| - Size and Range of oscillations
`----
I don't understand 'size' and 'range'. Can you please explain?
,----
| - Slides and other gamakas
`----
This case is covered.
,----
| - The Main and Repetitive signatures,phrasal patterns
`----
covered.
,----
| - Jhanta swara patterns and other peculiarities (eg. gG rs in
| Darbar)
`----
covered.
,----
| - Special and Supplemental phrasal patterns ,Accidental notes
`----
covered.
,----
| - Range / Sthayi limitations , Madhyama sruti requirements (navroj,
| kuranji)
`----
WOW. I didn't know such things existed! Thanks! We must think about
this. I'm tagging this as a TODO and putting this here:
http://swing.rutgers.edu/wiki/index.php/RGP_Requirements#Requirements_from_the_Experts
,----
| - Things like non-occurrences , varja phrases
`----
covered.
Time measure issues:
,----
| Kalapramanas sung in- speeds (Ahiri and pace)
`----
Added to above doc. Though, you will have to explain us in detail
what the issue here is.
,----
| - Janya Janaka Relationships and inherited properties acquired out
| of that Eg,. Gandhara in Suddhaseemantini,asaveri etc gets
| inherited from the property of Todi and other melas in 6-12
`----
I'm hoping, this would automatically appear when one views 'related
ragas'. But yes, we need to have a way to show 'parent', 'child',
etc.,. (added to doc).
,----
| - Metamorphosis of ragas - Changes that happened
`----
Is this really necessary? (I'm asking because I don't know :) ).
,----
| -Variants of the same raga.Variants in schools.
`----
Good point. Added to doc.
,----
| -Allied ragas
`----
Care to explain?
,----
| -Arriving at other ragas by shift , leaving out patterns, zigzag
| patterns etc
`----
This is already in our mind and will be one of the many 'tools' that
people can develop on top of this data. This 'bedham' tool, I'm very
interested in and we will keep it as part of RGP's native tools
itself. However, this is low priority. Our first goal is to provide
a platform to set/get raga info.
Experts, can you please explain in detail how 'bedhams' can happen?
(I have some minimal knowledge here but I would like to hear clearly
from one of you experts about this :) ).
,----
| I suggest you leave out time of singing (morning,evening etc) and
| this rasa theory.They are subjective and have no rhyme or reason.
`----
Yes, this is left out. However, we will have the concept of
'tagging' a raga (yes, the same tags as applied in the Web2.0 world
-- like how you can 'tag' your GMail mails or your flickr photos).
So its upto the user to create tags for such purposes if they wanted
(one could even create tags like "morning-raga", "creator:muttaiah
bhagavathar", etc.,. ...possibilities are endless :) ).
,----
| So one needs to come up with some way to encapsulate this info
| whether it is based on MusicXML or a simple database table
| structure or anything else that you have in mind.
`----
Yes, this will happen once we start the design.
,----
| Ranging from Kanada(repetitive key phrases), Begada (accidental
| notes), Darbar-Nayaki or Madhyamavati-Brindavana Saranga (same
| scale diff type of gamakas and note differences)
|
| Come up with a small subset of 5 ragas and get information on them
`----
Point. If its not too much to ask, can you please help put up this
info on the wiki in a format that we (developers) can understand? :)
(If you need any help with wikis, please let either me or pravin
know). Both of us logon to GTalk from our gmail IDs.
,----
| 1.I will supply some audio clips which I will be recording in
| August - September.I am actually working on recording some
| encyclopedic raga material sung and demonstrated by my father in
| law.We were planning to record it raga-wise but based on a
| historical evolution.For example we planned to record Nattakuranji
| - history, raga, key phrases, snippets plus pretrinity , trinity
| and postrinity compositions. I can share some of those clips with
| you (ie those that are relevant to the raga part alone)
`----
Awesome :) Thanks!
,----
| 2.Once some initial work is done on this - I can also write to
| these people (www.musicresearch.in) for their inputs.This will
| facilitate information on other and additional ragas, plus any
| ideas others may have.
`----
That would be a great boot-up!
I will try and summarise your inputs on a wiki when I get the time.
Initially Pravin and I had set a rough deadline of having some
design ideas, docs, etc.,. on the wiki by this weekend and some
minimal 'prototype' code by next weekend-ish. For all that I had
expected to happen, this is indeed going very well :)
Thanks,
-Suraj
sun...@gmail.com
Suraj Kumar wrote:
> So in effect, this UI simply records the 'movements' and not the
> syntax itself. Thus making it possible for the user to define any
"Not the Syntax itself" shouldn't be confused with the "syntax" that
the system will use to represent a 'gene'. What I meant by "syntax"
here is that it doesn't represent a movement as a kind of 'ghamakkam'
(ex: kampita, etc.,.) and instead it keeps it as just a 'movement'.
deponti
Yes....I have a sense of what you are trying to do, but certain things
are still confusing to me. (which aptly illustrates the difficulty of
this whole project.)
There is a theory of 22 notes in Carnatic music, which would help you
immensely in this project. Let me go through the music books that I
have and see if I can dig out this theory from them. Will let you know
if I can get my hands on some written pieces on this theory.
Deepa.
Suraj Kumar
Vidya Jayaraman
Pravin Shankar
Thanks for the link to Rasika2000. Its indeed a very interesting piece of work. They have designed a whole different notation for carnatic music, and their sound clips (purportedly auto-generated) are v cool too!
We need to read more about this software and ideally try it out to see how much it does what all we plan to do. But one fundamental difference is that this is a propreitary/commercial software written by one person, and a community platform will be far more beneficial to people than a propreitary one (the OSS philosophy!)
The software costs 39$. I'm thinking if its worth for me to buy it and try it out.. Will let you all know if I do.
Meanwhile let the discussions continue. The 22 sruti octave brings an interesting design issue, whether we should try to incorporate it in our design. I think even if we initially start with simpler ragams which fit well into the 12note octave, we should design our notations flexible enough to accomodate these microtones.
Among the "simple" ragams that you suggested we start with, Ramya was mentioning to me that Hamsadhwani actually has a Ni which is very close to Sa. I think we can approximate to 12 notes to start with, but should make the design flexible enough to be able to add microtone genes later.
- Pravin
Pravin Shankar
and also categorized Related work into book, software and website.
Please do let us know if you know of any other related work.
- Pravin