August 2010

After a long break, here is the 12th in the series (click here for the 11th one). It is an alapana in another rare, old and attractive raga. Can you also guess the artist?

Audio courtesy: Sangeethapriya

A hint: Your instincts may not be right 😉

Click here for the answer.


This is a Part 2 of a post that tries to highlight some of the rhythmic wizardry employed by Carnatic musicians during kalpanaswaras.1 Click here for Part 1.

Note on terminology – matra and akshara: It has been mentioned that Carnatic Music suffers from ambiguous and thus confusing use of terminology because there is no consistent adoption of standards. I have used akshara to indicate outer division of tala cycle (so 8 aksharas per Adi tala cycle), and matra to indicate sub-division within that. This is used by many, but some authoritative experts would deem this as incorrect, and what I refer to mAtra here i.e. the inner division should be akshara. At least, for the purposes of this article, I continue to use akshara for outer division, and matra for inner division.

In part 1, I discussed a sarvalaghu style kalpanaswara rendition by Semmangudi Srinivasa Iyer (SSI) and contrasted that with a “more complicated” example from T. N. Seshagopalan (TNS), the latter being one that would be labelled to involve “kanakku” implying rhythmic complexity. We saw that in that example, TNS was weaving a melody with a 5-based rhythmic pattern in the context of a song that up to that point had a 4-based rhythm (and although not shown eventually it all goes  back to 4-based pattern leading to the conclusion of the piece).

In this part, I elaborate on a more sophisticated form of rhythmic wizardy. Listen to this sample from Tanjore S. Kalyanaraman (TSK). It again starts with the violin playing a standard 4-based tune (i.e. like sarvalaghu style). However, pay particular attention to what follows (from 0:20 onwards):

What is happening here? Again, we can see a melody that appears to have a complex rhythm in comparison to the initial (violin) part. Here is the notation:

As you can see just like with TNS’ case, TSK has also switched to a 5-based pattern (swaras above are basically grouped in 5 mAtras), but there is a difference.

In TNS’ case, the duration of a matra (same as shortest swara/pause in this case) in the 5-based pattern was still the same as the duration of the 4-based pattern.  So although the performers keeping tala would stick to the original 4-matras per akshara spacing for tala-keeping, in this localized part the rhythm of the melody is no longer in sync with the inherent beat of the tala keeping (and that is part of the challenge of executing it). It is now 5/4th of that beat (i.e. 5/4 * 4-matras = 5-matras). Does it mean that the tempo has changed? I am not 100% sure but I think not. Atleast there is no perceivable shift in pace. There is definitely a perceivable shift in rhythmic dynamics, the “gait” changed from a 4-based one to a 5-based one during that sample.

Now, in TSK’s case, he has also switched the gait of the melody to a 5-based one but he keeps the akshara duration same but shrinks the matra duration so that five instead of four now fit in an akshara!  So for example, if say you had a 8 second tala cycle for Adi, and thus this is divided into 8 parts, i.e eight 1-second aksharas. If each of those are divided into 4 (equal of course) matras, then each matra is 250 milliseconds. Now, instead if you want to divide that 1-second into 5 (equal) matras with the same, then each matra would be  200 millseconds. Thus the shortest unit (swara or pause) in the notation above is 200 milliseconds, as opposed to 250 milliseconds in the preceding violin part.  So does this mean that the tempo has increased? Perhaps it is not apparent here – but will be so soon 🙂

The above is technically is in khanDa gati as per carnatic music i.e. the term for the gait where the akshara of tala is divided into 5 matras. Note that in a strict rhythmic point of view, the TNS’ sample is also in khaNDa gati since it is a 5-based rhythm.

Taking it to next level

Now in this particular rendition, Shri Kalyanaraman does not stop with this. Later, he turns it on with the following:

Now what is this – some of you may wonder!

He is basically taking the switching of gait by “shrinking” the matra duration to the next level, by doing it for 3, 4, 5 and 6 matras – all within one iteration of kalpanaswaras!  Note that here, initially the rendition is in a 3-based rhythmic gait i.e. tisra gati. TSK actually switches to it after the khanda-gati section shown in the first sample, and I have skipped that part. This 3-based tisra gati gait should be apparent from the mridangam beat at the start (123 123 i.e. ta-ki-Ta ta-ki-Ta). When TSK starts the swaras he first does tiSra gati (3 matras for the same original akshara duration), then catusra (4  matras), then khaNDa (5 matras), and then tiSra but in double speed (6 matras).

Do you perceive a shift in tempo in the above sample – an increase from the start to the end? It may seem so even though if you kept tala the beat that is there in the tala keeping will not change – since the akshara duration does not change!  Why there is a perceived tempo shift may be because the matras, and thus the swaras are  getting shorter and shorter, and in effect accelerating the tempo. We all can naturally “double” the tempo say go from 2 to 4 to 8 and that is no big deal. The interesting part here is that TSK is doing it in more gradual but precise terms.  From the initial 3-based one, he increases by 4/3 (133%), then 5/4 (125%) of that, then 6/5 (112.5%) of that.

Here is the swara notations for that part.

Note: The dashes (-) indicate musical phrasing separation. Also the notation for tisra-mEl kalam may not be 100% accurate in that it may not be starting on an akshara as illustrated above (see below). Nevertheless the duration of 6 of those swaras is the same as an 5 for khaNDa, 4 for catusra etc.

From the notation we can see the following:

  • In the first part, which is in tiSra gati i.e a 3-based rhythm, and we have three swaras per akshara – G D P  for first,  G R G for second.
  • For the second part in catusra gati (4-based) we have four swaras per akshara: D R’ S’ D for first, P S’ D P for second.
  • For  the third part in khaNDa gati (5-based), we have five swaras per akshara – D R’ S’ D P for first, and G P D S’ P for second.
  • For the fourth part, we are in tiSra gati again but double the speed of the first part and thus 6 swaras per akshara: D R’ , S’ D P  for first and D S’ , D P G for second.

Here is another way to look at the above. A video that illustrates that TSK is indeed fitting more swaras within the same duration (may be more illustrative to view in full screen):

Note: The blue lines correspond to the akshara boundaries. I have marked the individual matras in the audio waveform at some places using numbers 1, 2, 3 etc. Note also that per this, the 6-based rhythm actually starts “before” the akshara boundary – at least that is how it appears.

Continuing the above rendition, what follows is the violinist response for above i.e. doing 3-based, then 4-based, 5-based and then 6-based.  Then you have TSK doing an iteration where the above is done in reverse (i.e  6 to 5 to 4 to 3 and then 2 also) along with the violist response (which does not strictly follow it):

I leave it as an exercise to the reader to map the swaras TSK singing to notation 🙂

There is more!

But wait! There is more! Some of you may wonder – hey how come he left out miSra gati i.e. the 7-based one in that progression. He does that immediately afterward 🙂 :

Note that the 7 is used predominantly as a 2+2+1+1+1  (e.g: d,d,pdp) pattern.

Why all this? Granularity of Time

So what is it that TSK is exhibiting here?  I am always intrigued by gati switches but recently I looked more into the implications, and  am now even more impressed.

As I mention above, the overall akshara duration (the blue lines in that vide sample) is practically constant and in this case, it is about 600 milliseconds i.e. a little more than 1/2 a second. Given this, the shortest note in that second sample above (with the gati switches of 3,4,5 and 6)  is one that takes 1/6th of an akshara duration, and is thus 100ms. The next shortest note is 1/5th i.e. 120ms. So the artist is aiming to show a mastery of timing via 2 notes that differ in length by about 20ms i.e. 1/50th of a second! Once you consider in the miSra gati i.e. 7-based one, we now have same 600ms duration split into 7, and so each note is roughly it is about 86ms. Now the difference between this and last 1/6th one is 14ms, an even finer resolution!

Of course,  if you measure at the micro-level you will see some jitter around this target time resolution as  humans are not robots. But isn’t this impressive nevertheless? Isn’t the above concept of dividing a certain beat interval into different even and odd parts pretty interesting and also quite complex? This is why many people think carnatic music has one of the most sophisticated and demanding rhythmic framework.

Of course there are many different ways of exhibiting ones’ talent (vidvat) in kalpanaswaras – many more techniques, many of them I don’t even know about! The above is just a sample – a sample I find very intriguing. I hope you enjoyed understanding the innards of the above techniques as much as I did!

A Note on the demands on the Accompanying Artists

First of all, I apologize that I do not know the accompanying artists in the above samples.

One should also note that all of the above is pretty much done on the fly. TSK may not have come prepared as to whether he is going to do these things that day for that song. Most often these are a product of on-the-moment inspiration and hence it is improvisatory. Thus one can imagine that the challenge on the accompanying players is quite incredible. Both the violinist and the mrdangist have to immediately perceive the underlying rhythm pattern the vocalist is weaving with all its complexities. The mrdangist has to immediate follow along as soon as possible for best effect (you can see in the mix-gati sample, he catches on very quickly). The violinist has to be able to produce a response that captures all the essential details (pattern, gait, as well as swaras being dwelved upon) – it of course need not and probably shouldn’t be a exact copy. But it has to be “in the same spirit”.  I am always amazed how quickly adaptive they are although in cases like this where they are accompanying a supremely talented musician like TSK, they are often stretched to their limits and it may show i.e where things aren’t fully synchronized all the time (you can see a bit of this in the miSra gati sample)



  1. kalpanaswaras: For the uninitiated, this is an improvisatory section of a carnatic music piece where in the artist weaves melodic patterns consisting of the Indian solfa syllables adhering to the grammar of the raga of the piece.

A short detour before the 2nd of part of Kanakku vs. Sarvalaghu in Carnatic Music. But this one is related to it.

When I was analyzing that “rhythmically complex” piece by T.N. Seshagopalan featured in that post, I was trying various different ways of confirming my analysis that the rhythm of that piece was a 5-based one. As I had indicated in that post, I was having a hard time doing this since (a) I found the rhythm “non-intuitive” and complex and (b)  I always struggled with rhythmic aspects of carnatic music more compared with the melodic aspects (guessing ragas, figuring out swaras etc.).

One of the methods I tried was to use Garageband on the mac to recreate the tune, set to the same tempo as the original so that it syncs with it. This way, I can find out the correct “lengths” of each note which would confirm if it is a 5-based rhythm or really a unique use of 4-based rhythm or something else.

When doing so, pne time,  just for kicks, I assigned a distortion heavy electic guitar as the MIDI instrument to my recreation of the tune – and wow! I was stunned by the effect!!!  I thought This just rocks!!.  So one thing led to another and eventually to the end-result below.

A few heads ups that you should read before you listen to the piece.

  • I have used some “artistic liberties” here to present a “situation”. Someone (say me 😉 )  hears the original carnatic tune on a radio, the tune getting stuck in their head (this definitely happened to me), resulting in what follows. That is why at the start, you have the original carnatic snippet albeit with some “radio” disturbance, as well as some strange echoing i.e. the tune swimming all over someone’s head as they tossed in their sleep (no comment!).
  • A good headphones is the best way to get the full effect! And full, maximum volume is certainly mandatory!!
  • It is basically just one riff, one tune, repeated pretty much throughout but in different textures – carrying different energy level. And given that it is 4:30 long, it may seem like overkill, and probably is 😉 !!  But that is a reflection of how much the tune filled in my hand (that I just had to “get it out”!).
  • For those carnatic fans amongst you, the end tune has no resemblance to carnatic  because it was not intended to be so. I made no attempt to even try to introduce even a touch of it since I liked the rocking nature of the riff as such and just ran with it.

And oh-yes – “Five on Eight” is a play on words – can you see it?

Thanks to T. N. Seshagopalan and other artists for being the creative inspiration behind this.

And thanks to “vk” for being the guinea pig ;-)!!