Shruti intervals - practice versus theory

[vinsim]

I always thought Indian classical music, with its harmonically related 22 shrutis, was fundamentally more "consonant" than western music, with its equal temperament tuning.

Then, I read these research papers..
https://ccrma.stanford.edu/~arvindh/cmt/

...which claims that even top artists (like LGJ, Ramani, U Srinivas) find it humanly impossible to hit the right shruti intervals (16/15,9/8,6/5,5/4,4/3,45/32,3/2,8/5,5/3,9/5,15/8).

To paraphrase the researcher, "All that matters is the way music is perceived and if shruti intervals are close enough (+/- 10 cents, say) to theoretical value, nobody notices the small discrepancy."

So, there goes my bubble and I find myself a bit disappointed with the realization that Indian shrutis (the way it is practiced) are no more consonant/dissonant that western shrutis. We might as well switch to equal temperament tuning - after all Mandolin Srinivas, Guitar Prasanna are all producing great Carnatic music with it.

I, then, measured the fret lengths on my Veena (I am a novice student) and noticed that it is strikingly close to equi-temperament already - I hadn't noticed it all these years. It also seems to be what is recommended in this paper ("On the Twelve Basic Intervals in South Indian Classical Music" - section 5. Instrument Constraints).

I would like hear more from those interested in this topic, and find time to go through some excellent research posted on the above link.

[vasanthakokilam]

Good subject line, vinsim. That is what it is. We have had extensive discussions at the erstwhile sangeetham.com and probably at rasikas.org as well. I can not find the links now.

Arvindh currently works at Apple as an Engineering Manager, Senior Audio Scientist. Not sure if he has any further info to offer. If anyone knows him, see if he might offer some info in this thread. He has a linkedin and facebook presence.

Consonance or not, we still occasionally hear characterizations like 'in Raga X, the M2 is sharper'. Some rasikas hear the difference. I usually can not. I wonder, if such notes are executed absolutely flatly without any oscillation then these things can be talked about somewhat meaningfully but in the presence of oscillations which is the norm in CM, how do we determine the swara that is sharper or flatter than the nominal swarasthana by such small degrees?

Talking of executing notes flatly, on the flute, one note where I can differentiate is the G in a raga like Sivaranjani. There is an intermediate position between G3 and G2 which sounds different from G2 but does not sound apaswaramish. But people can tell it is not the Sivaranjani G2. Similarly ( not in the sivaranjani context ), there is a position between N2 and D2. I have no clue on the technical aspects of this. ( It is the same finger position for both pairs, in the latter case it is with the standard Sa position and in the former case, it is with the lower Pa as Sa. ). This may just be an artifact of the construction of the flute.

[arunk]

equi-tempered is a very good approx with intervals for major swaras like pa (700 cents) and m1 (500 cents) being extremely close to the natural ratios of 3/2 (701 cents) and 4/3 (498 cents) - but g3 (400 cents) overshoots "significantly" as the natural ratio for it 5/4 is 386 cents, and a flat long antara gandara is essential to many ragas.

Other swaras like r1, g2, d1 etc. are somewhat variable in practice - depending on raga, context (i.e. gamaka), and iirc Arvindh's analysis is also that different musicians also show variability within that. It has been a while since I read his articles and so I may be remembering wrong.

Arun

[mahavishnu]

Not sure how relevant it is to this discussion. Here's something I read by Jamshed Bharucha today on the topic of consonance. We discussed Jamshed's work when we talked about musical expectancy structures some months ago.

I spotted this piece when reading a larger article on edge.org where several scientists were asked what their most elegant explanation was in science.

Beauty and Tragedy In the Mathematics of Music

Some things are too good to be true. Others are too good not to be true. The elegant correspondence between musical consonance and simple mathematical ratios is too good not to be true. The infinitely creative system of musical harmony—based on a small set of canonical intervals and chordsarises out of this correspondence.

And yet the system has a flaw—a rip or tear in the musico-mathematical universe.

The most consonant musical interval is the octave, so much so that two tones an octave apart bear the same letter name, for example, A. The next most consonant interval is the fifth (for example, A to E). If you pluck the A string on a guitar and then pluck it again while placing your finger halfway up the fingerboard (allowing only half the string to vibrate and doubling the frequency), the pitch goes up by an octave to A. If instead you place your finger one-third the way up the fingerboard (allowing only the remaining two-thirds of the string to vibrate and increasing the frequency by a factor of 3/2) the pitch goes up by a fifth to E.

On a full size piano keyboard, if you play the lowest note, A, and then go up by twelve successive fifths, you reach A again—which happens to be seven octaves higher than the lowest A. Calculating by fifths, the highest A is (3/2)12 times the frequency of the lowest A. Calculating by octaves, the highest A is 27 times the frequency of the lowest A.

It should therefore be the case that:

(3/2)^12 = 27

Tragically, this is not the case: (3/2)^12 is 129.7463, while 2^7 is 128. Close but unequal. If you tune by octaves, the fifths are out-of-tune, and vice versa.

This discrepancy—a variant of the Pythagorian comma—shatters an edifice that rests on the beautiful principle that ratio simplicity underpins musical consonance. While each interval is most purely tuned using these simple ratios (the just tuning system: 2/1 = octave, 3/2 = perfect fifth, 4/3 = perfect fourth, 5/4 = major third and 6/5 = minor third), the mathematics falls apart when these intervals interact in music.

The Pythagorian comma has haunted string players for centuries, requiring annoying adjustments. On the violin, some notes must be played slightly sharper and others flatter to dodge dissonance. The comma held back the development of keyboard instruments, in which notes cannot be adjusted while playing, thus limiting a composition to the musical key for which the instrument was tuned.

The solution to this problem is to forego mathematical purity by fudging tuning. This discovery—which emerged in various forms over a period of centuries—unleashed new creative possibilities and inspired J.S. Bach to write The Well Tempered Clavier, which traverses all major and minor keys without discriminating on the basis of dissonance.

Today, several tuning systems achieve this end. In equal tempered tuning (used in some electronic synthesizers), the octave is divided into twelve equal semitones on a logarithmic scale. In this system, each successive semitone is 21/12 times the frequency of the previous one. Thus a fifth is 27/12 or 1.498 instead of 3/2 or 1.5; the perfect fourth is 25/12 or 1.3348 instead of 4/3 or 1.3333; the major third is 24/12 or 1.2599 instead of 5/4 or 1.25; and the minor third is 23/12 or 1.1892 instead of 6/5 or 1.2.

By anchoring the octaves and sprinkling the error elsewhere, consonance is compromised slightly but never disturbingly. The edifice of musical harmony is preserved by systematically fudging the very mathematical system upon which it is built. An elegant way to fix inelegance in an elegant musico-mathematical system.

------------

For those interested, there are several other interesting essays here

http://www.edge.org/responses/what-is-y ... xplanation

[vasanthakokilam]

Thanks Mahavishnu. I need to read up on the rest of the articles. The very first one, 'why is the world comprehensible' is my cuppa tea I have put it aside for now without reading it to get my own thoughts in order before I read the professor's take.

On this natural ratios, I am confused about one thing now.

There is musical consonance ( psycho acoustics ) and there are those simple mathematical ratios. Leaving side the Piano problem, what problem string instrument players have in sticking to natural ratios. I understand the Pythogorian problem of successive fifths but I do not understand why that is even brought in to the picture. Why do we care if things do not converge 7 octaves out. Is this a problem of self-consistency of the mathematical ratios themselves as applied to music, with octaves following the geometric scale and harmonics following arithmetic scale?.

Secondly, I understand that this equal tempered tuning is an acceptable compromise and why that is so. In this modern era with programmable keyboards, assuming we can instantly reprogram it for any adjustments that are needed, how many perfectly consonant notes can we get in an Octave, without encountering any of the problems that the string players of yesteryears seemed to have had? Could Bach have written his Clavier without requiring equal tempered tuning just as well, had he had access to programmable keyboards with the required flexibility in making small adjustments to the frequencies as he switched keys, while keeping the tuning to natural ratios?

I am asking probably the same question many different ways and that is because I do not quite get the real import of the tuning problem.

[cmlover]

You said it right VK: psychoacoustics..
It is actually psychological. We are conditioned to like certain sounds as melodious by conditioning and indoctrination.
The child (for that matter the snake) are not fascinated by nIlambari; but it is the sound that attaracts them.
CM will make no sense to one uninitiated as the sonorous chanting of Vedas would also not..
Trying to justify using mathematical mumbo-jumbo is post facto...

[rshankar]

No idea about the child, but snakes do not have ears. They respond to the movement of the charmer....

[mahavishnu]

CML: As Ravi says, snakes do not hear anything.

Secondly, I don't think you have the right definition of psychoacoustics. We use the technical term "psychoacoustics" in terms of relationship between the physical characteristics of a sound and the perception of the sound by a human listening system. In a sense, this is the science of how the ear calibrates to the physics of sound. This is based on basic sensory psychophysics and is not culturally specified in any way.

Should one have basic hearing problems and see an audiologist, the basic hearing tests that are administered are based on psychoacoustics and the underlying mathematics of sound structures. Again, this is not culturally specified in any way. Concepts of tuning and interval spacing come into play here.

The perception of consonance and dissonance might be subjective, "psychological" as you call it and culturally specified and here you may have a point. But that is not relevant to what we are discussing here.

VK, I need to give your point more thought.

Could Bach have written his Clavier without requiring equal tempered tuning just as well, had he had access to programmable keyboards with the required flexibility in making small adjustments to the frequencies as he switched keys, while keeping the tuning to natural ratios?

According to Bharucha's theory yes. It is these small adjustments that are made below the threshold of human detection that help to approximate the errors in equal tempered tuning.

[cmlover]

"A snake has no ears. However, its tongue is extremely sensitive to sound vibrations. By constantly flicking its tongue, the snake picks up these sound waves. In this sense, a snake "hears" with its tongue. "
http://www.funtrivia.com/askft/Question56272.html
Even more, even the cattle flocking to hear Krishna's Flute (in spite of their external ears) is just poetic exaggeration..
I am aware of the definition of 'psychoacoustics'.
My point is to bring attention to the fact that "suggestion" plays an important role in appreciating CM rather more than the physical/physiological equipments.

Most of the time we hear what we want to hear unless we want to be critical!

[mahavishnu]

I am aware of the definition of 'psychoacoustics'.

If you say so. although it is not evident from your previous post.

[Nick H]

Joining the by-the-way --- I believe that snakes do not hear sound in the sense that we do, and are certainly not aware of the music of the charmer, but they do "hear" as in "are sensitive to" low-frequencies. Thus they are able to sense the approaching steps and get out of the way.

If anyone wants to redefine psychoacoustics into the sense of the relationship between psychology and sound (and perhaps the should be a different word for this. Perhaps there is already?) they need only visit a hifi forum or two, where they will find complete madness expounded, sometimes as justification for spending huge amounts of money! They will quickly find that they have joined a group where some people leave technology far behind and espouse what seems to me to be a form of faith-based religion. Such people might even try to convince that changing a power cord will result in different harmonies.

It theorise that the reason that some of my British friends find that Indian music makes them uncomfortable is that they find it subtly out-of-tune. I'm not much good as a researcher, though, because I begin from the standpoint of wondering how anyone could fail to like it!

[vasanthakokilam]

I take my share of the blame in using psychoacoustics loosely. ( http://en.wikipedia.org/wiki/Psychoacoustics ) ( http://en.wikipedia.org/wiki/Equal-loudness_contour )

What I meant to talk about is probably in the realm of Music Psychology ( http://en.wikipedia.org/wiki/Music_psychology )
specifically how we subjectively perceive consonance ( as in sounding good ) and dissonance ( as in sounding not good ). Though my reference was in passing, since we got into this, I do have a question. Is there a scientific reason why those natural ratios sound pleasant to us? Like s-S, s-p, s-m, s-g etc.

[mahavishnu]

VK, here is a paper by Dale Purves et al at Duke University that claims that musical intervals arise from the relationships of the formants in speech spectra.
http://www.pnas.org/content/104/23/9852.full

His website has a demo of tones that you could play to determine which ones sound pleasant and which ones do not. Please see: http://www.purveslab.net/seeforyourself/index.html?6.00

And another paper from their group makes a similar claim that humans prefer musical structures that have conspecific vocalizations, meaning that scales in music were possibly derived from what was capable with the human voice.

"The similarity of musical scales and consonance judgments across human populations has no generally accepted explanation. Here we present evidence that these aspects of auditory perception arise from the statistical structure of naturally occurring periodic sound stimuli. An analysis of speech sounds, the principal source of periodic sound stimuli in the human acoustical environment, shows that the probability distribution of amplitude-frequency combinations in human utterances predicts both the structure of the chromatic scale and consonance ordering. These observations suggest that what we hear is determined by the statistical relationship between acoustical stimuli and their naturally occurring sources, rather than by the physical parameters of the stimulus per se."

[vasanthakokilam]

>"meaning that scales in music were possibly derived from what was capable with the human voice"

Mahavishnu: Thanks for putting that in layman's terms like that. It took me a few readings of the quoted paragraph to realize that is what they are saying You have what it takes to be a popular science writer, which is one of the hardest things to do.

That makes intuitive sense. Put another way, wolves in the wild probably consider the howl very consonant. I am usually a sucker for explanations that use such inter-dependencies and reversal of cause and effect.

[mahavishnu]

VK, thanks for your kind words. Although I was just trying to get the geek speak to make sense in my own head
In a sense, this theory could be extended to account for our indifference to certain scales, vivadhi ragams etc.

[Nick H]

Objectively, some languages sound very dissonant to me. In fact, I rationalise this with the thought that, if we do not understand the meaning, then listening to another language is very like listening to music ...and some music we like, some we don't. Thus, I can listen to Italian longer than I can listen to French; Japanese longer than Chinese. The sound and experience of some languages is so utterly different to some others that I wonder how far the research for the above quote reached!

Mahavishnu, I know this stuff is your profession, so feel free to tell me I'm talking rubbish!

[l.gopalan]

In continuation of my blog under manodharma and in reference to yours, I seek further clarification and enlightenment of my views below:
1 Basically ourCM spectrum of frequencies are not discreet but continuous like that of a rainbow
2 So our CM and HM musical instruments should simulate the conditions of the rainbow with additional properties of playing Gamakas producing finer frequency variations within the allotted band on all its Stayies / actual practice of allowed swara variations there on for the chosen raga etc And should be able to slide across Stayies too
3 Further well orchested and deftly handled amplitudinal (Decibal) and speed variations can also come in handy to embellish
4 Our age old tradition therefore promoted musical instruments like Veena, Flute, and Nadaswaram as they were admirably suitable to meet the above conditions and did not depend on musical instruments which can play only fixed frequencies like the erstwhile harmonium and all
5 later, it is said, thanks to Thirukkodikaval Krishna iyer Violin was admitted into our CM as they found that it too can meet the said requirements
6 In my reckoning the requirements are met as below in these instruments:
6.1 in Veena though frets are there the oscillating length and the fingering techniques (Tension / variations on the oscillating segment of Tanthi) meet the expressed need where as fixed frets as in conventional harmoniums or jalatharangam cant match it
6.2 likewise Flute and Nadaewaram bring this about by finguring, blowing patterns based on pressure, speed, etc of blowing the air
6.3 As regard Violin this requirement is met much better as the strings can be made to oscillate and any length
6.4 It is said that modern key boards are being designed and are tending to achieve these requirements of our musical systems- Others may give their views
7- These points are advanced by me to get further clarifications on the following issue:
7.1Since CMs and HMs requirements are different from occidental scale requirements is it not better we speak of ratios of the allowed band with respect to Adhara sadja for 12, 16 and 22 Swarastanas by fixing the beginning point and ending point ratios of each one of the swarasthanas instead of fixing the ratio of intervals which suits better for western music and fix the frets accordingly.
8 I am advancing these thoughts for further dissection to accept or reject my views
9- It is only your valuable inputs which has trigered my thoughts -thank you for that-gopalan

[mahavishnu]

Objectively, some languages sound very dissonant to me. In fact, I rationalise this with the thought that, if we do not understand the meaning, then listening to another language is very like listening to music ...and some music we like, some we don't. Thus, I can listen to Italian longer than I can listen to French; Japanese longer than Chinese. The sound and experience of some languages is so utterly different to some others that I wonder how far the research for the above quote reached!

Nick, you might be on to something here, but I have an explanation that has more to do with timing and the implicit rhythmic structure rather than tone. As it turns out, English, Italian and Japanese (mostly) are stress-timed (as opposed to syllable timed) languages. Fluent speech in these languages requires one to accentuate or stress parts of a word (lexical) or part of a sentence (prosodic).
For e.g: What did you eat today? and What did you eat today? carry very different semantics.

French is syllable timed, although there are prosodic elements to it. Cantonese chinese is also syllable-timed, but Mandarin Chinese is a pure tonal language, probably the one that has the closest to actual melodic variations in speech. A very large number of Chinese people have perfect absolute pitch, something that is correlated to the sing-song nature of the language.

So your preference for Italian over French and Japanese over Chinese might have a lot to do with your natural disposition for liking stress-timed languages!

However the origin of chromatic scales and how we divide intervals is culturally (and linguistically) universal to a large extent. Essentially the question is: why did we choose to divide a musical scale by twelve pitches, each a semitone apart? (seen with different calibrations in equal-tempered tuning, Indian solfege/sarigama and even Chinese tuning). Some say that the origins of this have to do with the human voice and the limitations of the ear in perceiving anything stable between these frequency intervals (probably co-evolved traits in the Darwinian sense.

Sri Gopalan: Your question is a very interesting one. I am not qualified to answer a large part of it. The continuous frequencies you speak about (if I understand you correctly) is a characteristic feature of all forms of music that have a chromatic scale. In fact it is because of the rainbow analogy loosely that we even call them chromatic. Fellow member Uday Shankar who has worked extensively in both the science and art of tuning is patently the best qualified to answer your very interesting questions.

[Nick H]

Thank you for explaining why, to my very personal taste (often at odds with others!) that French is not the natural European language of poetry --- but German is!

[VK RAMAN]

Very interesting discussion

[girish_a]

Found this website:

http://www.22shruti.com/

Looks like he is a seller of musical instruments, but there's also plenty of technical stuff....Check it out.

[Shivadasan]

The website mentioned by girish_a contains a lot of valuable information for those who want to solve the problem of consonence and 22 srutis.

Some top ranked musician noted for their tunefulness have told me that after using the 22 sruti Harmonium they found that some of the notes they were playing did not produce the type of sweetness gained after using the values in the hamonium. They say they realized how off key they were. They said that when playing the harmonium they found that the melody produced by it appeared to be faultless and drenched in melody.

Dr Oke is a serious researcher and his harmoniums are not mass manufactured. He prepares the harmonium after ascertaining the sruti used by the artist. It is specific to a certain sruti.

[SrinathK]

That site is perhaps the last word on this subject, thanks a lot girish_a. I never knew so much material existed about this topic. The author has done some truly outstanding work!

In my experience I noticed the difference when playing g3 and d2 together and then playing sa and d2 together. The da had to be played slightly lower to align with the sa. I am also aware of some minute differences in ragas -- the R2 of kharaharapriya is perfectly aligned to the lower pa while the R2 (or as per this website, is it R1?) in sama is slightly lower -- but I never knew there was so much research carried out on this subject.

There is a virtual tanpura in that site that will allow you to play all the shrutis and the difference between a d1 and d2 and a R1 and R2 is exceedingly minute -- It reminds me of a violin string that goes very slightly out of tune -- you start hearing the beats fluctuate. A fraction of a turn on the fine tuner will bring it back in tune and eliminate the beats -- that's how fine the gaps are. I'm playing around with that tanpura now and I can see that R1 and P are slightly dissonant and you hear slow beats while R2 and P are perfectly tuned to the fourth. I envy computers in this regard -- they will get the notes right every time forever. For a human being, to get the exact position of swaras for each and every raga correctly will take many years, even decades of sadhana and I don't know if a human being has enough lifespan for that.

But here's a question. How does the theory of the 22 shrutis apply to gamaka oriented music as in our system? This could be used for plain notes and harmonic music, but what about the heavy gamaka oriented Carnatic music where the notes are actually labels for fluid phrases?

Also regarding the subject of shrutis and intonation at faster tempos, there is only so much a human being can do. There is a limit to the capability of the human body. At high speed even the great western virtuosos of the violin only hit the notes right up to that extent. Actually compared to a computer, no human being ever sings or plays in tune all the time, they are just correct enough or make corrections before it's noticed. Even in the brighas of GNB or the taans of Hindustani singers the notes can only be precise to that extent. Only a computer can hit all the notes, with all the correct intervals 100% properly 10000 times in a row. Normally what human beings do, they just adjust here and there so it sounds in tune at the moment of playing.

Also I am not very fond of the equal temperament tuning for Indian music so much. I've played and listened on keyboards and harmoniums. The g3 (anthara gandharam), m2 (prathi madhyamam) and n3 (kakali nishadam) are too flat, the r1 (shuddha rishabham) and d1 (Shuddha dhaivatam) is too sharp and the panchama is slightly off too!

All said and done, despite the limitations of instruments and human beings and tuning systems, good music will always be good music.

[vasanthakokilam]

In my experience I noticed the difference when playing g3 and d2 together and then playing sa and d2 together. The da had to be played slightly lower to align with the sa.

Really? Interesting.

Of late, I have been thinking of a method to get swarasthanas right which is basically to sing each note with Sa as the base.
That is, instead of S R G M, it is SS SR SG SM and after some practice, do not vocalize the first Sa ( may be just a hiss under the breath), so it just sounds like one is singing the Arohana. This is even more drastically different from the usual method for Avarohana, since I will still be singing an ascending pair SS SN SD SP ( from low S to the higher notes ) but the general effect is one of descending pattern.

This is more for practice, so I do not have to figure out the arbitrary intervals between any two swaras, I only need to know only the interval from Sa to respective note. I am enjoying that process since I seem to hit the notes more precisely this way. Like, I know very well how S M and S D sound but I do not have to know how D M sounds to play DM since internally I am really playing SD SM. I thought it was a nice trick!

Given this, your finding is quite interesting. On that D2 being a bit lower, can you check if it feels that way, when you compare: SG3 SD2 vs SS SD2 with the first S being there for a minute time, if at all, compared to the G3 and D2? ( the fleeting first S is discrete and not a base for a meend gamaka).
Just curious.

[SrinathK]

Dear vk, Regarding S & D2 and G3 & D2 there is indeed a noticeable difference and I can assure you of that -- it is one limitation of the just intonation system which however is easily answered by the 22 shruti system, the higher one was D2 and the lower one was D1. Both those notes belong within the 22 shrutis.

There are also demonstrations of the 22 shruti instruments on youtube and that harmonium shows no dissonance whatsoever when playing any of the intervals -- 3rd major & minor, 4ths and 5ths. I must say it sounds wonderful and perfectly in tune to my ears. Despite playing all the notes of arohanas and avarohanas of some ragas it still doesn't show any dissonance. There is also a tanpura flash player on that site to show exactly how those shrutis sound in any key. On a side note, one of the co-launchers of the website was Dr. N. Rajam -- so we must stop and ask if TNK's and NR's beautiful tone has to do with this...

The interesting thing is that the algorithm used to generate the 22 shrutis actually gives rise to a 24-tone scale in which 2 variations (one of Sa -- slightly sharp and the other of pa -- slightly flat) are rejected because the position of Sa & Pa are fixed -- thus giving rise to the 22 shrutis -- S, r1, r2, R1, R2, g1, g2, G1, G2, M1, M2, m1, m2, P, d1, d2, D1, D2, n1, n2, N1, N2 & then upper S.

Note : This system has nothing to do with the present Carnatic system where R1 is shuddha rishabham and M2 is prati madhyamam, etc... -- so let there not be any confusion. In fact such a system would force us to redefine the present grammar altogether...

More interesting is the mathematical beauty of this system -- for e.g. is that all but 2 of these shrutis have their own perfect 5th, 4th and 3rd (R1 doesn't have it's perfect fourth and M2 doesn't have it's perfect fifth. That is because of the above omitted 2 tones). However it's not a practical issue as no tuning system in the world uses those 2 intervals -- neither Pythogorean, nor Mesopotamian nor Just Intonation. In fact a combination of all the correct tuning systems used in the world will show that they used 20 of these 22 tones in tonal, with the R1 & M2 being the remaining 2 unique to the indian system -- and it is probably for the reason mentioned above.

Pretty much everything you need to see is over here : http://www.22shruti.com/research_topics_list.asp

On a side note I first learnt swarasthanas with the Equally tempered harmonium (!!). Later when I switched to violin and the tanpura I found the intervals on the harmonium quite out of tune and dissonant to my ear -- the same went for electronic keyboards.

And one more thing, I have even tried to sync audio recordings of great western violinists to the tanpura and found that a lot of chords were played in ET (although in other places they also used Just Intonation (JI) or Pythogorean) and those chords played in ET didn't sound in tune with the tanpura -- the others did (I'd blame that on using ET in the piano as a training tool.) However the melodic lines were almost always resonating beautifully with the tanpura which means there are a lot of spot adjustments that are made while playing. Indeed, some pitch analysis done on great violinists renowned for their sound (e.g. Heifetz) show they did not stick to any one system in particular and relied on human ears ultimately. I think violinist.com was the source.

But right now I have an even more urgent need -- Can we have a 22- shruti app that can be played in any kattai to tanpura accompaniment ?