Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Ear vs. robot, music vs. statistics, commonsense vs. nonsense

I generally try to limit my comments here to the positive or informative, and keep my gripes to myself. But there is something I have been seeing here recently that repeatedly raises my hackles.

Perhaps unike some members here, I consider tune ID threads with accompanying sound or video clip very welcome and a wholly appropriate use of this forum - if I can identify the tune, I enjoy being able to help and if I do not, I enjoy hearing a new tune (and often some great playing). What gets me, however, is resposes of the type, “It could be this one http://[insert tune link here]” - and more often than not, the tune linked to is *not* the tune in the clip.

Seemingly, the contributor has played the recording into Tunepal (or some similar app) and simply posted the result, without first checking the sheet music (or abc) to see if it is the same tune. Naturally, such apps are going to have a hard time with Irish Traditional Music, since so many tunes have so much in common and there are so many possible variations and variants to any tune. Don’t get me wrong, Tunepal is a clever invention - and perhaps has its uses (although I have never used it myself) inasmuch as it sometimes gets it right and, when it doesn’t, you at least get to eliminate the incorrect results. But Tunepal has *no idea* what a tune sounds like - all it can do is identify what notes are in it and in what order, match it with tunes with similar *statistics*. (Statisics and music are about as far removed from one another as two disciplines could be, are they not?)

Quite apart from the fact that saying what a tune “could be” is not a particularly useful contribution (and unnecessary - if you have managed to find the sheet music or abc, just playing or reading through it would comfirm whether it *is* or *is not* the same tune), this is an appeal against trusting technology over the ear. Let us not lose sight of the fact that what music *sounds like* is the most (if not only) important thing about music.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

…and if anyone wishes to post a counter-thread about proof-reading before posting, be my guest. ;-)

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Couldn’t agree more, Creadur!

Unlike you, I *have* used TunePal. Sometimes Tunepal gets it right, sometimes it doesn’t. As with all things in traditional music, one needs to use one’s ears as well before making a determination.

A much bigger irritation for me though is the ridiculous number of "settings" now being posted here for any given tune. Surely this must be a great source of confusion for any newcomer to traditional music when visiting this site. For example, do we really need twenty settings of Drowsy Maggie?

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

As far as settings I find 90 percent of the time I find a better setting of a tune in the subsequent settings rather than the original tune post.

Posted by .

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

TunePal is a lot like my local weather forecaster - a 40% chance of rain/clouds/flog/sunshine etc. For single note melodies played into TunePal, it’s success rate is ≈ 65%. If there’s any background accompaniment it the success rate drops substantially. I generally use it when a tune is running through my head that I know but can’t remember the title. For that TunePal is a quick check rather than referring to a multitude of tunes in dot form. I get a kick out of watching some of the individuals at a session grabbing their phone and accessing TunePal to identify a tune they’re hearing that they don’t recognize. Chances are, TunePal won’t recognize it either. ;^) If it does though, that’s a good thing.

As far as multiple versions of tunes, variety is the spice of life. As a beginner or perhaps a novice at this point, I’ll refer to the multiple versions for a number of things. I can’t play a slur on a bouzouki - is there a version of the tune where a different approach to a phrase matches what I can play? I’m more comfortable applying something that exists than perhaps venturing out onto some phrasing that might be a tad bit erroneous compared to the "norm" for the tune. Drowsy Maggie does seem rather redundant with all of the different versions but there’s no doubt that the version played by say, Altan, would vary greatly from that played at any of the sessions I go to. I wouldn’t mind playing the Altan version - if I could only play that fast. Which I can’t. Still, it’s nice to have the dots available for something like that.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

I’ve never yet found Tunepal give a reliable answer when trying to identify an Irish tune. It does have its good points, but tune recognition is not one of them in my experience.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

That’s the way the world is going, CMO. Instant results, little effort, too much information.

Personally I think the same of many people requesting tune IDs. If they made a bit of effort, used the various search tools and their ears - they could often figure it out themselves and be the better for it..

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

I dunno… sometimes even a little bit of information is better than no information, even if the information is wrong. Because it might help lead you to the right information. There are certainly good reasons to post "It might be this" type answers. Often times someone will post something that is very similar to another tune, but different. So even if you don’t have a definitive answer for the tune, the knowledge that it shares a lot in common with a specific tune can be helpful. (And then you have to decide if it’s a different setting of the same tune, or a different tune from a ‘family’ of tunes, or a completely different tune. But you’re still better off with the partial information…)

And Kilcash, I remember being annoyed when the discussion forums started becoming more "what tune is this?" threads than actual discussion. This was before YouTube had so much trad available, and most of the times, the threads were using ABC to ask the question instead of a link to audio or video. This is why I originally wrote my ABC Tune Search page (which I know needs to be updated now, with the demise of the flash player, I just haven’t had time). But people convinced me that it’s OK to ask about tunes in the forums, because it sparks a conversation and often leads to more information about the tune. And it encourages people to be social.

As far as TunePal goes, it’s a fantastic resource, and I find it amazing that it can be as accurate as it is, because it can recognize tunes played with different settings by different instruments at different tempos, and even in different keys to some extent. So you may think of it as cold, impersonal data that has nothing to do with music, but I think of the app itself as a labor of love, written by someone (Mr. Duggan) who wanted to give the Irish trad world a tool to help make their lives better.

Having said that, it certainly makes sense to not just blindly post an answer to a tune query that TunePal (or my tune search, for that matter) gave you, without checking it out first. I can’t say that I’ve seen people being that lazy, but I generally only check out the tune threads to see if it’s asking for a tune that I actually know.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

"A much bigger irritation for me though is the ridiculous number of "settings" now being posted here for any given tune. Surely this must be a great source of confusion for any newcomer to traditional music when visiting this site."

I agree that it is confusing for beginners. But what is also confusing is when the version submitted is an unusual or idiosyncratic setting and the ‘standard’ session version is buried somewhere in the comments. At least, with the multiple setting format, all settings are given equal status, so the most session-friendly setting does not play second fiddle to Coleman’s or Casey’s variation-laden masterpiece.

"… do we really need twenty settings of Drowsy Maggie?"

Well, not if they are all basically the same version with a few nuances. But I wouldn’t be surprised if there were twenty distinct variants of Drowsy Maggie out there - in which case, bring ‘em on.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

"I dunno… sometimes even a little bit of information is better than no information, even if the information is wrong…. Often times someone will post something that is very similar to another tune, but different."

Yes, I agree - in the absence of anything better, I might sometimes add a comment to the effect of, "It sounds like a version of … " or "… related to …". The trouble is, Tunepal (and abc searches too) picks up on ‘statistical’ similarities, which do not necessarily equate to musical ones - the same sequence of notes can have quite different ‘meanings’ depending on context. So the search results may bear very little musical similarity to the input. (Someone that uses Tunepal will need to fill me in here: Does it rank results on a percentage match basis?)

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

"Don’t shoot the messenger"

If you can’t get tunepal to come back with the tune, in 95% of the cases it’s because you don’t know how to use tunepal to optimise its’ benefits.

as for the statement:
"Don’t get me wrong, Tunepal is a clever invention - and perhaps has its uses (although I have never used it myself) inasmuch as it sometimes gets it right and, when it doesn’t, you at least get to eliminate the incorrect results"

Someone please explain the above sentence to me, eg "I have never used it myself" but I’m going to talk about it like I have, from what, hear say, this site really suffers from people talking about something they know little about, just like those people who post “It could be this one http://[insert tune link here]” it the same level of redundant information.

This site has lots of great information, it’s just buried in a sea of useless information.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

@Theirlandais: You make a fair point - I have perhaps got carried away with detail that is based partly on assumptions. But I don’t believe I have said much that I could not have deduced from reading comments by those that do use it and a little commonsense. I was hoping those that use Tunepal would would either corroborate or correct my assertions - which has been the case thus far. If I have made any incorrect statements about Tunepal, please set me straight.

"Don’t shoot the messenger … If you can’t get tunepal to come back with the tune, in 95% of the cases it’s because you don’t know how to use tunepal to optimise its’ benefits."

‘Shooting the messenger’ is certainly not what I am doing. I am merely shooting (metaphorically) those that use it ineffectively.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Optimise TunePal? I’m looking at the web version (too much of a luddite to own a smart phone) and there are all of three items on the page: Record (text - not an active item), "tunepal" (the logo - also not an active item) and one active button "TAP TO RECORD". So… how do I "optimise" TunePal?

P.S. Not trying to shoot the messenger

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Well in my experience if I play a tune into it it tells me the name first time. Got to be played in the standard key though .thats what I use it for so far .

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Will, that’s great (amazing) you’re able to get such a high percentage.

Posted by .

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Well I don’t use it that much! But it’s been great so far. I use a tin whistle and it normally gets it right first time so I’m very happy with it.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

A single instrument with no backing is best. The more voices involved, the less likely you are to get a good result.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

"A much bigger irritation for me though is the ridiculous number of "settings" now being posted here for any given tune."

This is a rule by Jeremy - before being allowed to post newly written material, you are obliged to first post five to ten existing tunes, or settings thereof.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Tijn, that’s not quite right. Before posting your own compositions, you should post about five *trad* tunes—but they don’t have to be tunes that are already listed; they could be tunes that have never been submitted before. The important thing is that they’re part of the tradition.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

On statistics and Irish music, tune transmission, in spite of all the transcribed tunes available, is a beautifully stochastic process. Stats is about evaluating variability and that’s a primary difficulty that Tunepal has to address. Go Brian!

Jim

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Reverend’s ABC Tune Finder ranks the this according to best match (when I use it, I always check those with a high percentage than say 70% - anything lower than that could be any other tune). Tunepal probably works the same. I don’t use it myself but have seen friends have luck with it the first time. You must play a rather quirky setting in a different key if you can’t find it.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Regarding multiple tune settings, I would certainly agree that the first setting is not necessarily the best one (or the most commonly played one). What is needed is some kind of ranking system based on votes cast by session.org members. As I recall, several folks suggested some kind of a voting system when multiple tune sessions were first introduced on this site. Unfortunately, this idea was never followed up. I would also point out that many of these so-called "settings" often differ from one another by only one or two notes. I wouldn’t mind so much if a justification was always supplied when posting a setting - e.g. "As played by Altan" or "As played at our weekly session at the Slug and Lettuce" or "as in O’Neill’s 1001 Gems" etc, etc. Unfortunately, the majority of settings get posted without any supporting comment.

To sum up:

If you already know a tune, the chances are that you will just carry on playing it the way that you’ve always played it and you won’t bother looking it up it the tune section.

If, on the other hand, if you don’t know a tune at all and wish to learn it my referencing the tune section, the chances are you will want to learn the version that is the most commonly played. No point learning an obscure version, as this will probably conflict with others when you get to play it at a real session. So how are you supposed to know which is the most commonly-played version?

in defence of multiple settings

"So how are you supposed to know which is the most commonly-played version?"
You’re not.
You’re supposed to know the setting they play at your session, the setting that you like to play, with whatever variations that you use, and any other setting that you like, be it Kevin Burke’s Bobby Casey’s or whatever.
Part of being a good session player is being able to adapt to whatever the other players are doing. Usually this means playing follow the leader. If you have not yet acquired that skill, sit out the tune. It won’t kill you.
Part of the beauty of session music is the multiple settings. The idea of standardizing settings, so that you can walk into a session in Akron, Ohio and they’re playing the same settings as they do in County Mayo and in Birmingham UK, quite frankly makes me want to vomit. Like McDonald’s.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

If the tune has a known composer, this *may* but, not necessarily be the most commonly played version. Not necessarily though and even the composer may play it differently if it suits him or her.

Having said that, I think the original tune should have priority in the tune database. I was recently peeved when I noticed that an original tune I submitted https://thesession.org/tunes/2210 was merged into another thread where a variation was given priority.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

"Part of being a good session player is being able to adapt to whatever the other players are doing. Usually this means playing follow the leader. If you have not yet acquired that skill, sit out the tune. It won’t kill you."

And I agree totally.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

I usually limit my comments here to utter rubbish but as I can see the creaturemaddoggin is upset I will say ah who cares. Learning languages like music takes full immersion, full immersion gives u the know how to know what tune is from where and what kinda bounce it needs.

Ppl want to uplink their stuff, more power to their paddy. Some ppl like to misguide others and in the long run it makes the misguided and the misguider seem foolish. The music thrives in fluidity and flux which would indicate a constant changing tradition oscillating between pure drips, ripples, flowing tides and waves that crush and hammer our minds. That would define endless amounts of versions and tunes unfixed in mustard.

Posted by .

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Sounds like statistically TunePal has better results than the ABC tune searching of this site. I have only ever had one success and that was after dozens of attempts over multiple months. I think I need to find this TunePal thing and give it a try.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

‘Statisics and music are about as far removed from one another as two disciplines could be, are they not?’

I work with numbers and, especially, statistics, for a living. Though I’m a long time avid player of ITM, I don’t know much about music theory. However, what little I do know persuades me that music is very mathematical, and math, of course, is the underlying basis for statistical methodologies. So there is some common ground there - just my two-cents…

https://en.wikipedia.org/wiki/Music_and_mathematics

Posted by .

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

@5stringfool;@jeff_lindqvist

Not only have you both totally failed to understand my point, but I find your comments insulting.

I learn most tunes by ear and only very occasionally reference sheet music. The problem is, not everyone is able to do that - especially newcomers to the music. If everybody was able to learn tunes "by ear" as you say, the tunes database would be totally redundant, would it not?

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

I was only commenting on 5stringfool’s paragraph (and interpreted his "you" as a "generic you"). To your question which he quoted:
"So how are you supposed to know which is the most commonly-played version?"

Ask and/or learn by osmosis. There’s no guarantee that a ranking system will work for your session. I don’t play any tune the same way. Ever.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Mix-please understand that as Jeff says, my comments was not aimed at you personally. On the contrary I have every reason to believe that you knew what you’re doing at a session. I will post​ more tomorrow or if you prefer I will PM you, but please believe I meant no offense.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Steve, go shake yourself, I know times can be hard and while your quote below is not without merit it also is uncalled for and unkind to your fellow members on the forum. MMM!
"The idea of standardizing settings, so that you can walk into a session in Akron, Ohio and they’re playing the same settings as they do in County Mayo and in Birmingham UK, quite frankly makes me want to vomit. Like McDonald’s."

Posted by .

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

What did i miss? I see nothing uncalled for or offensive.

Posted by .

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

It’s a bit harsh on McDonald’s, but otherwise the point has a lot of merit!

Jim

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

@5stringfool: Since you say that your comments were not aimed at me personally, your apology is accepted. No need for personal messages … !

jeff_lindqvist@5stringfool

I’ve been attending sessions since the 1970s (maybe before either of you were born?) so hopefully I know roughly what I’m doing.

jeff_lindqvist: "There’s no guarantee that a ranking system will work for your session".

My experience isn’t just confined to *one* session. I’m a regular attendee at six different sessions in England and two other sessions in Wales. Plus playing in sessions at various festivals and also regular gigs with a ceilidh band.

Regarding "Drowsy Maggie", I don’t much care for the tune myself as it’s been overplayed and thus become rather hackneyed. However, if someone else starts it, I usually join in out of politeness. During all those years across all those sessions I’ve heard "Drowsy Maggie" played very, very many times. Not once have I ever heard played with any significant difference to the version that I know. Certainly not with sufficient difference to justify 20 different settings.

jeff_lindqvist: "I don’t play any tune the same way. Ever."

That’s true for a lot of trad musicians - including myself. But that’s not a justification for folks posting every tiny micro variation to the tunes database. If carried to a conclusion, it would lead to an infinite number of settings for every tune, which would hardly be helpful, would it?

I’m NOT against multiple settings, many of which are justified. The problem is, the majority of them aren’t! Especially so if posted without any supporting comment.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

For Tunepal you don’t need a smart phone, it can work on a laptop, PC whatever, you don’t even need an instrument if you don’t have one handy, if you’re a good singer you can sing into it, I only ever got that to work once though.

here’s an example of the simpliest way to get it working with no instrument, lets take a tune like "Lark in the morning"
K: Dmaj
|:AFA AFA|BGB BdB|AFA AFA|fed BdB|
AFA AFA|BGB BdB|def afe|dBB BdB:|

Now take the first fifteen notes (minimum, more if you want) and convert the lower case to upper and the numbers to letters eg A2 = AA a2 = AA abcd = ABCD

eg AFA AFA|BGB BdB|AFA = AFAAFABGBBDBAFA and now add that to https://tunepal.org/index.html#!/notesSearch/

to give you https://tunepal.org/index.html#!/notesSearch/AFAAFABGBBDBAFA click on the link to see what you get

obviously the quality of the results depends on how well your notes match with the person who entered the first setting on this site or other sites.

Sometimes tunes are played in different keys or don’t match the key on the various sites, you can either play the tune and use settings option in tunepal to change the fundamental, eg I play devaney goat in C, If I play it into Tunepal in C, it finds nothing, if I have the fundamental setting as C, it will come back with a match of the tune, but the D version, which is perfect.

you can also convert the notes to D with something like http://sessionite.com/jens-wollschlagers-abc-transposer/

When playing tunes into tunepal, quality is better than quantity, so if you only know a couple of bars well just play for 3 or 4 seconds (15 notes minimum), rather than fumbling for 15 seconds.

for Tunepal the sound is very important, some instruments are better than others, flute is probably the best, I believe the banjo isn’t great with tunepal (but maybe they were’nt setup correctly or were over playing the tune, you should keep the fancy stuff to a minimum), however you can’t go wrong if you know the notes and just enter the simple notes by hand, (15 of them)

all of the above is from trial and error..

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

@Mix O’Lydian I agree with your general point, (and your ‘as played at the Golden Guinea’ transcriptions are usually just what I am after) but my experience with Drowsy Maggie was similar that reported by ‘slainte’:https://thesession.org/tunes/27#comment78033

Which is well off topic. As someone who in the last week has twice linked to tunes that were not the one requested the OP may have me in mind as one of the culprits. So in my defence I will point out that in the first instance I should have said "it’s not this but have a look because it’s relevant to the discussion" and a day later I had learned and I hope made clear I was linking to a tune that had some similarities (which may have been interest to the OP).

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Since Mix O’Lydian has kindly accepted my apology I need not comment on that part of my previous post. However, to further clarify the points I was trying to make:
Multiple settings are useful in a number of ways. True, if you learn tunes by ear and have the ability to adapt your playing to the situation then multiple settings, and indeed the entire Tunes section may be of little use to you. However if you are not there yet, access to multiple settings are much more helpful. I have seen many posts asking which setting is closest to such and such a version, be it a you-tube video or old vinyl recording. Arthur G alone has posted dozens of such queries. They can also be useful for finding local settings- if I were planning a trip to Bristol I might want to look at a bunch of Mix’s settings, of which there are hundreds, in order to be better prepared. I also think that the number of settings posted is a good gauge of how widely-played the tune really is.
For myself, although I prefer to learn tunes by ear, I am a reasonably good sight-reader and I enjoy reading through multiple settings of tunes that I already know. Historical settings, regional settings and versions by well-known players are all of interest to me. Ofttimes I’ll find a little turn of phrase that I like better than the way I play the tune, and that might not have occurred to me on my own. Each different version gives me a little different insight into the essence of the tune.
Most importantly, the multiple settings serves to underline the point, critical for beginners to understand, that there is no one right way to play these tunes. Imagine if all the people who use the tunes section as a primary source all learned the same "most popular" setting of every tune they learned. Within a generation the vast majority of alternate, often better or more interesting settings could be lost. The McDonaldization of trad.
Therefore, I stand by the part of my post that AB finds offensive-I guess the word "vomit" is now politically incorrect? Well, that makes me want to…

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

"Someone that uses Tunepal will need to fill me in here: Does it rank results on a percentage match basis?"

Yes, both my tune search page and TunePal use a version of the "Levenshtein Distance algorithm" (https://people.cs.pitt.edu/~kirk/cs1501/Pruhs/Spring2006/assignments/editdistance/Levenshtein%20Distance.htm) to determine the degree of match. And both sites will give you the top of the list ranked by match ‘distance’, with the closest matches at the top…

This is unlike searching for ABC on this site or google, etc. Those are really doing "word searches", which can sometimes provide reasonable answers, but any variation breaks the "word" and lowers the success rate. For instance, people might play the first four notes of Cooley’s reel as ‘EBBA’ or ‘EBBE’. When those are treated as "words", they don’t match, but when they’re run through a Levenshtein function, they come back basically as a 75% match… And if you put in longer strings of notes, you get more accurate results (although, there’s diminishing return on that. For my tune search, it seems to work best if you just search for 4-8 bars at a time, instead of the whole tune.)

There are some other differences between my tune search page and TunePal. The obvious one, of course, is that TunePal can listen to you playing and go from there… But my site also takes octave into account, and has some features that try to help you find the tune even if your ABC is in a different key than the norm. (Basically, if you specify the "root" of the key your transcription is in, it does a rough transcription to every other key, and searches for each of them…)

I have an update for my search engine planned that actually converts ABC to a string of interval changes, and converts the entire archive to interval changes, and then searches simply on the pattern of interval changes, so it would automatically find the tune no matter what key you posted it in. But that will require some development time that I don’t have at the moment, because my job is pretty demanding (says the guy posting on an internet forum in the middle of a work day), and then playing music takes up a lot of the remainder of my time… ;-)

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

I find Tunepal mostly does find the tune I’m looking for (if it exists in the Tunepal ABC tune corpus).
But…
1. I use it rather seldom (I have most of the tunes in my head)
2. I only use it for tunes which are not already in my own collection and where I haven’t gotten the name along with the tune when picking it up. This means the tunes I search for using Tunepal are usually rather unusual ones, which might appear in only one or a few settings in the Tunepal ABC tune corpus.
3. I use it the "clean" way by playing the first few bars of the tune on the tin whistle or flute with no backing. (Not using a recording from a noisy pub session)

@Reverend:
Don’t use intervals, as that will double the Levenshtein distance. I recommend reading Bryan Duggan’s PhD thesis for more input about this and other problems with tune searches. http://arrow.dit.ie/sciendoc/71/

Quick explanation: Take first 2 bars of Cooley’s reel (long notes converted to shorter)
EBBA BBEB BBAB dBAG
Variation:
EBBE BBEB BBAB dBAG
This variation results in a Levenshtein distance of 1

Using intervals there would be two of them affected by the change of just one note and thus a Levenshtein distance of 2.

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

@David50 - The Golden Guinea session was quite a lot different from other local sessions my area in that many tunes were played there which you probably wouldn’t have heard elsewhere. I managed to transcribe most of these tunes and post them here before the Golden Guinea session closed for good (about 5 months ago).

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

Thanks for the input, Henrik! It’s true that the Levenshtein distance would increase, and I hadn’t thought of it that way. But the other option would be to calculate the intervals relative to the root note of the key. So a single note varied would still only give you one hit to your distance, not two. However, just like the feature that looks for it in all keys on my site, this would require you to input what key your ABC is in… I’ll have to think about that…

Re: Ear vs. robot, music vs. statistics, commonsense vs. nonsense

For the record the part of the quote I posted above which I found difficult to connect with previous replies was *not* the bit about MacDonalds or vomit. Where I was unable to grasp 5strings’ intention was this bit
(& how it relates to the other posts), " The idea of standardizing settings. "
Whose idea does that bit refer to?

Posted by .