[abcusers] RE : tune finder
John Chambers wrote - But if the software doesn't agree on what pieces of the notation mean, it can sorta interfere with getting the music across. I've been thinking along the same lines myself for quite a while. And abc has a quandary that's common in all other kinds of computer communications: You find something that can't be expressed using the standard language. What do you do? Discuss it with as many interested people as you can. Listen to their ideas and objections. Modify your proposal accordingly. Arrive at a consensus and only then implement your ideas. Isn't that what this list is for? Start with a rule Any notation you don't understand should be ignored (perhaps with a warning but not a fatal error message). Not always possible when different sets of non-standard notation impinge on each other such as the use of ! and the various incompatible versions of the V: command. When a small crowd finds something that seems to solve the problem, they present what they've done to the general population. And a lot of people like it so it gets used and becomes part of the system. Unfortunately it screws things up for other people who may cry Wouldn't it have been better if But it's too late. The damage is done. Eventually most of the new ideas get incorporated into the standard language Not any more they don't. The standard hasn't been updated for several years and there is no mechanism to do so. Alternatively, they don't get incorporated, and you get a collection of dialects or a family of very similar but incompatible languages. Yes, that's what's happening. Bryan Creer To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
[abcusers] Explicit key signatures
Than you Phil Taylor for a sensibly argued case. A welcome change from the this-is-my-opinion-and-I'm-sticking-to-it attitude that has been prevalent lately. Unfortunately I didn't agree with any of it. He wrote - In order to describe a piece of music completely, you need to know any two of the tonic, mode and key signature. The key signature is sufficient alone. Tonic and mode are implicit in the tune. Conventional musical notation gives only the key signature, which is inadequate, and has led to the practice in classical music of including the tonic in the title of the piece. I have been taking classical oboe lessons for three years and have never been given a piece with the tonic in the title. I think this practice died out years if not centuries ago. One of the tests my teacher occasionally springs on me is to put a piece in front of me and ask for the key or give me a key and ask for the key signature. If the law was changed to make IOUs legal tender It was. A long time ago. They're called banknotes. The economy suffers if there are a lot of counterfeits in circulation which, in abc, there are. People are blatantly passing off Edorian tunes as K:D. Please note Bryan, that I'm not objecting to this suggestion on the grounds that it's likely to be popular Perhaps not; it's a subtle distinction. You ARE saying that you think a large number of people are not that concerned about having the tonic and mode and that the explicit key signature format would be widely used. You think that they should be prevented form doing so because you know better. My concern is that they will still find it simpler to say K:D to get two sharps because it's easier than K:^f^c. Perhaps it's tonic alone that should be outlawed (except that we can't of course, because it's too late.) Finally, if we want to make life easier for people transcribing from manuscript by permitting them to use an incomplete description of key, perhaps we should do the same thing for those transcribing by ear, and permit them to specify only the tonic. After all, any competent musician who was familiar with the tradition concerned should easily be able to figure out where to put the necessary accidentals in order to make sense of the tune. But the user might not be familiar with the tradition so you'd have to include a sound file. In that case you wouldn't need a tonic (or the notes for that matter). abc software requires the K: command so the resulting abc would be something like - X:2312841276 T:The Pig and the Parlour Maid C:Trad N:From the playing of Gilbert Hairy Scrotum F:PigMaid.wav K: I don't think I'll be implementing that just yet. (Actually, come to think of it, it wouldn't be that difficult.) Bryan Creer To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] Explicit key signatures
[EMAIL PROTECTED] wrote: Than you Phil Taylor for a sensibly argued case. A welcome change from the this-is-my-opinion-and-I'm-sticking-to-it attitude that has been prevalent lately. Unfortunately I didn't agree with any of it. He wrote - In order to describe a piece of music completely, you need to know any two of the tonic, mode and key signature. The key signature is sufficient alone. Tonic and mode are implicit in the tune. Bryan Creer To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html I also respectfully disagree with Phil Taylor's rationalle. I pointed out my objections to the K:key-mode specification, instead of a direct key signature sharp or flat specification, in a communication here on Jan 25, 2001. Here is a slightly revised version of it. The first line of the Introduction on Chris Walshaw's ABC homepage says abc is a language to notate tunes in an ascii format. It says nothing about describing, or in any way characterizing tunes. That, at present, seems to be beyond the stated purpose of ABC. Notation can be unambiguously satisfied by giving the number of sharps or flats on the key signature; It doesn't require any interpretation of keynote or mode to do that. As many as 3 different 'descriptions' in terms of key and mode are supplied for some of the tunes given in 'Sources of Irish Traditional Music', 1998. Who decides which one to use? Keys and modes are subject to interpretation, an interpretation that is sometimes not unique. The tune is the tune; it doesn't need a 'description' in terms of key and mode. Circular modes, ones that don't end on the keynote, (which are 4% of the 6601 tunes coded in file COMBCODE.TXT on my website) can sometimes be difficult to figure out as far as key and mode go, and not all interpreters will come out with the same key-mode combination for a given tune. Bruce Olson -- Roots of Folk: Old British Isles popular and folk songs, tunes, broadside ballads at my website A href=http://www.erols.com/olsonw; Click /a Motto: Keep at it; muddling through always works. To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] Explicit key signatures
Phil Taylor writes: I don't mind if you give any random collection of sharps and flats in the K: field as long as you also give the tonic. It's clear that the twelve-tone crowd hasn't gotten into abc yet... In order to describe a piece of music completely, you need to know any two of the tonic, mode and key signature. Knowing any two of these makes it quite trivial to determine the other if you need it. Conventional musical notation gives only the key signature, which is inadequate, and has led to the practice in classical music of including One question strikes me: Phil, is the guitar your main instrument? I was glad to get the post, tho, since I'd been wondering what the controversy could be about. Now I see that it is more philosophical than musical, It hinges on what completely means, what information is sufficient, and whether or not abc should be content with minimal sufficient information, or insist on more. Which in turn depends on what one thinks is the purpose of abc. A subject I _really_ hope to avoid. I claim that the key signature alone is sufficient to allow one to decode the piece. (The proof is that even a computer---which is so dumb that it only understands explanations which are so clear they can not *possibly* be misunderstood---can correctly play the tune and display the staff notation with only the keysig as a guide.) It's true that the additional information of the tonic or mode is useful, and it's even necessary for some additional things---such as telling the guitarist at a session what key the tune is in---but it's not really necessary for the musician who just wants to play what is written. Why should we demand more? Phil gave three examples where the additional info is needed: automatic chord-setting, just intonation, and transposition. In the first case, chord-setting, I'd think it's up to the user to get this right. (Guitarists are free to disagree. In fact, I may be hard-hearted here, since I usually strip the chords off any piece before playing it, so I can hear what it sounds like. But I just tried chord setting on a couple of tunes, and found to my surprise that abcmus set the same chords to the major and the dorian key signatures. That's too few examples to allow any conclusions, tho.) In the second case---and by the way, does Barfly do just intonation now? Cool!--I'd think that anyone who knew enough to ask for just intonation would also know what he or she was doing, so the whole problem would be moot. I haven't thought much about automatic transposition, but wouldn't it work if the user knew what info to feed it? (Not necessarily including the mode or key note.) If you write an E dorian tune with K:D and trasnpose D to G, doesn't the tune transpose to A dorian? For the explicit key signatures, the program probably has to do extra work, but isn't it possible to transpose everything, including the key signature, up or down by x semitones without knowing the key note? At any rate, I'm in favor of minimality here: I'd like abc to accept the minimal sufficient information to convey the tune, meaning in this case that it would accept any of tonic and mode, OR explicit keysigs, OR tonic-plus-modifying-accidentals, OR other such, within reason*. I agree that tunic + mode is a great way to describe the music I play, and in fact, I've learned quite a bit about it from abc, but I don't want to force my own prejudices on others. [* Of course, not everything is within reason. For instance, consider this Question: what is K:DMixMix, the mixolydian scale based upon the D mixolydian scale? ;-)] thought to what the actual tonic (or mode) is. The problem with this suggestion is that it represents a degradation of the abc standard, since the resulting K: field contains less information, and while programs would still be able to display the staff notation or play the notes Loss of information, yes. Degredation? I'm not sure. One has to have a little faith in the user. Presumably, he or she wants to communicate some music to a certain audience via abc. If that audience understands, it's a success. Furthermore, human nature being what it is, the introduction of this K: format would encourage many users to give up trying to figure out tonic+mode and simply take the easy way out by entering the key signature only. This in turn would inevitably lead to most new abc transcriptions adopting it, and the whole corpus of abc music would suffer. But we already have this situation! Many people write the major key which gives the correct sharps and flats, even for dorian and mixolydian tunes. In fact, the O'Neill project even asked that its transcribers do just that. (I think there were several considerations here, tho.) As a practical matter, I doubt that explicit key signatures would ever be used that much, except possibly for keysigs with one or at most two accidentals---it's just too much
Re: [abcusers] Explicit key signatures
Bruce Olsen wrote: I also respectfully disagree with Phil Taylor's rationalle. I pointed out my objections to the K:key-mode specification, instead of a direct key signature sharp or flat specification, in a communication here on Jan 25, 2001. Here is a slightly revised version of it. The first line of the Introduction on Chris Walshaw's ABC homepage says abc is a language to notate tunes in an ascii format. It says nothing about describing, or in any way characterizing tunes. That, at present, seems to be beyond the stated purpose of ABC. Notation can be unambiguously satisfied by giving the number of sharps or flats on the key signature; It doesn't require any interpretation of keynote or mode to do that. As many as 3 different 'descriptions' in terms of key and mode are supplied for some of the tunes given in 'Sources of Irish Traditional Music', 1998. Who decides which one to use? Keys and modes are subject to interpretation, an interpretation that is sometimes not unique. The tune is the tune; it doesn't need a 'description' in terms of key and mode. Chris invented abc as a quick and dirty way of writing down tunes, and later wrote the first version of abc2mtex in order to simplify the task of translating his hand-written notes into conventional notation. The abc standard which we still use is part of the documentation for abc2mtex. abc2mtex is solely concerned with notation, and indeed could have used a simple collection of sharps or flats as its key information. However, the language has since developed into a storage medium and a notation system in its own right, independant of the programs which are used to interpret it. Chris was prescient enough to create a key description which is more useful than the conventional key signature, and which in turn makes the data available for purposes other than conversion to staff notation. Circular modes, ones that don't end on the keynote, (which are 4% of the 6601 tunes coded in file COMBCODE.TXT on my website) can sometimes be difficult to figure out as far as key and mode go, and not all interpreters will come out with the same key-mode combination for a given tune. Whatever system of key representation you choose, there will always be a small number of tunes which prove difficult to represent correctly. When you enter a conventional key signature, in some cases you have to decide whether a particular note should carry an accidental, or whether that accidental should be part of the key signature. In order to answer that question correctly you have to figure out where the tonic is, and what mode the majority of the tune is in. Hard cases make for bad law, and I don't think that the difficult tunes make a case either way. Of course, if abc were _only_ to be used for copying music from printed sources, and those sources were always unimpeachably correct, that problem wouldn't arise. Phil Taylor To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] Explicit key signatures
John Walsh wrote: I don't mind if you give any random collection of sharps and flats in the K: field as long as you also give the tonic. It's clear that the twelve-tone crowd hasn't gotten into abc yet... Yeah well, in that case I suppose what's really needed is K:none. Mind you, I'm strongly of the opinion that while it's quite possible to write completely atonal music, the result can't be described as a tune. In order to describe a piece of music completely, you need to know any two of the tonic, mode and key signature. Knowing any two of these makes it quite trivial to determine the other if you need it. Conventional musical notation gives only the key signature, which is inadequate, and has led to the practice in classical music of including One question strikes me: Phil, is the guitar your main instrument? How did you guess:-) I was glad to get the post, tho, since I'd been wondering what the controversy could be about. Now I see that it is more philosophical than musical, It hinges on what completely means, what information is sufficient, and whether or not abc should be content with minimal sufficient information, or insist on more. Which in turn depends on what one thinks is the purpose of abc. A subject I _really_ hope to avoid. However, it is a subject which is germane to this discussion. I claim that the key signature alone is sufficient to allow one to decode the piece. (The proof is that even a computer---which is so dumb that it only understands explanations which are so clear they can not *possibly* be misunderstood---can correctly play the tune and display the staff notation with only the keysig as a guide.) It's true that the additional information of the tonic or mode is useful, and it's even necessary for some additional things---such as telling the guitarist at a session what key the tune is in---but it's not really necessary for the musician who just wants to play what is written. Why should we demand more? It's not a case of demanding more - rather of not discarding what we already have. Phil gave three examples where the additional info is needed: automatic chord-setting, just intonation, and transposition. In the first case, chord-setting, I'd think it's up to the user to get this right. (Guitarists are free to disagree. In fact, I may be hard-hearted here, since I usually strip the chords off any piece before playing it, so I can hear what it sounds like. But I just tried chord setting on a couple of tunes, and found to my surprise that abcmus set the same chords to the major and the dorian key signatures. That's too few examples to allow any conclusions, tho.) In the second case---and by the way, does Barfly do just intonation now? Cool!--I'd think that anyone who knew enough to ask for just intonation would also know what he or she was doing, so the whole problem would be moot. BarFly does just intonation, Pythagorean intonation and highland bagpipe. You can also invent your own (subject to a few limitations). For the first two it has to know the tonic though. It doesn't yet do automatic chord assignment, so I don't actually know if it's possible to do that without knowing the tonic (I couldn't do it by ear though). I haven't thought much about automatic transposition, but wouldn't it work if the user knew what info to feed it? (Not necessarily including the mode or key note.) If you write an E dorian tune with K:D and trasnpose D to G, doesn't the tune transpose to A dorian? Yes, that would work. Even a wrong tonic+mode is better than nothing! For the explicit key signatures, the program probably has to do extra work, but isn't it possible to transpose everything, including the key signature, up or down by x semitones without knowing the key note? I was thinking more of the problem of how to transpose a tune in two sharps into three flats, but you've actually supplied the answer. Give it an arbitrary tonic+mode (anything that fits) and transpose that. At any rate, I'm in favor of minimality here: I'd like abc to accept the minimal sufficient information to convey the tune, meaning in this case that it would accept any of tonic and mode, OR explicit keysigs, OR tonic-plus-modifying-accidentals, OR other such, within reason*. I agree that tunic + mode is a great way to describe the music I play, and in fact, I've learned quite a bit about it from abc, but I don't want to force my own prejudices on others. [* Of course, not everything is within reason. For instance, consider this Question: what is K:DMixMix, the mixolydian scale based upon the D mixolydian scale? ;-)] Since only the first three characters of the mode are significant, it's the same as K:DMixogamous (but different from K:DMyxematosis). thought to what the actual tonic (or mode) is. The problem with this suggestion is that it represents a degradation of the abc standard, since the resulting K: field contains less information, and while
[abcusers] Re: Explicit key signatures
Phil Taylor wrote - Even a wrong tonic+mode is better than nothing! I don't think I need to comment. Bryan Creer To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
