craigsapp
commented
4 years ago Now you are trying to sneak in the transposition by semitone that I was complaining about initially 😛
I will implement Transposer class functionality to do what you want. This will also make it possible to do what you originally wanted, such as verovio --transpose +3 to transpose up 3 semitones. This system should be further enhanced to check for automatic enharmonic transpositions at every key / key signature change throughout the score if you want it to be fully generalized, but just doing assigning a single transposition at the start of the music will work in about 99% of real-world cases.
For automating the generation of capo knowledge inside of verovio, that will probably be dependent on having some formal way of indicating it in MEI. The informal way to indicate that a harm is a capo is to add harm@type="capo" (which is already possible in MEI). A more formal way would be to allow harm@form="capo" (should probably be formally allowed in MEI) or possibly harm@class="capo" (@class is probably free-form, so could be set to capo without any change in the MEI spec.). When using harm@type="capo", the @type attribute will be mapped to <g class="harm capo"> in the SVG rendering, so you could also stylize the capo in the SVG with CSS (change its color or hide it, for example).
A problem that you would have to negotiate with @lpugin is that you seem to want to generate the capo chords automatically from the regular chords. This is a transformation of the data which is not generally done inside of verovio (except that I get special dispensation to do it in the Humdrum importer 😉 ), but this sort of thing would be done with external processing data. MEI sorts of people would suggest using XSLT to do such a thing (in which case you would have to implement the transposition algorithms in XSLT). In the worst case, an independent C++ program that uses pugixml to load MEI data and borrows the Transposer class could be implemented and compiled to javascript with emscripten. The MEI data would then be passed through your preprocessor before being passed to veorvio.
If you are not planning on automatically inserting the capo harmony elements, but rather change existing elements to a new capo transposition, then adding an extra option to verovio such as --capo -3 seems more possible. This option would only transpose element that have a @type="capo" attribute on the transposable element. Or perhaps the options would be --transpose -3 --capo or --transpose -3 --type capo, or use of an XPath option in conjunction with the transpose option. (which can be discussed further with @lpugin).
If the capo is already inserted into the MEI data, then the verovio --transpose option will work on it correctly, assuming you are not changing the capo transposition amount. For example of G minor music is transposed to C minor, then the real chord on the first note will be changed Cm and the capo will be changed to Am.
Below are charts for transposition by semitones. For semitone transposition, there will ideally be enharmonic adjustments to choose between alternate possible notations. For each semitone transposition, there are two possible transpositions that could be done (or at least two reasonable ones). The example initially discussed was a +1 transposition which would convert C major into either C-sharp major or D-flat major. The best choice will be D-flat major since that add 5 flats while the C-sharp major transposition adds 7 sharps. In other words, there will be two options to choose for the transposition, and the option that gives the least number of accidentals will be the optimal solution (which will usually, but not always, be desirable).
The first column is the number of semitones to transpose by. The second column lists the two possible enharmonically equivalent key signature adjustments related to that semitone transposition, and the third column is the chromatic interval related to the circle-of-fifths number in the second column. When doing a transposition by semitone, you would take the key signature in the music as a circle-of-fifths value and add each of the values in the second column to this, and then choose the resulting value for the circle-of-fifths for a new key signature which is closest to zero (so a transposition to D-flat major will always be preferred to C-sharp major).
Once the option of when circle-of-fifths value is decided on, then the chromatic interval in the third column can be used to perform the transposition as usual with the existing code.
Example: Suppose that you are in C major and want to transpose up five semitones. The starting key signature is "0" and the two fifth-transpositions are "-1" and "+11". 0-1=-1 is a better choice than 0+11=11 since -1 is closer to 0 than 11 is. Therefore the chromatic transposition to select is +P4 rather than +A3.
Here is an example that involves enharmonic wrapping: Suppose that you are in C-sharp major and want to transpose up 7 semitones. The starting key signature is "7" and the two fifth-transpositions are "+1" and "-11". 7+1=8 is not as good as 7-11=-4, so the best transposition value will be +d6.
Here is the table for positive semitone transpositions:
| Semi | 5ths | Interval | Example |
|---|---|---|---|
| 0 | 0 | P1 | C => C |
| +1 | -5 | +m2 | C => D-flat |
| +1 | +7 | +A1 | C => C-sharp |
| +2 | +2 | +M2 | C => D |
| +2 | -10 | +d3 | C => E-flat-flat |
| +3 | -3 | +m3 | C => E-flat |
| +3 | +9 | +A2 | C => C-sharp-sharp |
| +4 | +4 | +M3 | C => E |
| +4 | -8 | +d4 | C => F-flat |
| +5 | -1 | +P4 | C => F |
| +5 | +11 | +A3 | C => E-sharp |
| +6 | +6 | +A4 | C => F-sharp |
| +6 | -6 | +d5 | C => G-flat |
| +7 | +1 | +P5 | C => G |
| +7 | -11 | +d6 | C => A-flat-flat |
| +8 | -4 | +m6 | C => A-flat |
| +8 | +8 | +A5 | C => G-sharp |
| +9 | +3 | +M6 | C => A |
| +9 | -9 | +d7 | C => B-flat-flat |
| +10 | -2 | +m7 | C => B-flat |
| +10 | +10 | +A6 | C => A-sharp |
| +11 | +5 | +M7 | C => B |
| +11 | -7 | +d8 | C => C-flat |
And here is the table for negative semitone transpositions (same as above, but the signs are all negated):
| Semi | 5ths | Interval | Example |
|---|---|---|---|
| 0 | 0 | P1 | C => C |
| -1 | +5 | -m2 | C => B |
| -1 | -7 | -A1 | C => C-flat |
| -2 | -2 | -M2 | C => B-flat |
| -2 | +10 | -d3 | C => A-sharp |
| -3 | +3 | -m3 | C => A |
| -3 | -9 | -A2 | C => B-flat-flat |
| -4 | -4 | -M3 | C => A-flat |
| -4 | +8 | -d4 | C => G-sharp |
| -5 | +1 | -P4 | C => G |
| -5 | -11 | -A3 | C => A-flat-flat |
| -6 | -6 | -A4 | C => G-flat |
| -6 | +6 | -d5 | C => F-sharp |
| -7 | -1 | -P5 | C => F |
| -7 | +11 | -d6 | C => E-sharp-sharp |
| -8 | +4 | -m6 | C => E |
| -8 | -8 | -A5 | C => F-flat |
| -9 | -3 | -M6 | C => E-flat |
| -9 | +9 | -d7 | C => D-sharp |
| -10 | +2 | -m7 | C => D |
| -10 | -10 | -A6 | C => E-flat-flat |
| -11 | -5 | -M7 | C => D-flat |
| -11 | +7 | -d8 | C => C-sharp |
earboxer
lpugin
Since we've implemented transposition, another helpful feature would be adding Capo'd versions of chords for guitarists (and other stringed instruments that use a capo).
It should receive a non-zero positive integer (the number of frets this is being capo'd), and for each chord symbol (
harmelement), it should add a transposed chord symbol (transposed down the number of semitones that the capo is set to).Example: