Open craigsapp opened 3 years ago
Thanks for picking-up on this Craig. My sense is that more precise data is always better, but then we also need to make it consistent. It shouldn't be too much of a problem to detect other examples with this issue in RDS and I will alert the RAs to the new system. I think for the duration we should include the upbeat, but maybe best to cut it at a certain number of decimals. Any suggestion?
But then, there is the issue of the total number of bars as well, which we should also edit.
It will be fairly each to check: Any example that starts with measure 1 can be double checked to see if there is a pickup beat. It will not be too big of a deal, except for some occasional analysis purposes. Basically when people say "measure 1" they mean the start of the music, even if there is a pickup measure before measure 1. Computers are more literal about these things and would like to know if you mean the start of the music, or the measure labeled "1" (with the pickup measure traditionally numbered measure "0". There may be cases where the example contains partial measure in the middle of the music which would be related to this sort of case.
Also, the import from Sibelius into Humdrum does not have real pickup measures via MusicXML so as those are fixed, the metadata can be adjusted in the spreadsheet. There can be various ways of encoding it in MusicXML, but the scores are encoded with padding rests at the beginning of the examples, so there will be no problem in terms of analysis with the data as it currently is.
For analysis purposes it probably does not matter too much: Currently the scores with pickup beats that are imported from Sibelius have a padded full measure, the the metrical information is preserved. In any case, if we ever need to extract individual measures, I can extract starting at measure "0".
Technically, if the measure start on the third beat of a three beat measure, then the start is 0.666666. Would it make sense to write it that way in the FDS metadata master?
I think that would be useful: It would give an idea of how much music is in the pickup measure. When extracting the measure with myank
it will be easy to identify that 0.666 is less than 1, so measure 0 is needed to be extracted.
There are 4 RDS examples with pick-up measures:
When an example starts with a pickup measure, that measures is (or should) be considered measure
0
(not measure1
).Consider R310, which I changed to measure
0
in the metadata (Example Start Measure
):This is because the
myank
program distinguishes between the first measure (myank -m 1
) which comes after the pickup measure from the pickup measure itself:myank -m 0
.How should
Example Total Measures
deal with pickup measures? In this case there are 8 full measures of 3/8, and one pickup measures containing an eighth note. Should the example be considered8
measures since there are 8 full measures or should it be8.333
since there are 8 and 1/3rd of a measure in the example?