Standardizing tempo changes vs. multi-movement structure across trios

[WolfgangDrescher]

op01n05c and op01n09b contain tempo changes within a single piece, while other trios use multiple movements for this purpose (e.g., op04n10a – op04n10b – op04n10c).

Should we standardize this?

[johentsch]

op01n09b indeed is an inconsistency, thanks for bringing it up. In order to fix this, we would need to split the file after four bars, move the rest to c) and rename the current c) and d) to d) and e). This is a major change that would result in a major release because such fundamental changes may break prior evaluation pipelines for this corpus. This is not a reason against the change but I would still ask if you think it is worth it? If it solves a problem for you, I'll definitely do it.

op01n05c appears to be different from the other cases (where a very short segment segues from one movement into the next without belong to one or the other) in that it constitutes a single movement alternating between Adagio and Allegro parts. Sure, beginning and end could be split off but, to me, it would seem like an arbitrary decision:

Also, the current split dovetails with the four movements indicated on IMSLP and Musicbrainz. My feeling towards this topic is that the harmonization that we should be doing is merging some of the files to make them conform to the Musicbrainz identifiers which are modelled after CD tracks (with future alignment to audio in mind). Then again, since different recordings might handle things differently, sticking to the more atomic splits might actually be beneficial.

What are your thoughts, @WolfgangDrescher, keeping your project in mind?

[WolfgangDrescher]

op01n09b indeed is an inconsistency, thanks for bringing it up. In order to fix this, we would need to split the file after four bars, move the rest to c) and rename the current c) and d) to d) and e). This is a major change that would result in a major release because such fundamental changes may break prior evaluation pipelines for this corpus. This is not a reason against the change but I would still ask if you think it is worth it? If it solves a problem for you, I'll definitely do it.

I don't necessarily need the files to be separated for my project. I have a filter function that allows you to search by tempo, so it would be nice to have that, but as it stands now, it's also good for finding special cases where there are multiple tempos per movement, whereas that's not the case with separated files such as op04n10a-c.

If this would lead to a major release on your side, it might not be worth the effort.

I have to find a solution for displaying different tempos within a piece correctly anyway, because I agree that op01n05c is a slightly different case than the others. To my knowledge, it is not possible in Humdurm to add tempo changes within a piece semantically correctly. I use the reference record OMD (https://www.humdrum.org/reference-records/) to mark the initial tempo. I will then use local comments for the tempo changes within the piece and simply create a text label for them there.

For the record, there is also op01n09d.

We have dealt mostly with Op.4 in the seminar so far, where we noticed op04n06c – op04n06d. In this case, it would actually make sense for me to merge the two pieces into one, as I am currently unable to create cadences across pieces. However, I can also solve this with a workaround using comments, as merging pieces would probably also be a major upgrade for you.

[WolfgangDrescher]

To my knowledge, it is not possible in Humdurm to add tempo changes within a piece semantically correctly.

Actually never mind: https://verovio.humdrum.org/?file=beethoven/sonatas/sonata32-1.krn

It looks like I can add multiple OMD records in one file. So this should not be a problem at all for me.