As a was up late last night (back ache),I thought I’d look at possibly converting a batch of songs in abc into lilypond. Using the abc2xml then musicxml2ly route as it does appear to give tidier .ly and the more direct abc2ly program is unsupported and may be dropped so isn’t really one for future use. I only got as far as the first trial file and noticed a couple of problems:

I think this is on the musicxml2ly side as importing the xml into muse score was ok but. The conversion seems to use a “blank note”, eg. “s4”. to handle an anacrusis. I think this explains a weird blank bar in the middle of a Kesh jig conversion I tried in the other thread. It is less of a problem in my sample song which just has one lead note with no repeats or multiple parts(eg, AB) but it still doesn’t work out right. In this case, “padding” (say s4 a4) before the first note would be ok but this came out as (say) a4 s4. Apart from looking a little odd on a score, a problem with midi production like this is you get an “instant” first note and then silence to fill the rest of the bar.

The B and W: fields did not get through. These exist in the musicxml as miscellaneous fields but I’ve been unable to find any information on mapping musicxml fields to ones that the musicxml2ly progam would take.

If (as I would be doing if I got to doing a batch for what I was playing with), I’d be extracting the abc from a database and running the conversions by a php script, I think I’d manage to add the missing fields to the .ly in the script (but wouldn’t be able to rewrite musicxml2ly) but don’t see how I’d make the substitutions for the first problem.

O well, not sure I’d use it anyway, just something to play with at 2am…