Hi Karl, hi all
Thanks for your thorough insights into the history of MIDI.
My initial question ( a) ) was about the way Lilypond implements the way
how Lyrics are rendered into MIDI. There are several standard-conforming
ways. But the way Lilypond currently does it by default does not suit my
purpose.
With regards of the issueing date of RP-017 (1997?) I realise that MIDI
has never been designed for this purpose. 20 years ago MIDI was perfect
to control physical synthesizers and maybe to feed some karaoke systems.
And I understand that investing time into MIDI implementation is a waste
of time nowadays since MusicXML has been invented to overcome those
shortcomings while being extensible for future (unforseen) scenarios.
What a pitty that Lilypond still does not offer MusicXML export.
Taking all the considerations, impediments and potential efforts to take
into account, I have come to the conclusion that I'm ending this
experiment although that I'm aware that with a decent programming effort
(parsing and transformation) at least some limitations might be
eliminated. However, the technical complexity exceeds my abilities and
my willingness to invest as a non-programmer.
I'll think about redesigning my arranging workflow and start in my DAW
instead of Lilypond.
Thanks for all your effort. I've learned a lot!
Best regards
Stephan
Am 19.09.2025 um 20:32 schrieb k...@aspodata.se:
Stephan Schöll:
I'm investigating whether I can produce vocal rehearsal tracks with
Synthesizer V Studio 2 Pro by importing MIDI files which I rendered with
Lilypond based on the score.
Just for reference, Stephan is talking about Standard Midi file format
(SMF) files. I have seen two versions of that format, 1.0 and 1.1, and
a version 2 seems to be in the works.
They should be available at https://midi.org/file-formats though I
cannot verify that at the moment.
I have uploaded copies of them at https://aspodata.se/choir/midi/
My research suggests that synthV has some specific requirements for MIDI
files in order for lyrics to be imported directly with the melody.
Initial tests have failed in any case.
So with the help of a HEX editor I started investigating the MIDI files
rendered by Lilypond. Find an MWE attached.
My findings so far:
Lilypond...
1.) uses SMF Type 1 -> ok for Synth
2.) saves lyrics as Lyric Events (not Text Events) -> ok for SynthV
3.) saves notes and lyrics in separate MIDI tracks -> nok! SynthV needs
notes and lyrics combined in one track
I think that you want to look at:
Recommended Practice (RP-017)
SMF Lyric Meta Event Definition
available somewhere at https://midi.org/ and also at
https://aspodata.se/choir/midi/rp17.pdf
Lilypond doesn't follow that practice, I guess since lyrics is separate
and more or less independent to the music.
My questions:
a) Lilypond-related: Can somebody confirm 3.) - if confirmed: Is there a
way to control/change that (for a non-programmer?), by juggling with
contexts e.g. (see MWE)?
Yes, it is so.
You can confirm it with https://aspodata.se/git/musik/bin/midi.pl
I provide two examples https://aspodata.se/choir/midi/example/
Use it as: midi.pl <SMF file>
I don't think you can control nor change that from within lilypond.
b) Off-topic: Is there anybody with a working workflow to render vocal
rehearsal audio tracks including lyrics rooted in Lilypond except for
recording one's own voice ;-) ? If yes, I'll kick off a separate OT thread.
Sorry I don't have a solution to offer you.
It would be nice if we could implement RP-017 for a subset of lilypond file
syntax, I guess at least \lyricsto would be requied.
Regards,
/Karl Hammar
