Hi, I am wondering if SYNTH V can import midi files with a correctly timed lyrics track in it… if and when I find out how to make midi files with a lyrics track ( are they karaoke ‘.kar’ files or is that yet another type?) it would be soooo helpful to a choir I try to help make some practice tracks. So I would hopefully get a midi-file with a lyrics track ( correctly timed according to the original score) and I just wonder if that can be imported onto a a vocal track in SynthV. Any hint if this is possible or help on how to convert a score with text to MIDI with text very much appreciated.
Yes, Synthesizer V works with midi files that contain lyrics. So you can export to midi and it will have the lyric information in there, and when you import a midi file that has lyrics, it will retain that info.
So, if you created a score in Musescore and added lyrics to it, export that to midi, when you import that midi into SynthV, the lyrics will be there.
Oh that would be marvellous! Would the lyrics have to be added separately in Musescore by hand or can MuseScore use the lyrics printed on the score that has been scanned into the app as PDF? If One would have to still fill those in by hand it might be the same amount of work in Synth V. But I must admit I am completely new to MuseScore and am just trying to help some non-pro choirs ( i.e. no money) to produce practice tracks on the basis of bought scores.
I just checked with ChatGTP and it says that Lyrics export inside midi in MS is not supported. Now ChatGTP can of course make mistakes so I’ll go check. But just as general info for everyone interested here’s what it has to offer : Step 4: Verify Lyrics in the MIDI File
- MIDI files can store lyrics in Karaoke MIDI format, but MuseScore’s MIDI export does not support lyrics natively.
- If you need a MIDI file with lyrics, consider:
- Exporting the score as MusicXML (
.musicxml
) instead. - Using a DAW (like REAPER, Cubase, or Cakewalk) or a MIDI utility to manually add lyrics to the MIDI file.
- Exporting the score as MusicXML (
Alternatively, if you need the lyrics for playback in a MIDI-based karaoke system, you might need to use a dedicated MIDI Karaoke editor like:
- Karaoke MIDI Editor
- VanBasco’s Karaoke Player
- Cakewalk
EDIT Later same day : Have tried all of this but can’t see any lyrics being imported into Synth V Studio Pro, nor in the AU Plugin in Logic. Any ideas what I’m missing?
Workaround for Lyrics in MIDI
If your goal is to use a MIDI file with lyrics in a DAW or Karaoke software:
- Export as MusicXML instead of MIDI.
- Open the MusicXML file in software that supports MIDI lyrics (like Finale, Sibelius, or REAPER).
- Export the MIDI from that software, ensuring lyrics are embedded.
Ok… I think I found a recipe. But will have to check some more tomorrow. Maybe some logic users can chime in to test?
Export MIDI as MusicXML from MuseScore
Import as XML into LOGIC
open score editor
MANUALLY select all
Save selection as midi file
Import as track in Synthesizer V. !! => Import BOTH tracks.
One track holds the lyrics and packs them on the notes.
Hopeful!!
Stemmer
The Musescore website has a utility to convert a PDF score into an MSCZ file. Open up the MSCZ file in the Musescore Studio program (this is a free program). Export this to midi. It should have the lyrics embedded. Import this midi into Synthesizer V Studio. You’ll probably have to do clean up, but I think this should work.
Thanks for your quick responses. I had tried that but could not get either MuseScore to embed the lyrics on midi-export or SynthV to display them even when importing both tracks. When I selected the score in Logic however and “saved selection as a midi file” then SynthV found the lyrics in track 0 and the notes in track 1. It put them together nicely. I will try your method again. Maybe I overlooked something. Cheers!
maybe worst case - import the MIDI notes, then bulk import the lyrics by selecting the notes and pressing ctrl-L. if the source of the lyrics and notes was already aligned, this should put the lyrics where they need to be. in some cases, you’re source program may use different notation for breaking words or extending words across notes, so a simply text editor to find and replace with the SV notation should do the trick.
Do you have an example of a PDF score with lyrics? I want to try to see if I can convert it properly into a midi file with lyrics using Musescore.
Ah, nevermind. I got my hand on a PDF score and tried it. Musescore does not capture the lyrics during the conversion.
Thanks for trying that! I just heard back from the choir that they tried to use MuseScore to scan PDF but it was not sufficiently accurate in notation anyway, plus as you have found out does not scan text. They seem to use a separate App called ScanScore to get it to PDF but hitherto only for notes, so I’m going to investigate if that has OCR that can actually embed the text from PDF… not high hopes though, as placement of text is way more imprecise than note notation…
I have finally managed to get back to do this (delay due to computer problems) and it still works perfectly for me, lyrics are tied to notes perfectly, where one word spans four notes in MuseScore it does the same in SynthV, some times for multi-note words the syllables don’t quite match SynthV’s phoneme splits and need a little help but otherwise, effortless.
I suppose ChumpGPT is a case of “garbage in, garbage out” like so many other so-called ‘AI’ attempts.
I do know MuseScore is a little tricky for entering lyrics, there are a few options on the submenu under ‘Add Text’ but once you figure it out it’s as easy as SynthV, just different.
I have successfully used ScanScore to create a midi file with lyrics for SynthesizerV. There are still some problems with timing and not well-placed lyrics or hyphenations. I am trying to find out if vocal score usually have “breathe” markers ( Like in : Breathe now! or suffocate later) because the score is contiguous but a vocalist could never do that. I will post a full How-To list when I have evened out the last questions. Thanks all! Now here’s to wishing Version 2 makes this process easier and with dynamics from the midi file…
Hi,
Recent MuseScore (since 4.4) can embed lyrics in the midi export. I was very excited once that was added and made a YouTube short on it: https://youtube.com/shorts/-V0SD1eZyNo?si=iziYQigOiQhCYpb-
I have been generating rehearsal tracks for a (mixed) choir for a while, originals with “piano” sounds and more recently with SynthV.
The old forum had a post where I explained my process, but I also posted on the Dorico forum at Creating vocal rehearsal tracks with PlayScore2/Dorico/SynthesizerV - Dorico - Steinberg Forums
I want to write down & share my process in more detail, but haven’t gotten the time to do that yet. Which part of the process are you most interested in?
Hey Jeroen, groetjes uit Berlijn , oorspronkelijk Utrecht, van een medelander
Switching to english for the forum.
Thanks for your very helpful input!
The small choir that I try to help has pdfs and uses MuseScore ( 3). I will definitely start them on 4.4 which I am downloading as I type as I was using what they use. So far I only had success with ScanScore ( Paid) to embed the lyrics. Just watched the short you made. Looks much simpler than via ScanScore and Logic. Thanks!
I have two questions that I hope you can help with.
- Pauses
Having worked with SynthV since version 1.0, I am quite used to converting my sung melodies to notes for any of my voice-databases and editing them a bit. What irritates the SINGERS around me and in the choir is the complete lack of “breathing” interrupts that some score produces when converted to midi. I have asked around and it seems these are not in any form or sign embedded in scores. For SynthV it would be a blessing of course if they were.
I know the API-Script (LUA) in SynthV that does “add breaths” but it does so indiscriminately. For instance in the happy birthday song in your video some voices do not take a breath-pause ( NOT referring to the BR note to actually HEAR it, only a small pause) which to singers makes it sound unnatural and “un-singeable”. So that would still mean a lot of manual editing.
Do you have any method ( other than manual edit of the blobs in SynthV) to add “breath-pauses” when reading from PDF ? A sign to add to the score maybe?
- Hyphenation
Do you know if there is a setting for correction script that can “repair” faulty notes because of hyphenation that was not intended as “note-separation” but strictly for “end of the right of the paper” in the lyrics? As I see in most scores they deliver there is some clear hyphenation for when multiple syllables are intended to be sung as separate notes, SynthV has no problem with that. But in cases f.i. like the word “ready” being chopped up in “read” and “y” on the next line because that is a long “ieeeeeeee” sound I find that SynthV has problems. Of course much depends on the quality of the score and the precision of the text placement. Any tips you can share on this ?
Thanks again for your time.
Frans (Stemmer)
Following this topic with interest, as I create rehearsal tracks for my choir (quite a big one!) using Synth V, and provide them per part and with a mix down giving an indication of the final performance.
It’s fairly quick to input the parts manually, at least for the pop stuff that we do, but I can see a role for MuseScore for more complex music. I was already planning to download the latest version to see how I could exploit it for our situation and this topic has given me some ideas.
I should have time over the weekend to try it out with music I’ve already converted. I’ll come back with my findings as soon as I can.
I think you might like that workflow when you get used to MuseScore’s myriad options. If you have, for example, a grand staff for keyboard and a soloist’ line then exporting the main score as MIDI will give you three tracks when imported into SynthV, the keyboard will have default words inserted. For your use case, each vocalist’s line will import as its own voice track in SynthV so audio can easily be isolated track exports, this is my way of working with final song assembly in my DAW. Musescore’s instrumental audio export rejoining the vocals there.
I have been trying to get “good” results with MuseScore but the audioveris transcription, even when used in 4.4…1, does not take along any lyrics on my Mac Pro… ( It says it is created with 3.6.2 when I download it and open it in 4.4.1. so maybe I need a PRO version to do this? ) I tried 6 scores.
I then tried ScanScore ( demo) again with a pdf but even there lyrics placement is too inaccurate to be of any use in 4 out of 6 cases… Maybe bad pdf’s are rife?
I am aware that this varies with the quality of the score and the text placement and probably textsize , but it also the case that in general “legatolization” ( if that is a word) from midi is a problem. Where a part is sung with clearly separate bass notes or melody-stabs , the midi somehow “binds all together” and manual editing of the part is then still quite extensive.
This is also partially due to the Vocal AI of course, interpreting the falls and rises with its “own” intelligence. Switching to manual mode for that part or maybe even to rap mode to get the more percussive bass notes may be the only possibility to forego this problem. If ANYone has this sussed ( Jeroen? japj ) I would gladly be schooled
So in conclusion : I was able to import midi with lyrics via the music.xml route via Logic. But text placement was too erratic and that sort of f*cked up note and syllable separation to the extend of it not being worth while.
For now I will stick with importing the midi and then typing in the text in SynthV, hoping that the just ordered version 2 (Hurrah) will make this obsolete or some new feature in ScanScore can actually compare midi-interpretation with lyric placement with the audio as a sort of “check-up”.
Thank you all for the input!
I think I am starting to understand the problem here…
you are using OCR to turn ‘printed’ scores into MIDI, I am entering notes and lyrics into MuseScore, I am linking each word (or part-word) to a note whereas the OCR program has to guess the placement based on position.
When you read a score how do you decide which note a word starts on, it isn’t explicitly stated in ‘printed’ music, just implied, how can any OCR be accurate?
Am I missing something?